• Energy: wind, solar, hydro, tidal, wave, geothermal etc
• Spatial: microscale, mesoscale, synoptic and global
• Temporal: seconds, minutes, diurnal, seasonal, interannual, decadal and climatological
• Approach: measurement, modeling
The success of wind power has pushed turbines and research need into increasingly complex environments — mountainous terrain, forested areas, high in boundary layer and offshore. For solar power, new installation sites such as floating PV both rivers, water reservoirs for artificial snow, and “alpine PV farms” are gaining more attention. We need accurate measurements and short-term forecasts of cloud fields and aerosol effects. For weather-dependent renewables, the challenge of integrating shares into the power grid requires advances in understanding forecast uncertainty and spatio-temporal variability. Furthermore, meteorological conditions define how much power can be sent through the power grid and could help prevent curtailment or negative energy pricing.
Specifically, we invite contributions including but not limited to:
• Measurement techniques and analysis for e.g., wind, solar, hydro, and marine resources.
• Wind conditions (resource, extremes, turbulence) on all scales in complex environments (mountains, forests, coastal, offshore, urban).
• Wake effect models and measurements
• Forecast performance and uncertainty of RE at different time horizons
• Forecasts and detection of extreme and adverse weather events (wind ramps, droughts, heatwaves, storms, compound and consecutive)
• Detection and forecasting of dynamic line rating suitable conditions
• RE resource and atlas development (wind, solar, hydro, wave, thermal)
• Hydro-meteorological analyses of inflow variability, snowpack, precipitation extremes, and their implications for hydropower
• Tidal and wave resource assessment and predictability
• Impacts of renewable power plants or their large-scale integration on local, regional, and global scales
• Tools for strategic planning of RE in urban areas and smart energy systems.
• Climate Change Impact studies for renewables and weather-driven energy demand
• Interannual to decadal variability of renewable resources
• Typical Meteorological Years and probability of exceedance metrics
• AI and Machine Learning for weather and climate forecasting and applications to RE
Wind turbine wakes can significantly impact the performance of downstream turbines, reducing power generation and increasing loads. The characteristics of these wakes are heavily influenced by conditions within the atmospheric boundary layer (ABL). We investigate the interaction between wind turbines and the atmosphere with focus on the near wake region, up to 4 rotor diameters downstream. A large dataset of inflow conditions and wake characteristics comprises measurements from a nacelle-mounted Doppler wind lidar, a meteorological mast and turbine operational data. The data are collected at the research wind farm WiValdi in northern Germany. The lidar scans multiple horizontal planes to derive wake characteristics and near wake lengths, which are then analyzed across a range of atmospheric conditions. The results show that wake velocity deficits are reduced in turbulent conditions and enhanced under stable conditions. Wind veering across the rotor layer is found to correlate with increased wake deflection and vertical tilting, while a high shear exponent and potential temperature gradient which are both characteristic of the stable ABL are associated with increased lateral asymmetry of the velocity deficit. The near wake length is observed to extend on average around 2.01 rotor diameters downstream and exhibits greater sensitivity to atmospheric conditions than to turbine operational parameters. In stable conditions with low turbulence, near wake lengths can be particularly long. Further analysis will explore the asymmetry of the near wake and its vertical tilting in more detail, with complementary measurements from a second, ground-based lidar scanning vertically through the wake during a campaign to improve understanding of the three-dimensional wake dynamics.
How to cite: Menken, J. and Wildmann, N.: Wind turbine wake characteristics in various atmospheric conditions investigated with lidar measurements at WiValdi, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10801, https://doi.org/10.5194/egusphere-egu26-10801, 2026.
Wind energy plays a key role in achieving carbon-neutral power generation, yet its deployment in complex terrain remains challenging. The WINSENT (Wind Science and Engineering Test Site in Complex Terrain) research facility addresses these challenges by operating research wind turbines in complex terrain.
The test site is located on the Swabian Alb near Stuttgart, Germany, in close proximity to a steep, forested escarpment that influences the local flow conditions. It is equipped with two research wind turbines (RWTs) and four meteorological masts with heights of 100 m. Unlike purely commercial turbines, the research turbines are operated under full experimental control, permitting deliberate activation and shutdown of the turbine and enabling wake studies under well-defined operating conditions.
Additional observation is provided by the University of Tübingen through campaign-based in-situ measurements using multicopter uncrewed aircraft systems (UAS). The UAS are simultaneously deployed at strategic locations, including the upstream inflow and multiple horizontal distances downstream of the turbines. They resolve turbulent structures down to sub-metre scales, allowing detailed investigation of flow variability, terrain-induced influences, flux measurements, turbulent kinetic energy (TKE), and mean wind statistics.
An extensive investigation of RWTs’ wake formation and horizontal and vertical structure is presented during multiple simultaneous UAS measurements. Despite the high surface roughness and the strongly heterogeneous flow conditions induced by the present complex terrain, turbine wakes can be clearly identified from the ultra-near-wake region at distances as close as 20 m downstream of the rotor, as well as at downstream locations corresponding to one-, two-, and three-rotor-diameter distances, with maximum observed wind-speed deficits reaching approximately one third of the inflow wind speed. Measurements acquired during turbine operation and under powered-off conditions are compared, revealing pronounced differences in wake structure, turbulence levels, and wake recovery, and confirming that the observed wind-speed deficits are primarily turbine-induced.
How to cite: Gruchot, L., Schön, M., Büchau, Y., Zum Berge, K., Rettenmeier, A., Bange, J., and Platis, A.: Investigation of Wind Turbine Wakes in Complex Terrain at the WINSENT Test Site Using UAS Measurements, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10118, https://doi.org/10.5194/egusphere-egu26-10118, 2026.
The Royal Meteorological Institute of Belgium (RMI) has been delivering offshore wind and power forecasts to Elia, the Belgian transmission system operator for high-voltage electricity, as part of a dedicated storm forecast tool, in an operational setting since November 2018. With an installed capacity of 2.26 GW fully completed by the end of 2020, the Belgian offshore zone (BOZ) is one of the highest density wind energy zones in the world. Each Belgian wind farm has a relatively high number of turbines and/or installed power per area. Moreover, due to lack of space in the Belgian North Sea, all Belgian wind farms lie close together in a narrow band, with the Dutch Borssele wind farm zone nearby. There is thus a considerable impact of intra-farm and inter-farm wakes on both power production and mesoscale wind.
In order to improve offshore wind and power forecasts in the Belgian North Sea, the BeFORECAST research project was funded by the Energy Transition Funds of the Belgian federal government, from 01 November 2022 until 31 October 2025. The project was coordinated by the von Karman Institute for Fluid Dynamics (VKI), in a consortium with KU Leuven, the Royal Meteorological Institute of Belgium (RMI), SABCA, 3E and Vrije Universiteit Brussel (VUB).
We give an overview of RMI's main results in the BeFORECAST project over the past 3 years. In particular a wind farm parameterization was implemented in RMI's operational weather model ALARO, and an artificial neural network for power forecasting was trained on power production data and NWP forecasts. Both wind and power forecasts were further compared with VKI's mesoscale WRF model, and against real-world observations from turbine SCADA, lidar and power production data. The influence of the planned second Belgian offshore zone, the future Princess Elisabeth zone (PEZ), on the BOZ production was also studied. Additionally, 3DVar assimilation of Doppler radar radial wind (VRAD) in ALARO was tested, with very promising results on offshore wind speed forecasts, showing a positive impact up to 24 hours in forecast lead time. Finally, two methods for postprocessing wind speed NWP forecasts using historical lidar and SCADA data were investigated. We developed a neural network for postprocessing of deterministic ALARO forecasts, and a modified member-by-member approach with special emphasis on storm events for the ensemble forecasts of ECMWF.
How to cite: Smet, G., Van den Bleeken, D., Van den Bergh, J., Dehmous, I., Degrauwe, D., Van Ginderachter, M., and Deckmyn, A.: Incorporating wake effects in Belgian offshore wind and power forecasts, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13162, https://doi.org/10.5194/egusphere-egu26-13162, 2026.
Machine learning-based limited-area models (LAMs) have been shown to rival or even outperform conventional numerical weather prediction models at local, high-resolution forecasting tasks. This study investigates how the Encoder-Processor-Decoder architecture, which has been successfully employed in numerous applications, can be adapted for wind power prediction. Leveraging the Anemoi framework developed by the European Centre for Medium-Range Weather Forecasts (ECMWF) and various national weather services, we implement graph-based neural networks over a spatial domain encompassing the North Sea region. Different weather models, including standard graph neural networks and attention-based methods, are trained using high-resolution weather data from the Copernicus Regional Reanalysis for Europe (CERRA). We explore several strategies for incorporating wind power at different stages of the training pipeline, including training weather models jointly with wind power data from scratch, as well as finetuning pretrained weather models specifically for wind power forecasting. Training and verification are performed utilizing the publicly available wind power production data from the European Network of Transmission System Operators for Electricity (ENTSO-E). The impact of input feature selection and architectural design choices on forecast skill is evaluated. In addition, the resulting wind power forecasts are benchmarked against those obtained from conventional physics-based methods and state-of-the-art data-driven approaches. This comparison provides insight into the benefits and limitations of end-to-end learning frameworks for renewable energy forecasting and their operational applicability.
How to cite: Van Poecke, A., Van Ginderachter, M., Van den Bergh, J., Smet, G., Van den Bleeken, D., Tabari, H., and Hellinckx, P.: Integrating Wind Power into Graph-Based Limited-Area Weather Forecasting Models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7982, https://doi.org/10.5194/egusphere-egu26-7982, 2026.
This study proposes methods of wind power predictions from Numerical Weather Prediction (NWP) models and evaluates wind power ramping event predictions in the Belgian Offshore Zone. We verify the operational deterministic model ALARO-4km at the Royal Meteorological Institute of Belgium, its enhanced version incorporating Wind Farm Parameterization (WFP), and the ECMWF ensemble prediction system. To convert meteorological variables into power forecasts, we implement both physical power curves and machine learning methods, including XGBoost and Transformer models. Within the machine learning models, we over-sample rare but high-impact events such as turbine cut-outs during high wind speeds, enabling the models to effectively learn these critical extreme states. While initial validation using traditional metrics suggests that the Transformer model achieves the lowest Mean Absolute Error (MAE) for deterministic and Continuous Ranked Probability Score (CRPS) for probabilistic, we argue that these aggregate scores may mask deficiencies in the capture of rapid power fluctuations, which is vital for stable grid operations.
Since ramping events pose challenges to power system operations, we further verify the capability of these models to predict significant ramps. We highlight the limitations of standard metrics like MAE and CRPS, as they often optimize average timing and magnitude errors in a way that rewards "over-smoothing", even though such smoothing renders the forecast ineffective for detecting ramps. To overcome this, we propose a verification framework that introduces an error buffer for deterministic contingency analysis (hits, misses, and false alarms) and adapts this buffer concept for probabilistic verification within the Brier Score. We apply these proposed verification solutions to our power model outputs and evaluate the models' useful skills. In deterministic forecasting, the XGBoost model achieves higher scores for most ramping events compared to other models, whereas the power curve approach proves more effective for capturing large-scale ramps within the ensemble-based probabilistic predictions. Our results demonstrate that the Transformer’s low CRPS is largely a result of its smoothed output, which is unfavourable for predicting actual ramping events. These findings emphasize the need for operational caution when identifying "optimal" models, suggesting that lower scores in average error metrics do not inherently guarantee reliability for managing critical power ramps. Our proposed verification solutions provide an intuitive framework for understanding and comparing the predictive skill of various models specifically regarding ramping events.
How to cite: Meng, R., Smet, G., Van den Bergh, J., Tabari, H., Van den Bleeken, D., and Termonia, P.: Evaluating the Operational Skill of Deterministic and Probabilistic Wind Power Ramping Event Predictions for the Belgian Offshore Zone, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12727, https://doi.org/10.5194/egusphere-egu26-12727, 2026.
Destination Earth (DestinE) is the European Union's flagship initiative to develop Digital Twin (DT) models of the Earth system. It leverages cutting-edge advances in numerical prediction, digital technologies, high-performance computing, and AI to enhance our understanding of climate change and evolving weather extremes. One of its key objectives is to support the European Commission's Green Deal by enabling the large-scale integration of renewable energy into Europe's energy system. This ambition is pursued through several energy-related use cases, such as the ongoing Onshore & Offshore Wind Energy Information project, or closed Energy Systems for making a resilient power system.
European Centre for Medium-Range Weather Forecasts (ECMWF) has developed Plume, co-funded by the European Commission under the DestinE initiative. Plume is a plugin mechanism for Earth system models that extends their processing capabilities through modular add-on functionalities. Plume dynamically loads plugins at runtime and provides read access to in-memory model fields via a well-defined interface (based on the Atlas library (Deconinck, et al., 2017)), enabling application-specific processing alongside the main model without costly I/O operations. This framework has been applied in the EU Horizon project DTWO, which develops a Digital Twin for wind energy applications. In collaboration with DTWO partners, ECMWF created two Plume plugins, introduced at the 2025 European Meteorological Society Annual Meeting, for wind farm modelling and extreme weather event detection, tested in Extremes DT-like experiments.
This presentation focuses on recent extensions to the Plume framework that enhance these plugins' usability and relevance for the wind energy value chain, while enabling broader development of renewable-energy applications through improved configurability. Wind energy applications require high-frequency, high-resolution data at turbine hub heights (50-200m). The Extremes DT, running the Integrated Forecasting System (IFS), computes wind fields on model levels, from which hub-height winds can be interpolated. However, certain heights, e.g., 100m, are only computed at output steps, limiting availability for plugins during model integration, and the typical output heights do not fully capture the required range. To address these limitations, Plume now includes its own data generation capability. Beyond interfacing with original model fields, Plume can manage derived fields and variables, feeding plugins with relevant data while centralising processing costs and methods, e.g., hub-height wind interpolation. This feature is implemented using an observer pattern, propagating updates from source model data to Plume-managed fields and triggering strategy-based recalculations. The design prioritises extensibility and avoids redundancy in plugin code by concentrating derived data generation within Plume. For the wind farm modelling plugin, this enhancement enables direct retrieval of wind data at configured hub heights, supporting more accurate resource assessments while keeping the implementation application-focused. By consolidating these capabilities within Plume, the framework fosters greater collaboration on iterative improvements and plugin development, engaging a broader community of stakeholders in shaping its evolution.
How to cite: Ducher, C., Bonanni, A., Sarmany, D., and Quintino, T.: Unlocking Renewable Energy Insights with Plume: Extensions for Wind Energy and Beyond , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21392, https://doi.org/10.5194/egusphere-egu26-21392, 2026.
Expansion of offshore wind energy in tropical-cyclone-prone regions (Japan, Taiwan, USA) requires robust and defensible estimates of extreme wind hazard to support engineering design, planning, and risk quantification. In current practice, extreme wind speeds for offshore wind projects are typically derived on a site-specific basis using a combination of historical hurricane records, atmospheric reanalysis products, and synthetic storm simulations. However, the limited length of observational datasets together with the spatial undersampling of rare events and methodological differences between hurricane hazard models may introduce substantial uncertainty and inconsistency in estimated design wind speeds.
This study presents a comparative evaluation of three storm track modeling frameworks commonly used in hazard modeling. The analysis focuses on the U.S. East Coast, extending from South Carolina to the Gulf of Maine. Extreme wind speed maps are generated on a 0.25° × 0.25° grid, consistent with the spatial resolution required for large-scale offshore wind energy design studies and portfolio-level risk assessment.
The evaluated frameworks differ in both their synthetic storm track generation methodologies and their associated hurricane wind field models. Using consistent regional domains and return-period definitions, we quantify differences in modeled extreme wind speeds and assess their spatial variability across sub-regions. The results reveal large discrepancies between hazard intensity maps, particularly in areas characterized by low historical storm frequency. These findings highlight the intrinsic uncertainty associated with hurricane hazard modeling for offshore wind applications and demonstrate the importance of systematic benchmarking across methodologies. By providing reference-scale comparisons of extreme wind hazard estimates, this work aims to support more consistent, transparent, and defensible assessment of design wind conditions for offshore wind energy infrastructure.
How to cite: Deskos, G.: Evaluating Hurricane Wind Hazard Models to guide the development of Offshore Wind Design Maps, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22131, https://doi.org/10.5194/egusphere-egu26-22131, 2026.
Accurate characterization of the vertical wind profile is essential for wind energy assessment, particularly in regions with heterogeneous surface conditions that are challenging for mesoscale modelling. In Latvia, extensive forest cover introduces significant surface roughness, increasing uncertainty when extrapolating wind conditions from mesoscale models to local, mast-measured wind profiles.
This study investigates wind speed and direction profiles based on measurements from communication masts at heights between 10 and 85 meters across three sites in Latvia. Two years of 10-minute averaged wind measurements at several height levels are analysed, with additional remote sensing data used where available. The measured wind profiles are compared against mesoscale model products commonly used in wind resource assessment, with a focus on the influence of surface roughness.
The analysis focuses on surface roughness related differences in mean wind speed, wind shear, and directional dependence between mast-based observations and modelled wind fields. The results demonstrate systematic differences between measured and modelled wind profiles over forested terrain, highlighting the limitations of mesoscale models in resolving local surface effects relevant for wind energy applications. The analysis identifies conditions under which these deviations are most pronounced, providing guidance for the interpretation of mesoscale model output at microscale sites.
These findings emphasize the importance of site-specific measurements for wind energy applications in Latvia and provide insight into the sources of uncertainty when applying wind atlases and reanalysis data in regions with complex terrain and surface roughness.
How to cite: Priedītis, J., Sīle, T., Bethers, P., Bethers, U., and Poļs, R.: A Comparative Analysis of Mesoscale and Microscale Vertical Wind Profiles for Wind Energy Assessment, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19096, https://doi.org/10.5194/egusphere-egu26-19096, 2026.
The amount of offshore wind capacity has been growing rapidly on the global scale. In Australia, there is currently no installed offshore wind capacity, but this is projected to change over the coming decades following government targets. Therefore, it is important to assess how wind energy availability varies in coastal areas, to understand potential opportunities and risks of offshore wind in the context of the broader energy system.
A key mode of local wind variability in coastal areas is the sea breeze, associated with daytime differential surface heating of the land and ocean, and the resulting thermal circulation with onshore flow near the surface. Although the sea breeze has been characterised by previous studies at individual coastal sites, there has yet to be a robust assessment of occurrences across the broader region of Australia, due to a lack of observational data and generalisable identification methods. As a result, several aspects of the sea breeze and associated wind variations have remained unexplored, including in regions relevant for future wind energy generation.
Here, we use a km-scale atmospheric reanalysis to characterise sea breeze occurrences over Australia. We investigate the spatial and temporal variability in their occurrences, as well as potential impacts on offshore wind energy. This includes the development and application of a new method for defining sea breezes as objects from reanalysis output, using a diagnostic of atmospheric fronts.
We find that there is more wind energy available during the afternoon over offshore wind areas on days with a sea breeze identified, compared to other days during the summer. Sea breeze days also tend to have higher average regional energy demand compared with other days, likely due to warmer surface air temperatures over the land that provide sea breeze forcing and lead to enhanced electricity demand from cooling. However, the amount of offshore wind energy also tends to be lower in the morning on sea breeze days relative to other days, likely due to weak prevailing winds that are then opposed by the formation of the sea breeze. Finally, due to the role of the prevailing wind direction in sea breeze formation, there is an anti-correlation in occurrences between opposite-facing coastlines.
These spatial and temporal variations in offshore winds associated with the sea breeze suggests a potentially important source of renewable energy. The sea breeze is shown here to drive local winds during the late afternoon in the summer, when demand is often high, and solar resources are reduced, representing large potentially value in the energy system. In addition, anti-correlation in occurrences between opposite-facing coastlines suggests that a diversity of offshore wind farm locations could be beneficial for energy reliability.
How to cite: Brown, A. and Vincent, C.: The impact of sea breezes on offshore wind energy resources in Australia, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1605, https://doi.org/10.5194/egusphere-egu26-1605, 2026.
PV generation is affected strongly by the short-term fluctuations in meteorological conditions - from clear sky to cloudy, which would be considered normal conditions, to more rare events like snow fall, freezing rain or dust that covers the modules. Rare events are hard to forecast reliably, they are usually not well represented in the training data and cause imbalances in training and prediction. However, they can have huge impacts on the energy system as for example snowfall usually covers many PV plants simultaneously over extended areas. Such events, if not foreseen in time, require balancing action of network operators and thereby cause large costs and possibly strain on the energy infrastructure.
State of the art PV forecasting models are overwhelmingly being trained on datasets without accounting for changing conditions and rare situations. To improve the prediction of such events we present a new method in the form of a data labeler and classifier for snow conditions in PV forecasting based only on meteorological and historical PV generation data to allow for a classification of the expected forecasting conditions over a time horizon of the next few hours up to days. With the classification performed, the best suited model trained for the expected condition can be employed to yield the most reliable PV forecast.
The method is site-specifically trained with historical PV generation data of the site, but no other metadata, module specifications, satellite or visual data are required. The data is combined with historical weather measurements (like irradiance, temperature and precipitation) from a close-by meteorological station. By classifying the conditions in the training dataset with the method, rare conditions are identified and labelled. The labels do not require exact validation, a high likelihood is sufficient. Expert models for those conditions can then be trained in a supervised setting. These are exposed to a training dataset that has dense samples of the selected rare condition and can include augmented samples of the condition. Thereby, a range of specialized forecasting models is created and benchmarked against each other to ensure selection of the best performing models for forecasting in case of a forecast rare condition.
Preliminary results from Austrian PV systems indicate a high accuracy of 99.6% and true positive rate of 96% for the labelling method with a false positive rate of only 0.05% on a test dataset. An LSTM neural network-based classifier to forecast conditions 24 hours ahead shows similar performance metrics and an LSTM regressor expert model achieved only 30% of the PV forecasting error of a similar non-expert model. Both classifier and expert regressor were trained on the labelled and condition enriched dataset.
The work was funded by the Austrian Climate and Energy Fund and carried out under the program "Energieforschung 2022".
How to cite: Reisenbauer, S. and Schicker, I.: Forecasting rare but impactful events in renewable energy generation - condition classification for optimal expert model training and model selection in PV forecasting, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19429, https://doi.org/10.5194/egusphere-egu26-19429, 2026.
We introduce a novel spatiotemporal framework for intraday photovoltaic (PV) power forecasting and apply it to a systematic comparison of seven PV nowcasting approaches, assessing their accuracy, reliability and sharpness. The benchmarked methods range from satellite-based deep learning and optical-flow techniques to physics-based numerical weather prediction models, and include both deterministic and probabilistic configurations. Model performance is first evaluated at the irradiance level using satellite-derived surface solar irradiance fields as reference data. The irradiance forecasts are subsequently converted into PV power estimates using a station-specific machine-learning-based irradiance-to-power model, which takes local solar irradiance and local solar azimuth and elevation angles as predictors. This approach enables the transformation of solar irradiance forecasts into PV power forecasts. The latter are validated against measured production from 6434 PV installations across Switzerland. To our knowledge, this work represents the first national-scale analysis of spatiotemporal PV power forecasting. In addition, we present novel visualizations illustrating the influence of mesoscale cloud dynamics on national PV generation at hourly and sub-hourly temporal resolutions. The results indicate that satellite-based models consistently outperform the Integrated Forecast System ensemble (IFS-ENS) at short forecast horizons, although their performance degrades more rapidly than that of IFS-ENS as lead time increases. SolarSTEPS and SHADECast yield the highest accuracy in both irradiance and power predictions, with SHADECast exhibiting the most reliable ensemble dispersion. While the deterministic IrradianceNet model achieves the lowest root mean square error, probabilistic forecasts from SolarSTEPS and SHADECast provide superior uncertainty calibration. Forecast skill is found to decline with increasing elevation. Moreover, cloudy and high-variability weather conditions remain the most challenging for PV power forecasting. At the national level, satellite-based models reproduce daily total PV production with relative errors below 10% for 82% of days during 2019–2020, highlighting their robustness and suitability for operational deployment.
How to cite: Lanzilao, L. and Meyer, A.: A spatiotemporal framework for intraday PV power forecasting using satellite-based and numerical weather prediction models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5577, https://doi.org/10.5194/egusphere-egu26-5577, 2026.
Satellite-based solar irradiance forecasting plays a key role in the short-term management of photovoltaic (PV) power generation. It provides intraday Global Horizontal Irradiance (GHI) forecasts more accurate than NWP models, but their performance remains highly sensitive to cloud cover dynamics and synoptic weather situations.
Large-scale circulation over the Euro-Atlantic area can be commonly described by four leading teleconnection patterns: the North Atlantic Oscillation (NAO), East Atlantic (EA), East Atlantic–Western Russia (EAWR), and Scandinavian (SCA) patterns, each characterized by positive and negative phases. While their influence on climate variability and seasonal renewable energy production has been widely studied, their impact on satellite-based solar irradiance forecasting errors has never been quantified. Previous analyses have shown that individual NAO circulation indices can modulate solar irradiance forecast errors, motivating a comprehensive daily assessment of Euro-Atlantic teleconnection phases.
Here, we analyze eight years (2016-2023) of satellite-derived GHI forecasts at the SIRTA observatory near Paris (France). Four-hour-ahead forecasts with a 15-minute temporal resolution are generated using CMV-based extrapolation of geostationary satellite cloud fields and evaluated against pyranometer observations. Daily Euro-Atlantic teleconnection indices (NAO, EA, EAWR, SCA) are computed from ERA5 500 hPa geopotential height anomalies using an EOF-based methodology. Each day is classified according to the dominant teleconnection pattern and its positive or negative phase.
Forecast errors are quantified using the relative root mean square error (RRMSE) up to a lead time of 4 hours, with a particular focus on the 2-hour forecast horizon as a representative forecast skill assessment. The RRMSE across the full period is 30.8%. Distinct error regimes emerge across the eight teleconnection states (NAO±, EA±, EAWR±, SCA±), with generally lower forecast errors during NAO+, EA+, and SCA phases, and higher errors during NAO-, EA-, and EAWR phases.
Pronounced seasonal contrasts are observed, with the highest (37.4%) and lowest (27.9%) RRMSE values occurring in winter and summer, respectively. Variations in forecast errors across teleconnection phases reflect both circulation dominance and phase frequency. For example, EAWR- exhibits elevated errors in winter (+18.5% relative to the seasonal mean), which progressively decrease from spring to autumn, while NAO- shows reduced errors in winter (-12.8%) but increased errors during spring, summer, and autumn. RRMSE were elevated in winter and spring (20.5% and 7.7%) and reduced in summer and autumn (-15.8%, -6.3%) during EA+. Similar but opposite error patterns were observed during EA- phases across consecutive seasons.
These results highlight the importance of considering the full Euro-Atlantic teleconnection framework when interpreting satellite-based solar irradiance forecast performance. By extending teleconnection analysis to intraday forecast errors, this study demonstrates that large-scale circulation phases provide valuable information for understanding and anticipating variability in solar forecasting skill, with direct implications for PV forecasting and energy system management.
How to cite: Singh, S., Cros, S., and Badosa, J.: Impact of Euro-Atlantic teleconnection phases on satellite-based solar irradiance forecasting errors, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17332, https://doi.org/10.5194/egusphere-egu26-17332, 2026.
West Africa has significant solar energy resources, and the growing number of photovoltaic power plants is increasing solar production. The establishment of a day-ahead market should make it possible to increase the share of this intermittent energy in the energy mix. However, this type of market requires estimating the production a day in advance, and thus addressing the challenges of weather and solar forecasting.
Dust advection and clouds are the two meteorological phenomena that most influence photovoltaic production in West Africa. They are still poorly represented by numerical weather models in this region, as no operational high-resolution regional forecasting systems exist. Moreover, the available global operational forecasting systems generally use a low-resolution aerosol climatology that does not account for high-frequency spatiotemporal variability of atmospheric dust content.
This study aims to evaluate the potential improvements achieved through a regional model that incorporates aerosol information and offers high resolution (3 km, 15 min) over Burkina Faso. The Weather Research & Forecasting Model (WRF), an atmospheric simulation system from NCAR, is supplemented by an extension for solar application: WRF-Solar. It can be used with different types of aerosol data, calculating the influence of Aerosol Optical Depth (AOD) on Global Horizontal Irradiance (GHI). In this study, WRF-Solar is used with three different configurations: without AOD data, with the monthly aerosol climatology built into WRF, and with the hourly 2D AOD forecast from CAMS (an atmospheric chemistry model produced by ECMWF). The WRF-Solar simulations are forced by ECMWF IFS forecasts. The simulations have a duration of 36 hours to meet the requirements of the day-ahead market. Two study periods were chosen: during the monsoon season, from July 2023 to September 2023 and during the dry season, from January 2024 to march 2024. The forecasts are evaluated against in-situ GHI measurements from a pyranometer located at Zagtouli photovoltaic power plant.
The results show that simulations using the CAMS 2D AOD forecast and those using the built-in monthly aerosol climatology give similar overall results, with their own specific characteristics. Both configurations simulate an overestimated GHI. They both have a clear advantage over the WRF-Solar configuration without AOD data. Slight differences between the configurations are observed in the calculated GHI on cloudy days during the monsoon, which are related to differences in cloud representation.
How to cite: Solbès, A., Cosme, E., Raynaud, D., and Anquetin, S.: Influence of aerosol input data on WRF-Solar global horizontal irradiance forecasts for solar energy in West Africa, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11439, https://doi.org/10.5194/egusphere-egu26-11439, 2026.
Accurate prediction of solar irradiation is important for renewable energy integration, agriculture, and environmental studies. However, solar power output is highly intermittent, with rapid fluctuations driven by cloud advection, formation, and dissipation. This intermittency increases operational uncertainty for grid operators and can raise reserve requirements. We present a deep learning framework for ultra-short-term forecasting of cloud evolution and solar irradiation up to 7.5 hours ahead using a 5-hour morning history ending at 09:00 (30-minute sampling). The model is trained with GK-2A geostationary satellite observations and auxiliary meteorological information. Conventional video prediction models often under-represent early-stage advection signals and tend to produce overly smooth forecasts, which limits their utility for irradiation prediction. To address these issues, we propose SimVP-Flow (Simple Video Prediction) with three components. First, we use the GK-2A water vapour (WV) infrared channel, infrared window channel and solar zenith angle (SZA) as inputs to provide both mid-to-upper-tropospheric flow cues and physically consistent diurnal geometry during pre-dawn and post-sunrise periods. Second, we incorporate optical-flow-derived motion fields as an explicit constraint to encourage sharper and more advective-consistent forecasts. Third, the decoder is modified with hybrid skip connections and PixelShuffle-based upsampling to better retain high-frequency cloud boundaries and reduce blurring artifacts in long-horizon predictions. We evaluate the proposed method on GK-2A case studies and compare it against single-channel baselines and the original SimVP. Performance is assessed using image-based metrics for cloud fields (e.g., MSE and SSIM) and error statistics for irradiation. This work aims to improve physically consistent short-horizon solar forecasting in data-sparse regions using satellite imagery and lightweight auxiliary variables.
This work was supported by the National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIT)(RS-2025-00515357).
How to cite: Kim, D. and Yeom, J.: Solar Irradiation Nowcasting with Flow-Guided Cloud Dynamics Prediction by SimVP-Flow, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8475, https://doi.org/10.5194/egusphere-egu26-8475, 2026.
In response to global climate change, photovoltaic (PV) power plants have been rapidly deployed over the past decade in order to reduce greenhouse gas emissions in electricity production. This massive deployment of large-scale solar parks in rural areas raises questions about the modifications in micrometeorology they cause in contrast to conventional rural land surfaces. This calls for physically based land surface models able to represent the specific land-atmosphere interactions induced by solar parks within weather and climate models. In these models, land surface schemes often neglect the alterations in radiative transfer, surface energy balance, and near-surface turbulence caused by solar panels, potentially leading to biases in weather and climate simulations over regions with large-scale PV power plants.
In this contribution, we present PV-LAND, a photovoltaic land surface model developed as an extension of the Interactions between Soil, Biosphere, and Atmosphere (ISBA) scheme, specifically designed to represent the surface energy balance of a coupled soil-vegetation-PV-atmosphere system. The solar park is represented as a periodic array of panel rows over a vegetated surface, and the surface energy balance is resolved using a nodal approach that explicitly describes the front and back surfaces of PV modules, the photovoltaic cell, the underlying vegetated or bare ground, and the air layers within and above the PV canopy. Shortwave and longwave radiative exchanges account for panel shadowing and multiple reflections between panels and the ground, while turbulent exchanges of momentum, heat, and moisture are computed using parameterizations adapted to the specific geometry and aerodynamic properties of PV canopies, as well as to the wind direction relative to the panel rows.
The model has been run in offline mode over an extensive solar park in southwestern France, where flux measurements (radiative, momentum, heat, and water vapor) have been collected for several years. The PV-LAND performances will be presented at the conference, with a focus on the model's ability to represent the surface energy balance and the surface temperatures.
How to cite: Bolsée, S. and Dupont, S.: Modeling the surface energy balance of a vegetated solar farm, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4000, https://doi.org/10.5194/egusphere-egu26-4000, 2026.
As solar photovoltaic (PV) systems are deployed globally to decarbonize energy production systems, atmospheric dust has emerged as a critical challenge due to its potential to drastically reduce production efficiency in many regions. Dust particles both attenuate incoming solar radiation and accumulate on photovoltaic module surfaces thereby reducing light transmission and power output. Soiling losses (defined as power production losses due to dust accumulation on PV panels) vary at daily, monthly and interannual timescales, as dust accumulation and removal processes depend on time-varying factors such as particulate matter concentration, wind, relative humidity, precipitation and cleaning operations. Capturing these dynamics thus requires assessments spanning several years.
Numerous studies have examined dust impacts on solar power generation, most relying on observations from solar farms or experimental sites. However, such observations remain scarce and often cover short time periods, particularly in data-scarce regions thus preventing comprehensive dust impact assessments. Dust simulation models offer an alternative approach: they enable the reconstruction of dust accumulation dynamics and their impacts on power production from meteorological data over extended periods.
This simulation approach was applied by Isaacs et al. (2023) for West Africa with atmospheric reanalysis (MERRA-2) and satellite-derived data. However, the extent to which input data and modelling choices may influence the conclusions of simulated estimates remains unclear. Reanalysis products are subject to substantial uncertainties and errors, especially in regions where ground-based observations used for their development are scarce. Dust models also typically rely on simplified process representations and poorly constrained parametrizations.
In this study, we introduce PVWAT, a simple dust simulation model developed for dust impact assessment as part of the ANR-funded NETWAT project, which examines water-energy nexus challenges in West Africa. Linking different sub-models from literature, it uses meteorological inputs from on-site observations or atmospheric reanalysis to simulate time series of dust deposition fluxes, deposited dust amounts and the resulting soiling losses.
We then use PVWAT to demonstrate how simulated dust impacts depend on input data and modeling choices. For this, we consider West Africa, a hot spot for dust-related PV production losses. The region's high solar potential and unmet energy demand are expected to drive large PV expansion in the coming years (10+ GW of solar capacity by 2030; IRENA, 2023) but the region borders the Sahara and Bodélé depression, the world's most prolific dust source. Our analysis considers three sites along a north-south transect, representing contrasting dust conditions, climates (arid to humid), and land covers (savanna to tropical forest), in order to draw recommendations for diverse solar production contexts.
Through systematic sensitivity analysis, we perturb model parameters up to 8× and meteorological variables up to 2× to quantify their effects on long-term soiling ratios. This reveals the dominant sources of uncertainty and assesses how the model responds to parametric versus variable perturbations across contrasting sites.
References
International Renewable Energy Agency. (2023). Scaling up renewable energy investments in West Africa. https://www.irena.org
Isaacs et al., 2023. Dust soiling effects on decentralized solar in West Africa. Applied Energy, 340, 120993. https://doi.org/10.1016/j.apenergy.2023.120993
How to cite: Banigo, A. T., Crochemore, L., Hingray, B., Marticorena, B., and Anquetin, S.: Simulating Atmospheric Dust Impact on Photovoltaic Performance: A sensitivity analysis to guide modelling choices in a data scarce region, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7339, https://doi.org/10.5194/egusphere-egu26-7339, 2026.
Aim and Approach
Increasing pressure on water resources, driven by the growing demand for drinking water, irrigation, and industrial uses, calls for improved water management strategies (Unesco, 2024). In this context, floating photovoltaic (FPV) systems have emerged as a promising solution. Initially developed to address land-use constraints, FPV installations also present a substantial potential for reducing evaporation losses from the reservoirs on which they are deployed (Sahu, 2016). By partially covering the water surface, these systems modify air–water interactions, reducing incoming solar radiation and altering convective heat and mass exchanges, thereby potentially limiting evaporative losses (Taboada, 2017).
However, despite this widely assumed benefit (Taboada, 2017; Gonzalez, 2025; Bontempo, 2021), evaporation reduction induced by FPV systems has not yet been robustly demonstrated or quantified at the scale of industrial installations. This lack of large-scale assessment primarily stems from the complexity of the physical processes involved, including the coupled effects of surface shading, altered turbulence, and modified atmospheric boundary-layer dynamics, which cannot be reliably captured by indirect or simplified approaches and require direct, high-resolution measurements (Tanny, 2008).
To address this gap, two eddy-covariance (EC) systems were deployed on a reservoir partially covered by an industrial-scale FPV plant (see Figure 1). This experimental setup enables a direct and simultaneous monitoring of evaporative fluxes over both covered and uncovered water surfaces, providing new insights into the impact of FPV installations on reservoir-scale evaporation dynamics.
Figure 1 - Location of EC measurements on the reservoir partially covered by an FPV power plant.
Results and Perspectives
Figure 2 presents the daily evaporation rates measured for several days in July over the covered area (EC) and the adjacent uncovered area (EUC), and compares them with evaporation from the reservoir assuming free-water conditions (Efree, PM). The results clearly indicate a substantial reduction in evaporation over the partially covered reservoir compared to the free-water reference.
Over the full observation period (2025-05 to 2025-10), an average evaporation reduction of 44% was observed above the FPV-covered area. More unexpectedly, this reduction extends beyond the direct footprint of the FPV installation. Evaporation over the uncovered area is also significantly reduced, with a mean decrease of 35%. This finding is particularly significant, as it challenges the common assumption in the literature that covered and uncovered areas behave as weakly coupled systems. Instead, our results reveal a strong coupling between these zones, indicating that FPV installations induce non-local modifications of the surface–atmosphere exchanges that affect evaporation at the reservoir scale.
Building on these observations, the next objective is to identify the key physical drivers controlling evaporation under FPV deployment and to explain the observed differences. Ultimately, this work aims to develop a simplified, physically based model capable of estimating evaporation losses from reservoirs partially covered by FPV systems.
Figure 2 - Daily mean evaporation from several days of 2025-07 over covered (orange) and uncovered (blue) areas. Gray bars correspond to the estimated free-lake evaporation of the reservoir.
How to cite: Berlioux, B., Le Berre, R., Ferrand, M., Knikker, R., and Pabiou, H.: Impact of floating photovoltaic power plant on reservoir evaporation: insights from eddy-covariance measurements, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11025, https://doi.org/10.5194/egusphere-egu26-11025, 2026.
In highly renewable power systems, weather-related variability increasingly translates into system stress. Dunkelflauten are multi-day to multi-week periods of unusually low wind and solar that can span multiple countries. These renewable energy droughts significantly shape storage needs, installed capacity, and transmission requirements, far more so than average conditions. Yet, simulated renewable output is highly sensitive to assumptions regarding meteorology, spatial layout, and plant‑level effects, complicating the detection of these extremes.
We address this by quantifying renewable output in a one-at-a-time sensitivity analysis varying bias-correction methods, spatial representation, technology settings, and wake-loss assumptions. Using hourly reanalysis data, we compute country‑aggregated wind and solar generation for 80 historical weather years to evaluate impacts on annual capacity factors, drought frequency and duration for wind and solar separately. These drought metrics are then linked to system outcomes by running PyPSA‑Eur for five critical weather years under a net-zero scenario, assessing changes in optimal capacities and system‑defining stress events. We find that capacity factors are driven mainly by technology specification, with bias correction exerting little influence on solar means and a moderate effect on wind, while spatial capacity layouts appear negligible for solar but more consequential for wind. Quantile‑mapping bias correction modestly improves energy drought detection, and certain technology configurations reduce risk of low‑generation. At the system level, these differences re‑order stressful periods and shift optimal capacity across technologies and regions.
By identifying the modelling choices that have the greatest impact on energy‑drought detection and associated system stress, this study helps strengthen power system resilience to weather extremes and can improve resource‑adequacy planning for a fully renewable European system
How to cite: Karkossa, L., Grochowicz, A., and Victoria, M.: How Technology and Modelling Choices Shape European Wind and Solar Energy Droughts and Stress Events, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3024, https://doi.org/10.5194/egusphere-egu26-3024, 2026.
To reach the goal of net-zero emissions and carbon neutrality, the European power system is changing towards more variable renewable energy production. The increasing share of weather-dependent energy production however, makes it more challenging to maintain a stable grid frequency. This results in larger penalties for energy producers contributing to instability.
Norway is well connected to the European energy system, exposing it to market conditions in neighboring countries. To ensure a stable grid frequency, the national transmission system operator is responsible for balancing production and consumption volumes. To support the balancing operations, all power producers must submit day-ahead production forecasts. Deviations from the predicted volumes are subject to imbalance fees. In addition, power producers need to buy and sell energy in a dedicated market to balance deviations. To avoid large imbalance costs and support grid stability, accurate high-resolution day-ahead production forecasts are essential.
In Norway, the largest variable renewable energy source is run-of-river hydropower. Forecasting run-of-river hydropower production is equivalent to forecasting streamflow. The industry has expanded rapidly lately, resulting in many newly commissioned plants with limited streamflow observations. Thus, there is a need for a forecasting model that can make accurate predictions with limited training data.
In this study, we explore the potential of using a Long Short-Term Memory neural network to forecast hourly streamflow. The model is trained on historical data from 215 Norwegian gauging stations. To improve training efficiency, we adopt a multi-frequency approach in which earlier time steps are processed at a daily resolution, while more recent inputs retain their original hourly resolution. We explore two approaches of improving model performance: including data from 139 run-of-river hydropower plants during training and including streamflow estimated from production data through a data assimilation approach.
The results show that both approaches improve the performance of the model and the final model outperforms both a persistence model and one of the leading providers of run-of-river hydropower production forecasts in Norway. The potential economic value of the improved day-ahead forecast is estimated on the basis of both reduced imbalance fees and reduced exposure to volatile prices in the balancing market. This shows that the model we propose has the potential to improve upon existing models and contribute to overall grid stability.
How to cite: Wergeland, K., Wulff, C. O. W., and Sorteberg, A.: The Economic Benefit of AI-Driven Day-Ahead Hydropower Production Forecasts, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20893, https://doi.org/10.5194/egusphere-egu26-20893, 2026.
A renewable energy resource with much potential, yet which is often overlooked in energy roadmaps, is tidal energy (both kinetic energy – tidal stream, and potential energy – tidal range). Tidal energy is particularly attractive in the energy mix due to its predictability. However, it is as yet under-developed globally, particularly the tidal stream resource. Morlais (“voice of the sea” in Welsh) is a 35 km2 grid-connected tidal energy site in the Irish Sea, UK. Although the site has the potential for 240 MW of tidal stream energy, currently developers have agreed 38 MW of electricity at a Strike Price of £261/MWh. To aid development of the site, we have conducted measurement campaigns over the last decade, including complete multibeam coverage of the 35 km2 site and the deployment of eleven acoustic Doppler current profiles (ADCPs), along with additional wave buoy and meteorological measurements.
Peak (spring) undisturbed power density exceeds 10 kW/m2 over much of Morlais, with the most energetic locations closest to the shore — facilitating power export to the grid. There is a large submerged sand bank extending from a major headland (South Stack) which is responsible for some of the most energetic tidal streams. This sand bank has a width of around 300 m, rises around 20 m compared to the surrounding sea bed, and there is evidence that it produces secondary flows that have been observed at many of the ADCP moorings. There is significant interaction of waves and currents across Morlais. However, this mainly influences wave properties, which could affect maintenance of moorings or devices (due to increased wave steepness), rather than directly influencing the tidal energy resource. There are large variations in flood/ebb asymmetry across the site, and this can largely be explained by the phase relationship between the principal lunar semidiurnal constituent M2 and its first harmonic, M4. Although prominent tidal energy test sites (e.g. EMEC in Orkney) also exhibit strong tidal asymmetry, it could be more of an issue for a commercial site like Morlais since it affects the timing of power export to the grid.
How to cite: Neill, S. and Chisholm, J.: Tidal Stream Energy Resource – a case study at grid-connected Morlais, Irish Sea, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3569, https://doi.org/10.5194/egusphere-egu26-3569, 2026.
Ireland is the first landmass between the northeast Atlantic Ocean and Europe; its geographical location endows it with high energy potential in the ocean. At present, around 40% of the energy in Ireland is generated by renewable technologies, the majority of which is produced by onshore wind turbines. However, given the marine energy potential, the integration of wave energy converters into the future energy mix is a plausible proposition. This integration would result in a more stable energy mix, which is not reliant upon a single renewable resource. This study aims to analyse the wave power potential in Ireland by employing long-term satellite altimeter and weather buoy observations. Satellite altimeters offer long-term measurements, and cover a broader area compared to weather buoys. Data spanning from 1995 until 2024 is employed, encompassing a total of 30-years. This timeframe is particularly pertinent in the context of the analysis, as it encompasses the projected lifespan of the wave energy converters to be potentially installed. The goal is to assess whether decadal variations on the available wave power would affect the devices’ performance. To this end, wave power variation maps are generated. As satellite altimeters do not measure wave period, a regression following the method proposed by Gommenginger et. al. (2003) is employed to estimate the zero-crossing wave period by relating the significant wave height and the back-scatter coefficient measured by the altimeters, with Irish moored buoy observations. The wave energy period is obtained from wave period ratio maps created from results by Haoyu Jiang et. al. (2022); ultimately wave power is calculated assuming deep water and panchromatic seas. Decadal variations are calculated with via Theil-Sen approach, after removing the lag-1 autocorrelation, and the statistical significance its evaluated by means of the Mann Kendall test.
How to cite: Martinez Iturricastillo, N., Ulazia, A., Ponce de León Alvarez, S., and Ringwood, J. V.: Decadal wave power variability form satellite altimetry in Ireland, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21779, https://doi.org/10.5194/egusphere-egu26-21779, 2026.
Due to the ongoing energy transition to variable renewable energy sources, climate variability plays a central role in energy system studies. Climate science routinely addresses this variability by simulating large ensembles spanning hundreds to thousands of model years. However, energy system and power-grid models used for industrial applications are computationally intensive and typically cannot process more than a few years to a few decades of climate data. This mismatch necessitates the selection of a small but representative subset of climate years.
A common workaround is the use of composite or “typical” meteorological years constructed from individual months. While computationally efficient, such synthetic time series disrupt temporal coherence, and fail to capture memory effects that are critical for adequacy assessments, such as storage dynamics of hydropower. As a result, many energy system studies instead select a limited number of complete climate years, typically ranging from one to fifty. Selecting such subsets from large climate simulations constitutes a combinatorial optimisation problem: choosing X years from N>>X, for which brute-force optimisation is computationally infeasible due to ‘combinatorial explosion’.
Current practices rely heavily on (pseudo) random sampling or heuristic selection methods, including clustering-based approaches such as k-medoids (or k-means). While useful, these methods provide no guarantee of near-optimal solutions and often struggle to balance representativeness across multiple, interacting climate variables relevant for energy systems.
In this study, we systematically review existing climate-year selection methodologies and introduce simulated annealing as a flexible and computationally efficient optimisation framework for selecting representative subsets of complete climate years. The method targets representativeness of the joint distribution of multiple energy generation and demand variables. We apply the approach to the Pan-European Climate Database, which comprises 85 years of simulations from six CMIP6 climate models under four SSP scenarios, together with associated energy demand and renewable generation time series. Two use cases are considered: the selection of a larger subset of 30 representative years for adequacy-type studies, and a smaller subset of 5 years for investment-type studies. Across both cases and for both national and contintental-scale applications, simulated annealing consistently outperforms existing methods, proving to be the most robust method for climate year selection in large-scale energy system modelling.
How to cite: van Duinen, B., van der Wiel, K., and Stoop, L.: Selecting representative climate years for national to continental-scale energy system studies, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1873, https://doi.org/10.5194/egusphere-egu26-1873, 2026.
Future energy systems increasingly rely on weather-driven variable renewable energy (VRE) sources. As a result, the accuracy, resolution, and statistical consistency of meteorological inputs have become key considerations in energy system modelling (ESM). In particular, wind power estimates strongly depend on local wind speed characteristics, including both distributional properties and temporal variability. However, existing wind datasets at continental to national scale often lack sufficient spatial detail, exhibit systematic or statistical biases, or are insufficiently validated against observations. As a result, substantial uncertainty is introduced into wind energy assessments and system-level analyses. To address these limitations, we develop a framework for generating high-resolution hourly wind speed time series for Europe by combining distributional information with statistical downscaling techniques. We estimate a two-parameter Weibull distribution for each region using linear regression across multiple gridded products, including the Global Wind Atlas, ERA5 and E-OBS. The distribution is then evaluated using leave-one-out cross-validation against station measurements. In a second step, we use the validated Weibull distributions to bias-correct and downscale existing wind speed time series using several statistical downscaling approaches. Using station data as an observational benchmark, we assess the accuracy of the reconstructed time series and quantify the structural uncertainty associated with wind speed inputs derived from gridded datasets. The resulting high-resolution, bias-corrected wind speed products provide more robust meteorological inputs for renewable energy system modelling, improving estimates of wind power generation potential and supporting more reliable long-term system planning across Europe.
How to cite: Scheiber, F., Wehrle, S., Nutz, M., Grabner, I., and Schmidt, J.: Bias-Corrected High-Resolution Wind Speed Time Series for Renewable Energy System Modelling, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7222, https://doi.org/10.5194/egusphere-egu26-7222, 2026.
The expected lifetime energy yield of wind turbines and wind farms is to a large extent determined by the wind climate in which they operate. Importantly, the wind climate of the coming 25 years might differ significantly from that of the past 25 years as a consequence of natural climate variability and/or anthropogenically forced climate changes. Research on the uncertainty in future wind resources often relies on bias-corrected surface wind output from General Circulation Model (GCM) projection ensembles. However, for locations in complex terrain, the accuracy of modelled near-surface winds by these GCMs may be severely impacted by their coarse grid resolution and therefore also the associated wind climate change signals. Here, we assess the added value of a statistical GCM downscaling algorithm which employs GCM output from higher atmospheric levels as predictors. More specifically, we compare it to the standard, surface wind-based approach for a Chilean wind farm located in complex terrain. Furthermore, we assess the performance sensitivity to the choice of statistical model, predictor set, training data and temporal resolution. Finally, we apply both approaches to a GCM projection ensemble to illustrate the necessity of more advanced approaches for quantifying the future wind resource uncertainty for sites in complex terrain.
How to cite: Borgers, R., Abiven, C., Buckingham, S., and van Lipzig, N.: Quantifying future wind resources in complex terrain using data-driven predictions from large-scale GCM inputs, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7808, https://doi.org/10.5194/egusphere-egu26-7808, 2026.
Ensuring reliable energy supply and infrastructure resilience in Africa requires renewable energy (RE) deployment that takes into account the continent’s pronounced weather variability. Here, we introduce a weather-aware framework that integrates multi-criteria decision analysis with assessments of meteorological variability to optimize renewable site selection. Optimal solar and wind energy deployment locations are identified using an adapted methodology. These sites are chosen not only by their highest average resource potential but also by evaluating weather variability at each location. We provide insights into generation variability from these optimal deployment sites under major climate oscillations, including the Madden–Julian Oscillation (MJO) modulated by the El Niño–Southern Oscillation. In addition, a set of novel Africa-centric synoptic regimes (AORs) are derived through Self-Organizing Map cluster analysis, providing insight into region-specific drivers of variability that are often missed by global modes like the MJO. Detailed country-level renewable yield estimates under these dominant meteorological patterns are provided along with their frequencies of occurrence. Our findings highlight a critical need for sub-seasonal to seasonal (S2S) forecasting of these regimes to enhance system resilience. While AORs linked to large-scale oscillations like the MJO may inherit its known predictive skill, the predictability of more localized African regimes remains a critical challenge. By explicitly linking generation variability from optimized RE deployment locations to underlying climate drivers, this framework offers a robust pathway for optimizing RE expansion across the continent.
How to cite: Kurup, R., Bloomfield, H., Tiwari, P., Acharya, N., and Hesse, E.: Advancing Resilient Renewable Energy Deployment in Africa: A Weather-Aware Optimization Framework, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3066, https://doi.org/10.5194/egusphere-egu26-3066, 2026.
How to cite: Strnad, F., Schmidt, J., Mockert, F., Hennig, P., and Ludwig, N.: Assessing the risk of future Dunkelflaute events for Germany using generative deep learning, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10500, https://doi.org/10.5194/egusphere-egu26-10500, 2026.
As Africa accelerates to become one of the world’s largest integrated electricity markets and sets the target to increase renewable generation capacity, the continent’s power systems are becoming more vulnerable to extreme weather and climate events. With growing shares of renewable resources in the power mix, events such as heatwaves, periods of low wind and solar availability or prolonged hydrological drought periods —so-called energy droughts— threaten to challenge the continent’s power system resilience. However, little is known about how these climate extremes interact with Africa’s rapidly evolving power infrastructure. In this study, we identify and characterise the climate extremes that could impact future African power systems. By integrating the power system design from the African Continental Masterplan with decades of weather and climate data, we examine how variability in wind, solar and hydropower generation, coupled with temperature-driven demand peaks, shape periods of power system stress. Power system stress is measured through load shedding in high resolution dispatch simulations, developed in the PyPSA modelling framework. We will evaluate scenarios of increasing inter- and intra- regional connections between power pools to investigate whether interconnection alleviates power system stress periods by leveraging Africa’s diverse resource potential and complementary spatio-temporal profiles. In this way, this research aims to inform energy planners and policymakers about strategies that enhance the resilience of Africa’s future power systems to climate extremes, ensuring sustainable electricity supply under a changing energy and climate landscape.
How to cite: Mast, T., Sterl, S., Thiery, W., and Gupta, R.: Modelling the impacts of climate extremes on Africa's future power systems, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-544, https://doi.org/10.5194/egusphere-egu26-544, 2026.
Posters on site: Mon, 4 May, 16:15–18:00 | Hall X4
Climate models become increasingly sophisticated over time, capitalizing on better modeling techniques, process understanding and computational power. Energy systems become more exposed to climatic changes owing to the increased deployment of weather-dependent renewables as well as heating and cooling systems. There is thus an urgent need for improved usage of climate model simulations in the energy sector.
Here, we present dedicated hourly climate model simulations with CESM2 and a new pipeline to translate climate model output to renewable generation timeseries and heating/cooling demand. We showcase the Climate2Energy workflow that combines bias-correction with existing open-source tools for individual energy sector components (GSEE, windpowerlib, demandninja). We include all relevant types of renewable generation, namely onshore wind, offshore wind, PV, hydropower, and heating/cooling demand in a consistent and synchronized manner. In contrast to assessments drawing from published climate datasets such as CMIP and EURO-CORDEX, we can use non-standard climate model outputs, such as model level winds, air densities, and river discharge.
Using the SSP370 scenario and sampling different phases of the North Atlantic Oscillation to account for climate variability, our results reveal strongly altered future heating (up to 50% reductions) and cooling demand (up to 20-fold increases). In line with previous studies, the impacts on renewable generation are substantially smaller in terms of mean capacity factors. For instance, onshore wind potentials drop by a few percent in many countries while PV potentials increase by similar amounts. More pronounced changes manifest, for example, in the seasonal cycle and in inter-technology complementarity. Furthermore, stochastic optimizations with AnyMOD reveal that a future cost optimal power system looks substantially different from a current one.
Overall, our results underline the need for further analysis of the combined effects of climate change on energy systems. We provide the Climate2Energy pipeline and the data with an open license, aiming to contribute to better and more standardized climate change impact assessments in the energy sector.
REFERENCE
Wohland, J. et al. Climate2Energy: a framework to consistently include climate change into energy system modeling. Environ. Res.: Energy 2, 041001 (2025) https://doi.org/10.1088/2753-3751/ae2870
How to cite: Wohland, J., Bloin-Wibe, L., Fischer, E., Göke, L., Knutti, R., De Marco, F., Beyerle, U., and Savelsberg, J.: Climate2Energy: a framework to consistently include climate change into energy system modeling, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3461, https://doi.org/10.5194/egusphere-egu26-3461, 2026.
Australia’s National Electricity Market (NEM) is in a period of transition. Decarbonization pressures, regulatory incentives, and consumer preferences are driving up the share of renewable generation in the NEM. Concurrently, the nation faces pressure to adapt to a changing climate and the extreme weather that entails. As renewable penetration increases variability of electricity supply, climate change reduces the predictability of the weather that fuels renewables. Extreme weather events are changing; heatwaves are getting more severe, more frequent, and lasting longer. While the physical processes caused by heat on generation technologies are well defined, quantifying and predicting the systemic impacts of extreme events is an ongoing line of inquiry. Modern electricity markets are relatively young and have evolved rapidly. Generally, market datasets are short in duration, poorly standardised, and have limited coverage relative to meteorological data. They are rarely publicly available, as data publication could be considered a risk to the interests of market participants. This presentation utilises Australia’s National Electricity Market’s (NEM) Market Management System Data Model, alongside the BARRA-R2 regional climate reanalysis, to analyse historical changes in the diurnal generation profiles of wind energy during heatwaves. By bootstrapping composite generation profiles of heatwave and baseline summer days, we present how heatwaves impact generation profiles. We then compare how these profiles vary through time and space. Impact curves (bootstrapped difference curves of heatwave and baseline generation) are calculated and used to analyse patterns of heatwave impact across the NEM using principal component analysis and timeseries clustering. Investigating the scales and patterns of heatwave impacts reveal the weather-scale drivers of generation variability. This allows us to identify how large-scale synoptic systems (such as heatwaves) have myriad localised impacts. We then discuss how these localised variations may contribute to larger shifts in generation dispatch and grid stability on heatwave days. This exploratory data analysis leverages a recent, unexplored dataset to develop methods that quantify the impact of heatwaves on wind generation. The primary contribution of this research is methodological; it also offers exploratory empirical findings, highlighting areas for further research.
How to cite: Salpadoru, M., Perkins-Kirkpatrick, S., Sturmberg, B., and Lu, B.: Changes in diurnal wind generation during heatwave events, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3213, https://doi.org/10.5194/egusphere-egu26-3213, 2026.
Coastal regions in India possess an exceptional wind energy potential, exceeding 8,000 MW, with wind speeds ranging from 6.8 to 7.1 m/s. However, these areas face critical data gaps in wind monitoring networks due to sparse instrumentation, station failures, and disruptions from tropical cyclones that frequently impact India's eastern coast. Accurate, high-resolution wind field data is essential for renewable energy planning, infrastructure resilience assessment, and identifying optimal sites for wind farm development in cyclone-vulnerable regions. This study presents novel approaches for filling spatial wind field gaps. We used two approaches based on Multiple-Point Statistics (MPS), which reconstructs wind patterns by learning spatial relationships from training images, and Deep Learning (DL) using ConvLSTM2D neural networks. We apply these methods to ERA5 reanalysis data at 25 km resolution spanning the Andhra Pradesh region. Two gap scenarios were tested: (i) systematic contiguous gaps, and (ii) random scattered gaps using MPS and DL methods. Preliminary results indicate that the MPS approach yields a Pearson correlation of 0.40 with a mean absolute error (MAE) of 0.42 m/s for contiguous gaps and a Pearson correlation (r) of 0.97 with an MAE of 0.34 m/s for random gaps. The DL method for both random and contiguous gaps exhibit better performance, with r > 0.998 and MAE < 0.16 m/s. Ground-based validation with operational wind farm data remains necessary to confirm site-specific accuracy for practical wind energy applications. These gap-filled wind datasets enable the identification of optimal wind farm locations and support climate risk assessments for existing renewable infrastructure and enhance resilience planning against tropical cyclone hazards.
Keywords: Wind field, Multiple-point statistics, Deep learning, Renewable energy.
How to cite: Prajapati, P., Pandit, S., and Jha, S. K.: Novel approaches for filling gaps in the spatial wind field in the coastal regions of India, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1045, https://doi.org/10.5194/egusphere-egu26-1045, 2026.
Global reanalysis products are indispensable for reconstructing historical meteorological conditions and are crucial particularly for estimation of solar and wind energy resources. Although previous studies have evaluated reanalysis performance for individual resources or regional biases, systematic assessments of their capacity to simultaneously simulate solar and wind energy as well as their complementarity remain limited. This study evaluates the performance of ERA5, MERRA-2, and JRA-55 in estimating solar and wind energy resources across China during 1980–2022, with the help of ground-based observations as a reference. Results show that ERA5 displays superior overall performance in reproducing spatiotemporal patterns of solar and wind energy. Reanalysis products generally reproduce interannual variations and declining trends in solar energy, none fully capture the observed “decline-then-recovery” pattern in wind energy. ERA5 also demonstrates a strong spatial consistency with observations in representing solar-wind complementarity at daily to monthly scales. At the annual scale, ERA5 performs best in representing solar-wind complementarity in southern China, while MERRA-2 overperforms in northern China. This study calls for caution in interpreting solar–wind complementarity in existing studies that rely solely on reanalysis products and provides guidance for their applications in supporting solar and wind energy planning and management.
How to cite: Song, X., He, Y., Zhang, M., Zhou, J., and Zhou, Y.: Can Reanalysis Products Reliably Represent Solar and Wind Energy Resources and Their Complementarity over China, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4858, https://doi.org/10.5194/egusphere-egu26-4858, 2026.
How climate impacts energy is widely recognised as a complex research area, given the diversity of phenomena and spatiotemporal scales at which climate and weather patterns influence the energy sector. While climate models have contributed to understanding climate risk in renewable energy implementation, the systematic use of large ensembles in the climate-energy interface still needs further assessment. This study therefore aims to evaluate changes and internal variability of the main resources for solar photovoltaic, wind, and hydropower energy generation based on large ensembles. To this end, we focus on the historical and SSP3-7.0 experiments from four Single Model Large Ensembles (SMILEs) that provide at least 40 realizations: CESM2, MPI-ESM1.2-LR, ACCESS-ESM1.5 and CanESM5. We evaluate solar radiation at surface and near-surface wind speed, and runoff across the globe because they are the primary resources for renewable energy generation. Given the different number of realizations per model, we identify the optimal ensemble size to assess trends and internal variability following the approach of Milinski et al. (2020). As suggested therein, we use the pi-control simulation and extract 200, 100, and 40 time series of 20-year duration that we consider as different realizations of each model. We report that the optimal number of realizations varies depending on the variable, region, and maximum number of realizations available. For example, starting from a 100-member ensemble, the optimal number of realizations to assess internal variability in solar radiation can reach up to 60 for some models while 40 are sufficient for runoff. Our findings provide additional insights into renewable energy resource changes around the world by leveraging multiple realizations of GCMs, which can increase our understanding of the impacts of climate variability and change on renewable energy resources. These results highlight the need to carefully consider the number of realizations when assessing large ensembles.
Reference: Milinski, S., Maher, N., & Olonscheck, D. (2020). How large does a large ensemble need to be?. Earth System Dynamics, 11(4), 885-901. https://doi.org/10.5194/esd-11-885-2020
How to cite: Correa-Sánchez, I. C. and Wohland, J.: Leveraging large ensembles for renewable resource assessments: how to subselect?, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5258, https://doi.org/10.5194/egusphere-egu26-5258, 2026.
We present the Microwave Radiometer for the Detection and Assessment of Offshore Wind Resources (MiRadOr) project, a year-long offshore measurement campaign designed to evaluate how microwave radiometer (MWR) technologies can improve the assessment of offshore wind resources. MiRadOr evaluates vertical profiles of temperature and humidity and compares them with traditional radiosonde and meteorological mast observations, as well as output from numerical weather prediction (NWP) and climate models.
The overarching goal of MiRadOr is to better characterize the dynamics of the lowest levels of the atmosphere in the context of wind energy. We will evaluate the quality and reliability of MWR observations for assessing atmospheric stability- a key metric for wind energy applications.
In November 2025, the MiRadOr project completed a week-long measurement campaign with an intensive radiosondes program, LiDAR measurements, and a 200m-tall met mast in Northern Germany. MiRadOr’s one-year measurements with MWRs and LiDARs are carried out in the Netherlands. Our main intensive observation period in the Netherlands will take place in March 2026 and will include data collection with several MWRs, LiDAR, and a radiosonde program.
Moreover, we evaluate simulated atmospheric stability from reanalysis and weather prediction models with measurements. Ground truth is provided by LiDAR, MWR, and meteorological mast observations from the 2025-2026 MiRadOr campaigns, paired with previously existing measurement data, e.g., from the 2021 FESSTVaL campaign and the Tall Tower Dataset. We assess the performance of atmospheric models against the observations concerning metrics relevant to wind energy.
How to cite: Schrepfer, J., Juchem, H., Mu, F., Shenolikar, J., Czekala, H., Gottschall, J., and Fiedler, S.: Towards a better understanding of Atmospheric stability for Wind-Energy Applications with the MiRadOr Project, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5004, https://doi.org/10.5194/egusphere-egu26-5004, 2026.
Soiling losses are a major source of uncertainty in photovoltaic energy yield, particularly in regions exposed to high aerosol concentrations and intermittent precipitation. These losses are strongly modulated by meteorological conditions making their quantification a key challenge in energy meteorology. Estimating soiling losses is challenging due to complex interactions between deposition processes, cleaning events such as rain, wind-driven dust transport, and proximity to local aerosol sources.
Soiling losses can be derived from irradiance measurements using paired modules subjected to differing cleaning schedules. In this work, one year of measurements from monitoring networks in West Africa and Pacific islands are used. Meteorological drivers are extracted from ECMWF reanalysis products, including precipitation and particulate matter.
We evaluate two widely used semi-physical soiling models as benchmark, HSU and Kimber, and develop a hybrid physical-machine learning framework that integrates a physics-based empirical model with XGBoost trained on meteorological reanalysis data. Model performance is assessed using temporal cross-validation across all stations and a leave-one-out approach to evaluate spatial portability, followed by an application to a real-world photovoltaic case study in Mali.
The hybrid model significantly improves soiling losses estimation compared to semi-physical benchmarks across most sites. However, its performance decreases in environments characterised by persistently low soiling, highlighting the importance of physical constraints for extrapolation beyond the training domain.
These results highlight the potential and limitations of hybrid physical-machine learning approaches for meteorology-driven soiling assessment, supporting maintenance decisions and photovoltaic energy yield optimization.
How to cite: Turpin, M., Dalmard, A., and Schmutz, N.: Hybrid physical-machine learning estimation of photovoltaic soiling losses from meteorological reanalysis data in Africa and Pacific islands, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6300, https://doi.org/10.5194/egusphere-egu26-6300, 2026.
Combining Climate Digital Twin (DT) and Extremes DT offers significant benefits for renewable energy planning. Climate DT provides long-term simulations, while Extremes DT focuses on detecting high-impact events. Although Climate DT includes wind energy aspects, it lacks emphasis on extreme events. Integrating both approaches can address uncertainties in renewable energy supply under current and future climate conditions, as PV and wind are highly sensitive to short-term changes. Within the presented study we aim to evaluate the added value of downscaling Climate DT data from ~5 km to hectometric resolution (400–800 m) to better represent local conditions. Further, we analyse the usability of Climate DT output for the renewable energy sector, either directly or as stated above, as input for setting up dynamical climate simulations at the hecto-scale by using regional climate simulation models WRF and ICON. We therefore have the following objectives: (i) to assess the skill of the Global Climate DT scenarios with respect to representativeness of extreme (meteorological) events, synoptic patterns, and their impact on renewables; (ii) estimate the added value of highly resolved climate scenarios dynamically downscaled to hectometric spatial resolution (and higher vertical resolution) with respect to selected renewables extreme events (negatively affecting either the supply or the infrastructure itself).
For assessing the added value of hecto-scale simulations, on the one hand, regional climate simulations using the WRF and ICON model were conducted – initiated by ERA5 data – for 5 km, 1.6 km, 800 m and 400 m. These simulations display that higher regional climate model resolution from 5 km down to 1.6 km, to 400 m increases the model skill to represent local wind patterns.
On the other hand, to evaluate the skill of Climate DT versus hecto-scale simulations initialized by Climate DT, a model year representative of a real year is selected and simulated using ICON and WRF. Consequently, the meteorological parameters (e.g. wind speed, radiation, temperature) as well as the post-processed energy production (e.g. mean annual and mean monthly values) data are validated.
How to cite: Bügelmayer-Blaschek, M., Baier, K., Gazzaneo, P., Hasel, K., Lexer, A., and Schicker, I.: From Climate DT to Hectoscale Forecasts For Renewable Energy Systems , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6521, https://doi.org/10.5194/egusphere-egu26-6521, 2026.
In addition to wind speed, turbulence intensity (TI) is a key atmospheric variable in the wind energy sector, as it affects both mechanical loads on wind turbines and their power production. Enhanced turbulence levels can increase structural fatigue and wear, while also influencing electricity generation. Consequently, reliable measurements and forecasts of wind speed and TI are essential for technical planning, safe operation, and accurate power yield forecasting for grid integration.
The German Meteorological Service (Deutscher Wetterdienst, DWD) runs the ICOsahedral Nonhydrostatic (ICON) model operationally as a numerical weather prediction model with horizontal grid size resolution of 2.1 km. This model provides wind data and subgrid-scale turbulent kinetic energy (TKE) using the TURBDIFF turbulence parameterization scheme. In parallel, Doppler Lidar (DL) systems are deployed at DWD's Lindenberg Meteorological Observatory to measure wind and turbulence profiles, including TKE, within the lowest 600 m of the atmospheric boundary layer. TI can be derived from both model output and observations by combining wind speed and TKE, enabling an evaluation of ICON with respect to wind-energy-relevant parameters.
In the presented study the model results for wind speed and TI from ICON simulations are compared with DL measurements over a five-day period with a typical summertime convective boundary layer evolution during daytime, while low level jets (LLJ) were observed during nighttime. Although the comparisons show reasonable overall agreement, it also becomes clear that uncertainties in both variables vary depending on atmospheric stratification.
In addition, the results of theoretical investigations into the potential benefits of using ICON forecasts of wind speed and ambient TI as inputs for wind energy power forecasts are presented. For this purpose, a performance model with a single turbine was used, which was driven with measured and simulated wind speed and TI in order to estimate the power output. The respective power outputs were compared with each other and the results suggest that incorporating TI information from ICON into wind power modelling can be advantageous, particularly under convective boundary-layer conditions. However, under stable stratification, the impact of simulated TI appears to be less significant, as uncertainties in LLJ forecasts can outweigh the effect of TI on electricity generation.
Although higher-resolution atmospheric models may better resolve ambient turbulence at rotor scales, their operational applicability is often limited by computational costs and data availability. This study therefore focuses on assessing whether freely available, operational ICON turbulence forecasts, which are available continuously and spatially consistently across Germany, can provide added value for wind energy applications under realistic practical constraints. The studies are limited to the investigation of ambient turbulence. Wake effects and turbine-turbine interactions, which will additionally occur in wind farms with more than one turbine, are not taken into account.
How to cite: Päschke, E. and Ahlgrimm, M.: Turbulence Intensity from ICON: A study of the potential for Wind Energy Applications, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9666, https://doi.org/10.5194/egusphere-egu26-9666, 2026.
Future energy systems in the Maritime Continent are expected to be increasingly dominated by solar power as part of the broader decarbonization and energy transition agenda, with substantial growth in solar potential projected across Indonesia. While overall resource availability is likely to remain high due to Indonesia’s equatorial location, climate change may increase spatial and seasonal variability in surface solar radiation through shifts in cloudiness and atmospheric circulation, underscoring the need for climate-informed energy planning.
However, the robustness of the reanalysis products under intensifying hydro-climatic extremes remains insufficiently assessed. This study evaluates the performance of the ERA5-Land reanalysis in reproducing surface solar irradiance relative to observations from 23 ground-based stations over the period 2019–2025, using the 2020–2022 triple-dip La Niña event as a natural stress test. Results are further contextualized within observed multi-decadal climate trends spanning 1981–2024.
The evaluation reveals a systematic clear-sky bias in ERA5-Land that is strongly dependent on the atmospheric regime. While the reanalysis captures the phase of the diurnal irradiance cycle reasonably well under moderate conditions, its performance degrades markedly during high-impact weather regimes. During the deep convective phases of the 2021 La Niña, in situ observations show pronounced attenuation of surface solar irradiance associated with persistent cloud cover, whereas ERA5-Land frequently maintains elevated irradiance estimates. This behavior points to limitations in the representation of cloud optical properties, especially for thick stratiform cloud decks characteristic of the Asian Winter Monsoon. As a result, ERA-5 reproduces rainfall occurrence but underestimates the magnitude of associated solar dimming, leading to a systematic overestimation of solar resource availability during periods of heightened system vulnerability, which may translate into biased generation forecasts, inadequate reserve allocation, and increased operational risk for solar-dominated power systems.
Other characteristics of climate data that are particularly relevant for future energy systems include emerging climate trends, especially those reflected in extreme climate indices. Analysis of ground stations indicates widespread asymmetric warming, with minimum temperatures increasing more rapidly than maximum temperatures, alongside a statistically significant intensification of wet extremes (RX1DAY) and changes in dry spell characteristics. The increasing prevalence of hydro-climatic extremes implies that the atmospheric regimes under which reanalysis performance is weakest are likely to become more frequent.
Overall, this study identifies a critical resilience gap in renewable energy resource assessment for the Maritime Continent. Reliance on unadjusted reanalysis data may lead to systematic underestimation of solar power drought risk. We argue that future energy planning should move beyond uniform bias correction and adopt regime-aware approaches that explicitly account for limitations in the representation of cloud-radiative processes under extreme monsoonal conditions.
How to cite: Bestari, D., Bloomfield, H., Robson, C., Fowler, H., and Putra, A. W.: Evaluating Reanalysis Reliability under Compound Climate Extremes for Energy Resilience in the Maritime Continent , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10089, https://doi.org/10.5194/egusphere-egu26-10089, 2026.
As solar photovoltaic (PV) generation becomes increasingly central to global renewable energy systems, cloud-induced intermittency of solar irradiance remains a major challenge for power system stability and economic dispatch. Due to the sparse spatial coverage of ground-based measurements, high-resolution geostationary satellite imagery (e.g., Himawari-8/9) has become essential for real-time solar forecasting. However, satellite observations provide only two-dimensional projections of integrated atmospheric optical effects, lacking explicit information on cloud vertical structure and microphysics, which fundamentally complicates the inference of physically meaningful irradiance dynamics. Despite recent advances, deep learning–based satellite forecasting methods continue to face three key limitations: limited interpretability due to black-box model structures, excessive parameterization that constrains real-time or edge deployment, and strong sensitivity to quasi-static background signals embedded in satellite imagery. To address these challenges, we propose a Physics-Constrained Latent Dynamics Framework that reframes image reconstruction as an auxiliary constraint governing latent dynamical evolution rather than a prediction target. By minimizing reconstruction errors between predicted and observed satellite images, the framework guides neural physical operators to learn physically consistent cloud motion in latent space. Inspired by PhyDNet, the model decomposes prediction into two parallel pathways: a physics-based branch that governs latent state evolution through neural physical operators, and a data-driven residual branch that compensates for non-physical visual components beyond simplified physical representations. The framework comprises three core components: (i) neural physical operators that approximate partial differential equations (PDEs) via architectural constraints in latent space, enforcing conservation and temporal continuity; (ii) a clear-sky background representation to isolate deterministic irradiance patterns; and (iii) a Global Horizontal Irradiance (GHI) prediction head. In parallel, a ConvLSTM-based residual branch captures cloud formation and dissipation, illumination variability, and sensor noise, forming a dual-branch architecture that integrates physics-based structure with data-driven flexibility. To further decouple stochastic cloud variability from quasi-static background signals, a bootstrap-based extreme-quantile method is employed to construct clear-sky deviation maps, enabling more effective separation of dynamic cloud processes. Preliminary experiments using multiple ground stations in Tokyo, Japan, demonstrate that, without direct irradiance inputs, the proposed framework achieves an R2 of 0.801 and a mean absolute error of 0.5 MJ m-2 for one-hour-ahead GHI forecasts. Analysis of the learned higher-order PDE coefficients suggests that the latent dynamics capture nonlinear physical behaviors beyond simple translational motion. Ablation studies further show that, compared with a pure ConvLSTM baseline, the proposed decoupled architecture reduces parameter counts by approximately 30% while improving forecasting performance by about 12%. While autoregressive frame-based prediction remains susceptible to error accumulation at longer horizons, ongoing work explores replacing the autoregressive formulation with Neural Ordinary Differential Equations to model temporal evolution as continuous dynamical flows, aiming to mitigate long-horizon error growth and establish a more robust foundation for physics-informed solar forecasting and dynamical analysis.
How to cite: Ding, J.-W. and Hsieh, I.-Y. L.: Physics-Constrained Latent Dynamics for Solar PV Forecasting via Interpretable Deep Learning, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13902, https://doi.org/10.5194/egusphere-egu26-13902, 2026.
Germany aims to substantially expand its offshore wind energy by 2045, increasing the installed capacity from about 10 GW today to almost 70 GW, with offshore wind expected to supply up to 25 % of the national electricity demand. Achieving this target requires the development of offshore wind in increasingly remote areas, where long-term observational reference data are scarce and meteorological and oceanographic conditions are less well understood. However, a key factor for the safe and cost-effective installation, operation, and maintenance of the offshore wind farms is the assessment of “weather windows”, defined as periods during which meteorological and oceanographic conditions like wind and waves are below the operational limits of the vessels used. The frequency and duration of such weather windows directly affect installation schedules, turbine accessibility during operation, as well as the vessel requirements, and thus the financial viability of offshore wind projects. At the same time, this has a major impact on the corresponding bids submitted in tenders of new offshore wind sites.
To achieve Germany’s offshore targets, new offshore wind sites have been tendered annually since 2021 by the Federal Network Agency, in cooperation with the Federal Maritime and Hydrographic Agency (BSH), according to the Offshore Wind Energy Act (WindSeeG). The German Weather Service (DWD) supports the BSH in compiling detailed information on the prevailing meteorological conditions at the tendered sites and in continuously providing new and improved products. The meteorological dataset for each site typically combines one year of site-specific in-situ measurements obtained with floating LiDARs and several long-term reanalysis datasets. Both provide the basis for the comprehensive report on the expected conditions at an offshore wind site. All data and reports are publicly available via the BSH’s PINTA portal – https://pinta.bsh.de.
This study presents a new comprehensive assessment of combined wind and wave conditions for selected offshore wind sites, using multi-decadal atmospheric and oceanographic reanalysis data. For the first time, the new regional reanalysis product ICON-DREAM-EU from DWD is included alongside well-established reanalysis datasets. The resulting weather windows are evaluated in terms of their frequency, duration, and seasonal variability, considering both average and extreme cases. Generic thresholds relevant to the offshore wind industry are used with a focus on near-surface wind speed and sea state.
The results show distinct patterns of favourable conditions and reveal substantial differences between the reanalysis datasets. These differences highlight uncertainties inherent in assessments based solely on reanalyses and underscore the importance for high-quality, site-specific in-situ measurements. The study supports improved planning and risk assessment for the offshore wind development and emphasizes the value of the in-situ and reanalysis data provided year after year via the PINTA portal for the energy transition.
How to cite: Möller, T., Michaelis, J., Hansen, A., Hanse, F., Spangehl, T., Hüttl-Kabus, S., Brast, M., Hahn, J., Outzen, O., Andersson, A., Grüter, M., and Kühn, B.: Assessing weather windows for the offshore wind development using combined meteorological and oceanographic reanalysis data, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11402, https://doi.org/10.5194/egusphere-egu26-11402, 2026.
With the rapid growth of wind farms worldwide, it is increasingly relevant to identify reliable measurement approaches for characterizing wind turbine wakes. Scanning wind lidars are commonly used for this purpose, but they are constrained by limited spatial coverage and can face performance challenges under certain atmospheric conditions (i.e., precipitation). In contrast, a dual-Doppler radar setup, in which two radars sample the same scanning volume, has emerged as a promising approach. It retrieves wind velocities over larger areas and can capture the spatial extent and evolution of turbine wakes, particularly during precipitation when radar returns are strongest. Recent field operations as part of the American WAKE ExperimeNt have demonstrated the value in using this dual-Doppler radar approach over a large domain encompassing several wind farms. However, there still remains uncertainty about its ability to resolve winds in different locations and under various radar configurations and atmospheric regimes.
The 2025 Krummendeich field experiment in northern Germany provided an ideal testbed to address this gap. This onshore campaign took place at a wind farm consisting of only a few turbines to target finer, turbulence-based measurements. Along with the dual-doppler setup of the radars, the site was equipped with scanning lidars, met masts, laser disdrometers, a commercial vertical profiling lidar, and other instruments. By leveraging observations collected from Krummendeich, dual-Doppler radar wind measurements can be validated against datasets previously used extensively in wind-energy research, and a systematic evaluation of this novel dual-Doppler setup can provide new insights into how its performance responds to different external factors. As part of an ongoing effort, this study examines under what conditions the dual-Doppler radar approach does and does not supply optimal data availability for resolving turbine wakes. Preliminary results suggest that data coverage and quality increases with rainfall intensity, motivating a more in-depth analysis across precipitation regimes, atmospheric conditions, and scan configurations.
How to cite: Jordan, A., Hung, L.-Y., Wolken-Möhlmann, G., and Gottschall, J.: Evaluating Dual-Doppler Radar Wind Speed Performance Under Different Scanning Strategies and Atmospheric Conditions , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12138, https://doi.org/10.5194/egusphere-egu26-12138, 2026.
Accurately representing complex spatio-temporal wind fields in mountainous terrain requires high-resolution atmospheric models, but these come with substantial computational cost. Although generally less accurate than physics-based models, machine learning-based wind downscaling offers computationally efficient alternatives for many energy-relevant applications; however, its performance depends on training data and local conditions, limiting its broad applicability.
We present an enhanced version of the deep-learning-based near-surface wind downscaling model Devine (Le Toumelin et al., 2023), trained on controlled atmospheric simulations over synthetic topographies covering a wide range of slopes and terrain features. Wind-direction-dependent descriptive features facilitate deployment across different mountainous sites. We evaluate the model using high-resolution atmospheric simulations and ground-based observations in two mountainous regions with contrasting climates and topography, performing a spatio-temporal assessment of its strengths and limitations.
The enhanced Devine model reproduces fine-scale wind patterns for terrain-induced flow as used in the training data, demonstrating transferability across mountainous sites. Its rapid generation of high-resolution wind fields enables applications such as wind resource assessment, atlas generation, climate impact studies, and short-term operational forecasts for wind farm operation. Overall, the evaluation shows how the enhanced Devine model can guide energy-related applications, indicating where it performs reliably and where caution is needed due to unrepresented wind regimes such as large-scale pressure-driven flow.
LeToumelin, L., Gouttevin, I., Helbig, N., Galiez, C., Roux, M., and Karbou, F. (2023). Emulating the adaptation of wind fields to complex terrain with deep-learning. Artificial Intelligence for the Earth Systems, 2(1):1–39.
How to cite: Helbig, N., Hammer, F., Duine, G.-J., Carvalho, L., Barber, S., and Jones, C.: Potential and limitations of efficient machine-learning wind downscaling for energy-relevant applications in mountainous environments, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14164, https://doi.org/10.5194/egusphere-egu26-14164, 2026.
Accurate estimation of wind speed at turbine hub height is a critical prerequisite for reliable wind energy resource assessment and project development. In recent years, the hub heights of modern wind turbines have steadily increased and now commonly exceed 100 m. However, direct wind measurements at these elevations remain scarce due to the high-cost constraints associated with tall meteorological masts. This observational gap introduces substantial uncertainty in pre-construction wind resource assessments, while relying solely on near-surface wind measurements often leads to significant biased estimates.
A widely adopted solution is the power-law extrapolation derived from Monin–Obukhov similarity theory, which assumes a power-law relationship between wind speed and height. Owing to its simplicity and flexibility, the power-law method has become the most extensively used approach in both engineering practice and scientific studies, as the power-law exponent can be readily derived from wind measurements at two different heights. Nevertheless, existing applications typically treat the exponent as height-invariant, overlooking its potential dependence on altitude. In reality, the power-law exponent varies with height, and directly applying the power-law exponent estimated from low-level measurements to increasingly taller hub heights results in great uncertainty.
To address this limitation, five years (2020–2024) of radiosonde observations over China were analyzed to characterize the vertical variations of the power-law exponent. Statistical results indicate a clear increasing trend with height: the mean exponents at 50 m, 100 m, 150 m, 200 m, 250 m, and 300 m are 0.130 ± 0.192, 0.162 ± 0.207, 0.179 ± 0.221, 0.188 ± 0.212, 0.194 ± 0.199, and 0.197 ± 0.194, respectively. And we found that the relationship between power-law exponents at different heights can be significantly represented by a cubic polynomial model. The fitted models between adjacent height levels exhibit high consistency, with coefficients of determination (R²) generally exceeding 0.96. For height separations of 200 m, the fitting performance remains robust (R² > 0.80), whereas larger vertical gaps lead to a noticeable decline in reliability.
Based on these findings, power-law exponents derived at lower heights—such as those obtained from short meteorological masts—can be reliably extrapolated to turbine hub heights using the proposed polynomial framework. Comparative experiments demonstrate that, relative to using a fixed exponent of 0.14 or directly adopting low-level exponents, the cubic polynomial extrapolation approach consistently achieves the highest accuracy across all combinations of height extrapolation. On average, mean absolute error and root mean square error are reduced by 74.6% and 68.0%, and by 27.9% and 25.7%, respectively. These results highlight the importance of explicitly accounting for the height dependence of the power-law exponent and demonstrate that the proposed framework offers a practical and effective solution for improving hub-height wind speed estimation, particularly in regions lacking direct wind observations at turbine hub height.
How to cite: Tong, Z., Liu, B., and Ma, X.: Improving Hub-Height Wind Speed Estimation by Accounting for Height-Dependent Power-Law Exponents, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15623, https://doi.org/10.5194/egusphere-egu26-15623, 2026.
When aggregating contiguous segments by the minimum ampacity across spans, wind’s contribution decreases with line length due to low spatial coherence: beyond ~20 km at low temperature ratings, the contribution of temperature and solar exceed wind's; for >100 km, wind’s share falls below 20%, and both mean ampacity and variance decline with length because of recurring locally calm spans. Operationally, these results support targeted wind sensing on short or high‑temperature corridors, while recognizing that ambient temperature (available without new sensors) becomes the primary driver for long, low‑rating lines.
How to cite: Rousseau-Rizzi, R.: Wind, Temperature, and Solar Contributions to Dynamic Line Rating: A Sobol Analysis from Span to Corridor Scale, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14644, https://doi.org/10.5194/egusphere-egu26-14644, 2026.
The rapid expansion of solar power capacity necessitates the development of advanced, high-resolution meteorological tools to ensure national grid stability and efficient energy market integration. This poster presents a high-visibility collaboration between the Danish Meteorological Institute (DMI) and Energinet, the Danish Transmission System Operator (TSO), focused on the end-to-end development and operationalization of a satellite-based solar nowcasting system.
The project is strategically structured into three distinct development tracks designed to modernize grid planning capabilities through improved short-term forecasts. In the first track, we have successfully transitioned a current optical-flow model based on Meteosat Second Generation (MSG) data into a live production environment. This system currently operates at a 15-minute temporal resolution and update frequency. As of early 2026, the first operational version of this system is live within a containerized Kubernetes environment orchestrated by AirFlow, which triggers automated updates every fifteen minutes. This infrastructure utilizes stable S3-storage for data handling and is transitioning from temporary researcher-led server solutions to permanent, integrated data flows. Validation of this system against current state-of-the-art operational numerical weather prediction output will be presented. The second development track involves adapting these models for the newly available Meteosat Third Generation (MTG) data, which provides significantly improved spatial resolution at 10-minute intervals. This transition includes establishing routines for skill comparison to quantify improvement over the first nowcasting system. Finally, the third track explores the development of an AI-based nowcasting model designed to learn realistic cloud development from historical MTG satellite imagery to further reduce nowcast uncertainty. This project serves as a technical blueprint for the integration of meteorological research into operational IT infrastructure to support the ongoing green energy transition.
How to cite: Kruse, I. L., Holten Møller, K., Stener Hintz, K., Vedel, H., Ankjær Borch, U., and Sommer, J.: Operationalizing Satellite-Based Solar Nowcasting: A Collaborative Partnership between DMI and Energinet for National Grid Planning , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17646, https://doi.org/10.5194/egusphere-egu26-17646, 2026.
A supervised machine-learning regression framework is presented for forecasting wind and photovoltaic (PV) power generation by integrating local and synoptic-scale meteorological information. The approach is evaluated across multiple sites, including 39 wind and 18 PV stations in Mexico, and 3 wind and 8 PV stations in China. For each station, an XGBoost regression model is trained to predict hourly energy production using local meteorological variables, derived from ERA5 reanalysis data for Mexico and on-site measurements for the Chinese stations.
To assess the added value of large-scale atmospheric information, dimensionally reduced synoptic-scale predictors extracted from ERA5 using self-organizing maps and principal component analysis are incorporated. These predictors are designed to represent dominant atmospheric circulation patterns potentially influencing local renewable energy production. Model performance is assessed through station-specific cross-validation, comparing configurations with and without synoptic-scale features across multiple predictor combinations.
Results indicate that the inclusion of synoptic-scale atmospheric patterns can improve short-term power forecasts at several locations, although the overall gains are generally modest. The analysis suggests that improvements in local meteorological inputs are likely to yield larger increases in forecast skill than further refinement of synoptic-scale representations. Nevertheless, the proposed framework demonstrates clear operational relevance: when customized for individual stations, synoptic-scale information can contribute to improved forecasting performance while maintaining the computational efficiency of machine-learning-based methods.
How to cite: Lezana Duran, F. and Ochoa Moya, C. A.: Harnessing Synoptic-Scale Information in Wind and Photovoltaic Energy Forecasting Using Machine Learning, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16324, https://doi.org/10.5194/egusphere-egu26-16324, 2026.
Hydropower provides more than half of Brazil’s electricity, making the energy system sensitive to climate variability. Recent droughts have had severe impacts on hydropower generation, including the 2014/15 event that affected production in Southeast Brazil. Heatwaves in densely populated areas of the country can also drive an increase in energy demand for cooling. The co-occurrence of these drought and heatwaves events can be considered a spatially compound event which occur when interconnected locations experience hazards concurrently, amplifying impacts beyond that of the individual hazards. Therefore, these events represent a substantial risk to the energy security of Brazil.
This study investigates how the occurrence of these drought-heatwave compound events has changed since 2004. Impacts metrics, including the Standardised Precipitation Evapotranspiration Index and Cooling Degree Days derived from ERA5, combined with energy demand and energy production data were used to investigate both the univariate and compound nature of the changes observed, and the implications this has for the energy system. The results indicate that the nature of the most extreme compound events vary, and that a compound approach offers a more comprehensive assessment of climate impacts on the energy system than a univariate approach. These findings also have the potential to aid adaptation research by providing a basis to explore how climate-energy stress events may change under future climate projections.
How to cite: Bradley, A., Hartley, A., James, R., Alves, L., and Mitchell, D.: Compound Drought-Heatwave Events: Brazil’s Energy Sector, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18742, https://doi.org/10.5194/egusphere-egu26-18742, 2026.
This work is a multidisciplinary approach that analyzes the growing vulnerability of renewable energy production systems to climate variability dominated by just a few large-scale atmospheric regimes. In particular, the Portuguese electricity system is heavily influenced by atmospheric vulnerability, due to its current high dependance on solar and wind energy, which increases the risk of energy shortages as a consequence of poor meteorological conditions, particularly in situations of low production and high demand. These episodes, known as energy compound events (ECEs), compromise the security and stability of the Portuguese energy system.
The main objective is to investigate the relationship between ECEs and large-scale atmospheric patterns, as well as to assess their evolution in the context of climate change. The methodology is structured in three phases: i) defining and characterizing ECEs in the Portuguese electricity system; ii) identifying the meteorological patterns associated with these events; and iii) assessing the impact of climate change on the frequency and intensity of these patterns. To this end, data on national electricity demand and solar energy production for the period 1989-2025 were obtained from the energy dataset of the European Copernicus Climate Change Service (C3S, https://climate.copernicus.eu/). Wind energy production was calculated from the CERRA atmospheric wind speed fields at 10 meters. The meteorological regimes affecting the Portuguese energy system were calculated from the 500 hPa geopotential height of the ERA5 database, based on the decomposition of the daily Z500 into empirical orthogonal functions and their grouping using k-means clustering. Finally, data from 14 global climate models (GCMs) obtained from the CMIP6 ensemble were used to analyze the evolution of the frequency and intensity of the identified regimes, as well as their consistency with the ERA5 observations (during the historical period) and in the future using different climate scenarios.
From the analysis, six meteorological regimes were identified as having an impact on renewable energy production in Portugal. Out of the 234 ECEs detected throughout the period, 144 occurred under the predominance of the positive phase of the North Atlantic Oscillation (NAO+), indicating an important contribution for ECEs occurrence. It is expected that the analysis of future projections will enable a robust assessment of the evolution of ECE risk in a constantly changing climate, contributing to adaptation and mitigation strategies and ensuring the reliability of energy systems.
This work is supported by FCT, I.P./MCTES through national funds (PIDDAC): LA/P/0068/2020- https://doi.org/10.54499/LA/P/0068/2020 , UID/50019/2025, https://doi.org /10.54499/UID/PRR/50019/2025 ,UID/PRR2/50019/2025
This work has also received funding from the European Union’s Horizon 2.5 – Climate Energy and Mobility programme under grant agreement No. 101081661 through the 'WorldTrans – TRANSPARENT ASSESSMENTS FOR REAL PEOPLE' project
How to cite: Ganhão, C., Molina, M., and Trigo, R.: Analysis of synoptic conditions that lead to Energy Compound Events (ECEs) in the Portuguese electrical system, in current and future climates, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19171, https://doi.org/10.5194/egusphere-egu26-19171, 2026.
To investigate the differences in microclimatic and eco-environmental effects of centralized photovoltaic (PV) power stations under diverse climatic backgrounds, high-altitude desert PV stations in Qinghai Province representing hyper-arid, arid, and semi-arid climates were selected. Micro-meteorology factors and vegetation evolution characteristics inside and outside the PV arrays were analyzed by employing paired inside-outside observations and long-time-series NDVI retrieval. It is indicated that the micro-meteorology and eco-environmental effects exhibit differential responses along the aridity gradient, with water availability identified as the core regulatory factor. A significant “heat island effect” with no vegetation recovery was observed in the hyper-arid zone; nocturnal warming and slight humidification with a trend of vegetation recovery were exhibited in the arid zone; while positive ecosystem feedback was demonstrated in the semi-arid zone, where the shading and wind-blocking effects of PV modules facilitated soil moisture conservation, leading to rapid vegetation recovery that offset physical warming through transpiration cooling. The evolutionary pattern of PV ecological effects transitioning from physical disturbance to ecological regulation is elucidated, and the feasibility of synergy between PV development and ecological restoration under suitable water conditions is confirmed.
How to cite: Cui, Y., Ma, H., Ye, D., Zhang, J., Ma, Z., and Tang, F.: A Comparative Study on Micro-meteorology and Vegetation Effects of Centralized Photovoltaic Power Stations in High-Altitude Desert Regions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21603, https://doi.org/10.5194/egusphere-egu26-21603, 2026.
Accurate short-term forecasting of surface solar irradiance (SSI) is essential for renewable energy integration and trading considerations. In operations, it enables flexibility mechanisms, provides a hedge against rapid weather transitions and overall facilitates decision-making for intraday arbitrage. The ability to anticipate rapid ramp events in particular allows for the proactive management of renewable assets, maximizing capture prices and minimizing imbalance settlements.
To this end, we present a new probabilistic framework for SSI forecasting over the contiguous United States (CONUS), developed within the NVIDIA Earth-2 platform. The framework builds upon Stormscope, NVIDIA's latest generative model for short-term Geostationary Operational Environmental Satellite (GOES) imagery forecasting, which serves as its core component. Stormscope predicts the spatio-temporal evolution of cloud fields, producing probabilistic satellite imagery sequences that capture atmospheric variability at high temporal resolution across eight spectral bands.
On top of this forecasting backbone, we apply a diagnostic diffusion model to estimate surface solar irradiance from GOES imagery using the National Solar Radiation Database (NSRDB) as reference data. This diagnostic model converts predicted satellite imagery into uncertainty-aware irradiance fields. Real-time inference is performed through Earth2Studio, providing continuous processing of live GOES data streams suitable for operational deployment.
We evaluate the system’s performance against the High-Resolution Rapid Refresh SSI forecasts, demonstrating improved skill in capturing rapid irradiance fluctuations and cloud-driven variability at short lead times. The integration of Stormscope and the diagnostic diffusion model represents a significant expansion of TotalEnergies' global weather forecast capabilities, bridging the gap between real-time and medium-range weather forecast. This work advances the reliability of solar resource prediction and contributes to improving the profitability of renewable asset portfolios in increasingly volatile merchant markets.
How to cite: Ertl, G., Carpentieri, A., Albergel, S., Benhaiem, D., Oublal, K., Guichard, M., and Le Borgne, E.: A Generative Approach for Surface Solar Irradiance Nowcasting, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21675, https://doi.org/10.5194/egusphere-egu26-21675, 2026.
Accurate and transferable photovoltaic (PV) power forecasting is essential for grid operation and energy system planning, particularly as PV installations continue to expand and energy communities increasingly rely on decentralized, locally managed generation. However, PV production is inherently site-specific, and many community-scale systems lack sufficiently long and continuous observation records to support robust data-driven forecasting approaches.
We present a scalable machine-learning nowcasting framework designed to support PV forecasting for energy communities. The approach integrates (downscaling) spatial radiation nowcasts and combining openly available meteorological data with local information to generate PV power forecasts tailored to individual PV systems or entire communities. It builds on semi-synthetic data generation and post-processing techniques and is specifically designed for data-scarce environments. Local high resolution weather prediction model output such as our in-house post-processing model INCA is used as a primary source of covariates, complemented by available satellite-derived radiation products from CAMS and reanalysis data from ERA5.
The methodology follows a two-fold strategy to address insufficient historical PV data. Where individual PV systems or communities provide a sufficient amount of measured production data for supervised learning, semi-synthetic PV time series are generated using classical approaches based on auxiliary meteorological and radiation data. In this setting, Random Forest models are employed due to their robustness for limited, seasonal datasets and their ability to capture nonlinear feature interactions without excessive overfitting. In cases where observational data are extremely scarce, an alternative strategy is applied using pre-trained foundation models. These models are driven by a set of meteorological and temporal covariates and calibrated using forecast radiation fields converted into site-specific PV power via PVLib and detailed PV meta-data (e.g. system geometry, technical parameters, and location). In both cases, semi-synthetic PV time series are effectively used to augment training data and optimize data driven nowcasting.
Model performance is evaluated across a diverse set of PV sites and compared against persistence and climatological baselines. Results indicate that semi-synthetic data combined with local covariates provide a robust approach for transferable PV power nowcasting and is useful for energy community use cases.
How to cite: Papazek, P. and Schicker, I.: Synthetic PV Data for Energy Communities , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19728, https://doi.org/10.5194/egusphere-egu26-19728, 2026.
To decarbonize its power sector, the European Union plans a major expansion of wind energy in the North Sea. However, closely spaced turbines can cause wake losses, which may aggregate at the wind farm scale and extend tens of kilometres. This study examines the economic impact of inter-farm wake effects, accounting for the correlation between wind speed and electricity prices. As a case study, we assess the planned Princess Elisabeth Zone (PEZ) and its potential impact on the existing Belgian North Sea cluster. Previous work used the meso-scale climate model COSMO-CLM with the Fitch wind farm parameterization to estimate wind farm energy production for both the current and a potential future layout that includes PEZ. The difference in energy production of the existing Belgian cluster between both runs is attributed to the PEZ’s wake effect and parameterized by wind speed and direction. The energy deficit is applied to ERA5 wind velocity time series, enabling synchronous multiplication with historical electricity prices. We find that energy is lost at about the average price at which wind energy is sold. This price is below the average market price due to a negative wind speed – price correlation. Wind farm owners may thus expect about the same relative revenue loss as their energy deficit. However, different locations for the PEZ as well as higher wind penetration in the electricity market lead to different outcomes, nuancing this statement.
How to cite: Winters, W., Borgers, R., Delarue, E., and van Lipzig, N.: Economic implications of inter-farm wake losses, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22824, https://doi.org/10.5194/egusphere-egu26-22824, 2026.
From the perspective of the energy transition in Europe, the Fit for 55 package outlines a comprehensive set of measures aimed at achieving climate neutrality by 2050 and reducing net greenhouse gas (GHG) emissions by 55% by 2030 relative to 1990 levels. In line with these objectives, the latest Italian Integrated National Energy and Climate Plan foresees a rapid expansion of renewable energy by 2030, with solar capacity rising from 37 GW in 2024 to 80 GW, and wind capacity growing from 13 GW to 28 GW.
As renewable generation expands and electricity demand rises due to increasing electrification, the power system becomes progressively more sensitive to meteorological conditions. This growing dependence highlights the need to better understand how the variability of solar and wind resources affects renewable power production throughout the year, as well as whether this variability has changed over recent decades.
In this context, weather regimes provide a valuable framework for energy system analysis, as they describe large-scale, physically consistent, and persistent atmospheric patterns that are inherently more predictable than local grid-point variables. Several studies suggest that weather-regime-based methods are more effective at predicting medium to long-term weather patterns, making them particularly useful for planning energy systems over the subseasonal-to-seasonal timescale.
Against this background, this study aims to characterize the variability of renewable energy production over the Italian peninsula as a function of weather regimes. In fact, while this approach has been widely applied in northern and central Europe—especially to investigate winter energy droughts (Dunkelflauten)—its application to Italy remains limited.
The methodology involves the estimation of solar and wind capacity factors using dedicated datasets. For solar energy, surface solar radiation from the Surface Solar Radiation Data Set – Heliosat, version 3 is combined with near-surface temperature data from the MEteorological Reanalysis Italian DAtaset - MERIDA to assess changes in solar production efficiency under increasing temperatures. Wind resources are characterized using the wind atlas Atlante EOLico ItaliANo - AEOLIAN, which provides wind speed data at multiple heights representative of wind turbine hub levels and has been specifically adapted for the Italian peninsula. Weather regimes are identified from ERA5 sea-level pressure fields using Principal Component Analysis.
The results show that distinct synoptic regimes are associated with markedly different renewable energy production patterns across Italy. For example, wintertime high-pressure regimes are generally linked to reduced energy production, although notable differences emerge depending on the specific high-pressure configuration and between northern and southern regions of the country.
Overall, these findings highlight the added value of a weather-regime perspective for interpreting and anticipating variability in renewable energy production in Italy, providing a robust basis for improving energy system management and resilience in a weather-dependent power system.
How to cite: Bonanno, R. and Collino, E.: Renewable Energy Variability in the Italian Peninsula: A Weather Regime Perspective, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3466, https://doi.org/10.5194/egusphere-egu26-3466, 2026.
The rapid expansion of rooftop photovoltaic (PV) systems in urban areas provides substantial renewable energy capacity while also modifying surface radiative and turbulent energy exchange in the urban boundary layer. As a result, PV installations can contribute to phenomena such as the photovoltaic heat island (PVHI), which refers to increased ambient temperatures associated with heat absorbed and emitted by PV panels. Understanding these coupled effects is essential to assess PV impacts on the urban surface energy balance and boundary layer structure. Despite growing observational and mesoscale modeling studies, building-resolving large-eddy simulation (LES) investigations with direct comparison to rooftop measurements remain rare. In this study, we evaluate a newly developed rooftop PV energy balance module implemented in the LES model PALM. The module solves the PV surface energy balance with temperature dependent conversion efficiency, providing a physically consistent link between radiative forcing, PV surface temperature, thermal and turbulent exchanges, and power production. Simulations are conducted for a large industrial rooftop near Dresden, Germany, equipped with approximately 2,700 PV panels, using realistic building geometry and multiple representations of rooftop PV layouts. Three clear-sky days representing summer and winter conditions are simulated and compared against rooftop observations, including eddy-covariance (EC) measurements of sensible heat flux, near-surface air temperature, PV surface temperature, and recorded power output. We analyze the ability of the PV module to capture the observed diurnal evolution across these thermal, turbulent, and electrical variables. Sensitivity experiments investigate the influence of grid resolution and different rooftop PV layout representations on thermal and turbulent exchange processes. This work aims to advance the understanding of interactions between rooftop PV systems and the urban boundary layer and to support future interpretation of PV impacts on the urban boundary layer.
How to cite: Zhai, H., Anders, J., Maronga, B., and Mauder, M.: Validation of a rooftop photovoltaic module in large-eddy simulations using eddy-covariance observations, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5124, https://doi.org/10.5194/egusphere-egu26-5124, 2026.
Modelling daily electricity demand is an essential step to ensure grid stability and to meet society’s needs. Temperature is a key driver of demand, as it not only influences the seasonal variability but also the extremes. Day number is commonly used as a proxy for seasonality and is especially efficient at capturing the lower demand of the summer holiday. This is, however, a static feature and therefore not a sufficient choice when modelling demand in a changing climate. It is therefore of great interest to further investigate how to best resolve the true impact of temperature in demand models.
This study quantifies the gain in model performance when utilizing meteorological parameters directly versus using day number only. Furthermore, we evaluate feature engineering strategies to improve the model's ability to leverage the predictive information embedded in temperature. This was done using Generalized Additive Models (GAMs) to model the weather- and calendar-dependent daily electricity demand for nine European countries and assessing different feature combinations.
The results demonstrate an overall improvement in model performance when temperature is included in the modelling across all countries. The most significant improvements are seen in the Nordics and France, with up to 51.5% decrease in mean absolute error (MAE) compared to using day number alone. The significance of temperature is most pronounced when assessing model performance on the upper 5th percentile of daily demand, where the reduction in MAE is up to 69.0%. These findings underscore temperature’s critical role in capturing extreme demand events and highlight the need for climate-responsive modelling strategies.
How to cite: Nesbø Gjøsæter, I. K., Sorteberg, A., and Scheuerer, M.: Towards climate-responsive demand modelling: quantifying the value of temperature and strategies to resolve the true dependency, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5419, https://doi.org/10.5194/egusphere-egu26-5419, 2026.
Previous studies of offshore operational weather windows have typically relied on relatively short records (often less than a decade), limiting the characterization of low-frequency variability and its climate drivers. Here, we use more than 60 years of ERA5 reanalysis data to examine weather-window variability relevant to Taiwan’s offshore wind development and to identify the dominant climate processes governing this variability across timescales. Summer months provide the greatest number of operational weather windows and exhibit relatively stable year-to-year variability, making them the primary season for offshore operational activities. Interannual variability of June-July0August mean weather-window counts is dominated by a coherent regional wind pattern across the Taiwan Strait, with secondary contributions from modulation by the western North Pacific summer monsoon, ENSO, and episodic tropical cyclone activity. Together, these multiscale processes explain more than 50% of the variance in summer weather-window availability. Notably, during the period corresponding to the onset of Taiwan’s offshore wind development (2018-2024), summers have exhibited near-maximum accessibility relative to other time windows in the 60-year record, indicating that such favorable conditions may not persist and should be considered in long-term planning. Outside of summer, weather-window variability displays pronounced low-frequency behavior, including decadal oscillations and trends, with transitional months (e.g., October) associated with the Pacific Meridional Mode and colder months modulated by ENSO. These results highlight the importance of accounting for low-frequency climate variability when assessing offshore operational risk, with implications for reducing weather-related delays and supporting sustained progress toward offshore wind deployment goals. The framework presented here is transferable to other offshore wind regions with appropriate regional adaptation.
How to cite: Tseng, W.-L., Wang, Y.-H., Wang, Y.-C., and Wu, Y.-S.: Variability of Weather Windows in the Taiwan Strait and Their Linkages to Various Climate Drivers, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8814, https://doi.org/10.5194/egusphere-egu26-8814, 2026.
We consider wind speed and power output time series from six turbines of a wind farm located in the Guadeloupe archipelago, in the eastern Caribean Sea. Simultaneous measurements of wind speeds and power outputs were sampled at a 10-minute temporal resolution throughout the year 2024, using an anemometer mounted on the nacelle of each turbine at a height of 48 m above ground level.
We first study their power spectral behavior and scaling statistics in the framework of fully developed turbulence and Kolmogorov’s theory and also in relation with atmospheric boundary-layer effects producing an inertial range with a power-law slope different from 5/3. We obtain an inertial range between scales from 10-7 ≤ f ≤ 10-4 Hz (10 min ≤ T ≤ 56 days), where f is the frequency and T the time scale, for both the velocity data and the power output.
On this inertial range, the Fourier power spectra E(f) follow a scale-invariant relation of the form E(f)=Cf-ß , where C is a constant, f is the frequency, and ß is the slope of the power law. We determine the values of ßv = 1.24 ± 0.07 for the wind velocity and ßP= 1.18 ±0.08. for the power output. We find a one-to-one relationship between both slopes: the steeper ßv , the steeper ßP . Furthermore, over the detected inertial range, using structure function analysis, we obtain intermittent and multifractal properties. In the framework of a lognormal model for the intermittency, we extract the different parameters to characterize this intermittency: the Hurst index H and the intermittency parameter µ. Within this intermittency and turbulent framework, our aim is to better understand the multi-scale relationship between the wind speed and the output power of the turbines.
How to cite: Rised, A., Schmitt, F. G., and Calif, R.: Analysis of the intermittency of simultaneous wind speed and power output data of two groups of wind turbines from a wind park., EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7748, https://doi.org/10.5194/egusphere-egu26-7748, 2026.
Wind direction is critical for wind energy assessment, as it influences turbine yaw alignment, wake effects, and energy production estimates. This is especially relevant in mountainous regions, where complex terrain and atmospheric processes contribute to directional variability. Despite its importance, wind direction from global atmospheric reanalysis has received little attention in wind resource assessments, which mainly focus on wind speed. This study identified clusters of daily-cycle patterns of angular errors in hourly ERA5 wind direction and evaluated a machine-learning calibration using ERA5 surface meteorological variables. The analysis was applied in the complex terrain of the Ecuadorian Andes (3800 m a.s.l.), using one year of data (2021) of ERA5 wind direction (100 m height) and ground-based wind measurements (80 m). K-means clustering was applied to the sine and cosine components of wind direction from both reanalysis and observations. A Random Forest model was trained independently for each cluster using wind speed at 100 m, sine and cosine components of wind direction, 10 m wind gust, near-surface air temperature, dew point temperature, skin temperature, surface pressure, radiation fluxes, and precipitation. Results revealed three clusters related to the daily-cycle and the magnitude of the angular error: Cluster 1- predominantly nocturnal and early morning (8 pm-10 am, minimum at 4 pm); and small angular error (median 16°); Cluster 2 - daytime and predominantly afternoon (10 am - 8 pm, peak at 4 pm), and large angular error (80°); and Cluster 3 - evenly distributed throughout the day, with a slight maximum at 3 pm; and medium angular error (47°). The largest errors coincided with lower wind speed and post-midday decreases in air temperature, skin-surface temperature, and surface pressure. They also coincided with large variability in wind direction since Cluster 1 was dominated by easterly to southeasterly winds, Cluster 3 by westerly, while Cluster 2 showed a large dispersion from easterly to westerly flows. Calibration substantially improved wind direction representation. For the nocturnal cluster, the most informative predictors were 10 m wind gust, skin temperature, and surface pressure, reducing the median angular error to 8° and improving the wind direction distribution (Perkins Skill Score - PSS from 0.50 to 0.69). For the high-error afternoon cluster, wind speed, total precipitation, and surface pressure were the dominant predictors, decreasing the median angular error to 15° and improving PSS from 0.32 to 0.60. Finally, for the evenly-distributed cluster, surface pressure, dew point temperature, and wind speed were most relevant predictors, yielding a median angular error of 8° and PSS increase from 0.36 to 0.68. The findings highlight the strong dependence of the angular error of ERA5 wind direction on the daily-cycle and thermal processes.
How to cite: Ballari, D. and Contreras, J.: Daily-cycle patterns of angular errors in ERA5 wind direction: clustering and calibration using surface meteorological variables, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12857, https://doi.org/10.5194/egusphere-egu26-12857, 2026.
Wind energy research relies on remote sensing technologies like dual-Doppler (DD) lidar and radar, or space-born SAR data to estimate complex meteorological conditions and the flow field around wind farms. Offshore measurement campaigns over the last decade accentuate the potential of DD wind radar technology for wind energy application. Onshore, promising results were also reported as part of the American WAKE ExperimeNt (AWAKEN).
The main drawback of wind radar technology is that further characterisation of radar measurement accuracy is required for industry implementation. We investigate X-band wind radar measurement accuracy in a comparison study, using meteorological mast data and wind lidars from an onshore measurement campaign.
Despite their cost, meteorological masts provide very accurate point measurements over specific heights, and are the preferred reference for wind validation. Wind lidars are a proven alternative, as demonstrated in several offshore campaigns. Like for lidars, radar's measurement principle is based on the Doppler effect, and both devices allow for wind field estimates. Yet, radars can scan over larger distances at a much higher sampling rate, with increased resolution along the beam due to the use of a compressed pulse. However, unlike pulsed lidars that emit collimated beams, the radar beam expands with distance, which arguably adds uncertainty to the measurements at far ranges.
A dual-Doppler lidar-radar set-up is used with the remote sensing devices collocated in space, but measuring a-synchronously. We conduct the analysis on wind component basis (u, v) and wind speed and direction, focusing on inflow in front of a large wind turbine (OPUS 1) at the onshore Wivaldi Test Site in Northern Germany, as part of the radar Krummendeich Campaign. Lidars and radars are deployed approximately 3 km to 4 km away from the collocated lidar - meteorological mast - radar measurements. Although the influence of increased probe volume averaging, due to beam expansion, in distance is unclear due to campaign set-up, the results presented set a ground base for further use of long range wind lidars as validation for upcoming radar measurement campaigns.
How to cite: Trindade, A., Rott, A., Schneemann, J., and Kühn, M.: On the accuracy of X-band Dual-Doppler Radar for wind energy applications: a comparison study., EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21748, https://doi.org/10.5194/egusphere-egu26-21748, 2026.
Continuous wave lidars have been widely applied in wind site assessment in recent years. The CW technique uses the adjustment of the beam focus for ranging. A known constraint of this technique is the poor definition of the range weighting function, particularly at upper ranges. In case of inhomogeneous reflectivity distribution, for example caused by low hanging clouds, the center of the scattering volume does not necessarily agree with the center of the adjusted focus range leading to a wrong range allocation of the wind measurements.
As a solution to this fundamental issue, a frequency modulation (FM) of a CW lidar provides independent information of the actual measuring height. The beat frequency of the FMCW lidar depends on the real range of the center of the scattering volume, which may differ from the assumed range based on the focus adjustment. Based on this real range information, the wind profile can be regridded to the expected or defined measuring heights. We will showcase the impact of regridding FMCW wind profiles using a Wind Ranger 200 for cases with inhomogeneous reflectivity distributions and compare the results with a reference pulsed wind lidar.
How to cite: Baumer, F., Markmann, P., Burgemeister, F., and Peters, G.: Potential of regridding of FMCW lidar wind profiles to improve data availability, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12889, https://doi.org/10.5194/egusphere-egu26-12889, 2026.
Understanding how wind turbines interact with large-scale atmospheric phenomena is an increasingly important issue for wind farm developers. With the latest 15 MW turbines reaching heights of 270 m, several studies have predicted that farms of these turbines will be able to induce Atmospheric Gravity Waves (AGWs). These buoyancy-driven waves are triggered as a result of a farm vertically deflecting thermally stratified flow above the turbine array. In particular, as the temperature inversion above the North Sea is typically located at heights near the top of a 15 MW turbine, farms in this region may be especially susceptible to generating AGWs. Hence, with most European offshore wind farms based in the North Sea, understanding the interactions with, and impact of, these waves is vital for yield prediction.
Recent studies have shown that AGWs cause a redistribution of flow at the farm scale, altering wind farm power production. As a result, AGWs provide a new challenge for wind farm planning and raise questions about whether farm design can influence how the AGW is triggered, and if these AGWs also have impacts at the turbine scale. To address these questions, large eddy simulations using actuator line turbine representations were undertaken. These simulations replicated the typical atmospheric turbulence, Coriolis force and thermal parameters seen in the North Sea.
First, the middle turbine of an infinitely wide row was simulated. The turbine triggered an AGW, and the flow field was compared to a wave-free case. The AGW was found to cause upstream flow deceleration, accelerate bypass flow above the turbine wake, and cause pockets of acceleration within the wake itself at AGW troughs. Overall, this led to faster wake recovery than in a wave-free case.
Following this, simulations were conducted using two turbines aligned in the streamwise direction, each representative of the middle turbine in an infinitely wide row. Introducing a second turbine triggered stronger AGWs, magnifying their effects on the flow. Furthermore, by varying the position of the downstream turbine, it was possible to both amplify and dampen the AGW produced, along with causing a shift in the wave phase. The power of the second turbine was found to vary sinusoidally with the change in turbine position. When in line with an AGW trough, the second turbine even outperformed the first despite sitting in its wake. However, the increased power came at the cost of a higher mean blade loading and an increase in cyclic loading.
This work demonstrates that AGWs can impact intra-farm flows and turbine performance. Additionally, it confirms an interdependence between AGWs and wind farm turbine spacing. Given the variation in the AGW with spacing, it may become an important factor in design which considers both intra-farm and farm-to-farm scale flows.
How to cite: Rafferty, T. and Vogel, C.: The implications of atmospheric gravity waves for wind farm and turbine design, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13601, https://doi.org/10.5194/egusphere-egu26-13601, 2026.
The growing demand for renewable energy has accelerated the development of offshore wind farms (OWFs), particularly in the German Bight, where significant expansion is planned. The construction of these installations in the shallow southern North Sea can significantly alter the lower atmosphere by introducing turbulence and modifying wind profiles. Observations and atmospheric simulations suggest that OWFs reduce near-surface wind speeds and affect vertical wind structure, depending on the size and layout of the turbines, as well as the expansion scenario.
In order to evaluate the potential impact on ocean waves, we use atmospheric simulations representing various OWF development scenarios as input for the spectral wave model WAM (v4.6), investigating changes in the regional wave climate over a 10-year period. The results suggest that OWFs affect not only local wave conditions, but also lead to a reduction in significant wave height and wave power downstream over larger areas. These findings emphasise the importance of considering OWF-induced atmospheric changes when modelling waves and assessing the impact on the coast.
How to cite: Groll, N., Akhtar, N., and Geyer, B.: Impact of Offshore Wind Farm Expansion Scenarios on Wave Climate in the German Bight, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17575, https://doi.org/10.5194/egusphere-egu26-17575, 2026.
Gridded short-wave surface radiation components are essential for meteorology, hydrology, and renewable energy forecasting. In particular, solar power prediction for photovoltaic (PV) and concentrated solar power (CSP) systems depends critically on accurate short-range forecasts of global, direct, and diffuse irradiance. Delivering such high-resolution, site-specific estimates is a core objective of the FFG-funded PV4Community project and the focus of the presented work.
A hybrid analysis–nowcasting approach has been implemented in the INCA (Integrated Nowcasting through Comprehensive Analysis; Haiden et al. 2011) radiation module. It combines global irradiance and sunshine duration observations from the Austrian monitoring network, MTG satellite retrievals, and high-resolution NWP guidance from AROME and C-LAEF. Strong coupling to INCA’s cloud analysis and cloud-motion nowcasting enables high spatial detail and very short-range accuracy, while accounting for low-sun-angle conditions and the effects of Alpine topography (terrain shading, slope, aspect).
Radiation fields are produced on a 1 km × 1 km grid at 15-minute frequency with lead times up to 48 h. A key advancement is the derivation of diffuse and direct radiation components using an adapted version of the Gassel (1999) algorithm. The original Gassel method describes a physically consistent partitioning of global horizontal irradiance into its beam and diffuse components based on solar geometry and atmospheric transmissivity. Our adaptation extends this approach for operational nowcasting by: (i) dynamically coupling the algorithm with INCA’s global irradiance output, (ii) incorporating MTG-based cloud physical properties, and (iii) adjusting the clear-sky and turbidity assumptions to the Alpine environment. This yields a robust irradiance decomposition that remains stable across rapidly changing cloud scenes and complex terrain.
Validation against measurements from the ARAD radiation network (Olefs et al. 2016) demonstrates high correlation and low bias for both diffuse and direct irradiance, confirming the suitability of the new components for operational solar energy applications. Their integration into the INCA framework ensures sustained, near-real-time availability and opens the door for improved PV nowcasting, solar ramp detection, and future energy system applications.
Funding: This work was supported by the Austrian Research Promotion Agency (FFG; www.ffg.at).
Haiden, T., Kann, A., Wittmann, C., Pistotnik, G., Bica, B., & Gruber, C. (2011). The Integrated Nowcasting through Comprehensive Analysis (INCA) system and its validation over the Eastern Alpine region. Weather and Forecasting, 26(2), 166-183.
Gassel, A. (1999). Beiträge zur Berechnung solarthermischer und exergieeffizienter Energiesysteme (Doctoral dissertation, Fraunhofer-IRB-Verlag).
Olefs, M., Baumgartner, D. J., Obleitner, F., Bichler, C., Foelsche, U., Pietsch, H., ... & Schöner, W. (2016). The Austrian radiation monitoring network ARAD–best practice and added value. Atmospheric Measurement Techniques, 9(4), 1513-1531.
How to cite: Hadzimustafic, J., Schicker, I., Madjdi, N., and Wind, G.: Hybrid Analysis and Nowcasting of Surface Solar Radiation Components in the INCA Framework, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3732, https://doi.org/10.5194/egusphere-egu26-3732, 2026.
Mountainous regions worldwide offer substantial yet underutilized wind energy potential. A key challenge limiting the expansion of wind energy in such areas is the difficulty of obtaining accurate wind resource estimates in complex terrain. Traditionally, long-term wind speed series are derived from short-term site observations combined with reanalysis products. Conventional reanalysis products such as the ERA5 single levels at 10 m and 100 m often misrepresent local orography, resulting in biased wind speed predictions and unreliable inputs for Measure-Correlate-Predict (MCP) methods used in wind resource assessment. Our study addresses this challenge by employing high-quality mast observations at high-elevation sites in the tropical Andes and by leveraging ERA5 model level wind fields, which remain largely unexplored in wind energy research and industry. We compared wind speed estimates at different atmospheric heights of ERA5 model level data with hourly wind speed observations at 80 m from four meteorological masts (2829–3796 m a.s.l.) in the tropical Andes of southern Ecuador. We developed site-specific Random Forest (RF) models to calibrate ERA5 wind speeds. Our results indicate that wind speeds extracted from upper ERA5 model levels (approximately 1000–1500 m for most sites) are stronger correlated with mast measurements than those at the hub-heights (near the surface). Relative to single level inputs, RF estimates driven by model level data show mean improvements of 59% in the Perkins Skill Score, 40% in R², and 23% in MAE and RMSE. In addition, the bias in annual energy production is reduced to below 7%, compared to 22% when ERA5 single level data are used. The largest gains are observed at sites located on exposed ridgelines and peaks, typical targets for wind farm development, where upper model levels more effectively represent the local atmospheric flow. Our results demonstrate that selecting optimal ERA5 model level offers a strategy for generating reliable site-specific wind time series in complex terrain providing useful information for wind resource assessment studies accelerating the development of wind energy projects in mountainous regions.
How to cite: Contreras, J., van Lipzig, N., Samaniego, E., and Ballari, D.: Improving wind energy estimates in mountainous terrain using optimal ERA5 model level heights, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14460, https://doi.org/10.5194/egusphere-egu26-14460, 2026.
Hydropower stands as the dominant source of renewable electricity worldwide, playing a pivotal role in global transitions to low-carbon energy systems and climate change mitigation. Yet, the planetary distribution of hydropower infrastructure remains poorly quantified at a global scale—a critical gap that hinders accurate assessments of energy security, freshwater resource allocation, and environmental sustainability. Current public inventories, which are largely compiled through fragmented bottom-up reporting schemes reliant on national or regional submissions, are plagued by pervasive incompleteness, inconsistent geospatial referencing, and significant lags in updates, rendering them inadequate for evidence-based global policy and conservation planning. Here, we present a multimodal artificial intelligence (AI) framework that enables the automated identification of hydropower plants from remote sensing imagery via a globally uniform, top-down methodology. Applied to 8,330,487 river segments across the globe, this framework detects 12,640 hydropower installations, 55.7% of which are unrecorded in leading contemporary public inventories. The resultant global dataset uncovers striking regional disparities and transboundary clustering in hydropower development. It further demonstrates that hydropower infrastructure impacts 56.97% of the world’s protected areas, with marked biomass loss occurring during the construction phase. Complementary hydrological analyses reveal that 29.9% of these installations have experienced declining runoff over the past two decades, while 12.0% are exposed to high flood risk. This work establishes a scalable framework for monitoring global hydropower expansion and its associated environmental and climatic risks, providing a critical foundation for evidence-based energy and conservation policy. The study releases a topdown remote sensing-based hydropower monitoring platform https://glohydro.cn.
How to cite: Li, J. and Huang, X.: Revealing Global Patterns of Hydropower Plants via Multimodal AI, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4045, https://doi.org/10.5194/egusphere-egu26-4045, 2026.
The rapid integration of renewable energy into national and global electricity systems is a cornerstone of climate mitigation strategies consistent with the Paris Agreement. Photovoltaics (PV) are central to this transition, with global installed capacity exceeding 800 GW by 2021 and projections indicating multi-terawatt deployment by mid-century (IRENA, World energy transitions outlook, 2023). While large-scale PV expansion is essential for decarbonization, it also constitutes a substantial land-surface modification that can influence surface energy fluxes, radiation balance, and atmospheric circulation. Quantifying these interactions is therefore important for understanding the broader environmental implications of renewable energy systems at scale.
Here, we investigate the climatic response to spatially extensive PV deployment using the intermediate-complexity climate model PLASIM (Fraedrich et al., 2005). We perform idealized global simulations with varying fractions of land surface covered by PV, across multiple horizontal resolutions (T21, T31, T42) and three model configurations: atmosphere-only, mixed-layer ocean, and a large-scale geostrophic ocean. This framework allows us to contrast short-term atmospheric adjustments with longer-term, ocean-coupled responses, and to assess the sensitivity of results to spatial resolution and coupling timescales.
Our results show that the climate response to PV deployment is strongly dependent on the albedo contrast between PV panels and the underlying surface. Low effective panel efficiency leads to surface warming due to reduced albedo, while intermediate efficiencies yield mixed regional responses. At high efficiencies, cooling emerges relative to the control climate. These non-linear responses highlight the importance of background land properties and surface–radiation interactions in shaping the climatic impacts of renewable energy deployment.
While the simulations represent idealized and prospective scenarios, we discuss pathways for linking such model-based assessments with long-term field measurements and remote-sensing observations of existing solar installations. Although a clear scale mismatch exists between climate-model grid cells and observed PV sites, observational datasets provide valuable constraints on surface temperature, albedo changes, and land-cover effects. Combining retrospective observations with prospective climate-model experiments offers a promising avenue for cross-examining renewable energy impacts across spatial and temporal scales.
This study contributes to the spatial and temporal modelling of renewable energy systems by bridging climate-system modelling, land-surface impacts, and future deployment scenarios, and by outlining how modelling and observations together can inform sustainable pathways for large-scale solar energy expansion.
How to cite: Samanta, A. and Rehfeld, K.: Simulating Climate Responses to Large-Scale Photovoltaic Deployment with PlaSim, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20178, https://doi.org/10.5194/egusphere-egu26-20178, 2026.
Wind loads are a major design constraint for photovoltaic (PV) systems, in particular when modules are installed on flat roofs not connected to the building. Such PV system designs must be heavy enough to assure safe and durable operation under varying and peak wind conditions, but should not be much heavier. The additional weight required for a selected configuration cannot be easily deduced from wind engineering standards (codes) without a calibrated aerodynamic model and without knowledge of the local wind environment. Consequently, risk and lifetime analysis, by means of critical loads for PV panel disposition and fracture initiation due to extreme wind events, as well as fracture worsening due to unsteady aerodynamic loads, cannot be addressed.
To overcome the mentioned limitations, case-specific loading rules have to be developed based on design-specific aerodynamics and site-specific wind conditions within the atmospheric surface layer (ASL), potentially in an urban environment. Numerical simulations provide means to develop such case-specific loading rules. For this purpose, the simulations need to offer sufficient fidelity to enable the prediction of lift and drag forces that act on the selected PV system, simulataneously providing further insight into the flow. Nevertheless, the computational approach is limited by numerical approximations and modeling assumptions. The corresponding numerical and modeling errors manifest themselves by a dependence of the simulated wind loads on the mesh, timestep, selected turbulence model, and inflow condition, among others.
In the contribution, large-eddy simulations (LES) of wind loads on PV systems placed on the ground and a flat roof will be presented. First, starting from the case of a single (South-oriented) PV panel placed on the ground, LES results obtained with OpenFOAM and PVade are compared to each other in order to establish a minimal reference set-up. Second, the geometry is extended to a double-panel (East-West) configuration, which is likewise simulated with both solvers. Third, a single building is introduced in the OpenFOAM-based set-up so that LES for a building without and with a roof-placed PV system are conducted. For comparison, the same PV array is simulated placed on the ground. These results demonstrate the significant influence of the local wind environment on panel-based wind loads and the derived case-specific loading rules. Last, an outlook is given to fluid-structure interaction and fracture initiation.
How to cite: Klein, M., Kabongo, M., Schmidt, H., and Meyer, R.: Numerical simulation of wind loads on PV systems placed on the ground and a flat roof, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21298, https://doi.org/10.5194/egusphere-egu26-21298, 2026.
Posters virtual: Tue, 5 May, 14:00–18:00 | vPoster spot 4
EGU26-6552 | ECS | Posters virtual | VPS19
Assessing future wind energy resources in the Iberian Peninsula under climate change scenariosTue, 05 May, 14:18–14:21 (CEST) vPoster spot 4
This study evaluates annual changes in wind power density (WPD) in a domain covering the Iberian Peninsula and adjacent areas using several CMIP6 global climate models and the ensemble mean under historical (1961-1990) and SSP5-8.5 scenarios for two-time horizons—near future (2041–2070) and far future (2071–2100).
Results from the ensemble indicate a robust and generalized decrease in WPD throughout the 21st century. The most pronounced declines occur in windows starting mid-century (2050–2055), with reductions of about -90 W m-2 century-1 persisting for up to 40-year periods. Short-lived positive trends (≈ 50 W m-2 century-1) appear around 2030 and 2045, suggesting temporary peaks before a marked decline (≈ -100 W m-2 century-1) in later decades. Comparisons between future and historical periods reveal strong WPD decreases (-70 W m-2), mainly offshore, particularly in far-future scenarios.
Inland areas may experience annual mean WPD values falling below the cut-in threshold (3 m/s, ≈ 15.5 W m-2), rendering some older wind farms economically and technically unviable. Offshore regions, despite current technological priorities, face substantial WPD reductions (up to -60 W m-2), while inland declines are significant in northeastern Spain, where major wind farms are located. These projected reductions—especially offshore (10–20%)—could challenge the financial viability of future wind energy projects.
How to cite: García-Miguel, A., Calvo Sancho, C., Díaz Fernández, J., González Alemán, J. J., López Reyes, M., Bolgiani, P., Martín Pérez, M. L., and Luna, M. Y.: Assessing future wind energy resources in the Iberian Peninsula under climate change scenarios, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6552, https://doi.org/10.5194/egusphere-egu26-6552, 2026.
EGU26-14485 | ECS | Posters virtual | VPS19
Integrating Micro-Scale Urban Geometry with Macro-Scale Climate Projections to Improve Rooftop Photovoltaic Potential Assessment: An Application to Selected Urban Areas in the Southeastern MediterraneanTue, 05 May, 14:21–14:24 (CEST) vPoster spot 4
This study presents a comprehensive methodological framework that integrates micro-scale urban geometry with macro-scale climate projections to improve the assessment of rooftop photovoltaic (PV) potential in urban environments. High-precision solar resource estimation is achieved through the use of very high–resolution Digital Surface Models (DSMs; 0.8 m) within the Solar Energy on Building Envelopes (SEBE) model, enabling detailed simulation of shading effects in dense urban fabrics.
Historical and present-day atmospheric inputs—including surface solar radiation, cloud cover, and aerosol optical depth—are obtained from the Copernicus Atmosphere Monitoring Service (CAMS) and combined with meteorological variables from ERA5-Land. Future rooftop PV potential is projected using a multi-model ensemble of CMIP6 climate simulations under the SSP2–4.5 and SSP5–8.5 emission scenarios. Statistical downscaling techniques are applied to translate large-scale climate projections to local urban conditions.
In addition, the study evaluates PV system performance during specific atmospheric episodes, quantifying the effects of dust intrusions and compound events—defined as the co-occurrence of high temperatures and elevated dust concentrations—on energy yield. Finally, cluster analysis is performed on the urban building stock of selected southeastern Mediterranean cities using key performance indicators, including received solar radiation, total energy yield, rooftop area, and building height.
The results demonstrate that integrating micro-scale urban morphology with macro-scale climate projections is critical for accurately estimating rooftop PV potential, particularly in regions characterized by complex urban structures and climate-sensitive atmospheric processes.
How to cite: Agazarian, N., Cartalis, C., Philippopoulos, K., and Agathangelidis, I.: Integrating Micro-Scale Urban Geometry with Macro-Scale Climate Projections to Improve Rooftop Photovoltaic Potential Assessment: An Application to Selected Urban Areas in the Southeastern Mediterranean, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14485, https://doi.org/10.5194/egusphere-egu26-14485, 2026.
EGU26-22088 | ECS | Posters virtual | VPS19
Hierarchical Bayesian Modeling of Solar Irradiance under Extreme Weather in the Tucson Electric Power RegionTue, 05 May, 14:24–14:27 (CEST) vPoster spot 4
Accurate forecasting of surface solar irradiance is needed, as it helps in PV power system planning, particularly under extreme weather conditions. Deterministic and persistence-based forecasting methods generally fail under extreme weather conditions. The present study develops a hierarchical Bayesian spatio-temporal model to forecast solar radiation in the Tucson Electric Power (TEP) region, Arizona, United States. Satellite-derived (CERES SYN1deg) and reanalysis (MERRA-2) solar radiation data have been used in the present study to identify variability across the four TEP stations. The hierarchical Bayesian spatio-temporal model outperformed the persistent model. The findings also highlight that, instead of focusing on point forecasts, we should focus on uncertainty-aware forecasts.
How to cite: Singh, J.: Hierarchical Bayesian Modeling of Solar Irradiance under Extreme Weather in the Tucson Electric Power Region, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22088, https://doi.org/10.5194/egusphere-egu26-22088, 2026.
