1) Forecasting and simulating high impact weather events - research on using advanced artificial intelligence and machine learning techniques to improve numerical weather model prediction of severe weather events (such as winter storms, tropical storms, and severe mesoscale convective storms);
2) Development and improvement of model numerics - basic research on advanced numerical techniques for weather and climate models (such as cloud resolving global model and high-resolution regional models specialized for extreme weather events on sub-synoptic scales);
3) Development and improvement of model physics - progress in research on advanced model physics parameterization schemes (such as stochastic physics, air-wave-oceans coupling physics, turbulent diffusion and interaction with the surface, sub-grid condensation and convection, grid-resolved cloud and precipitation, land-surface parameterization, and radiation);
4) Verification of model physics and forecast products against theories and observations;
5) Data assimilation systems - progress in the development of data assimilation systems for operational applications (such as reanalysis and climate services), research on advanced methods for data assimilation on various scales (such as treatment of model and observation errors in data assimilation, and observational network design and experiments);
6) Ensemble forecasts and predictability - strategies in ensemble construction, model resolution and forecast range-related issues, and applications to data assimilation;
7) Advances and challenges in applying data from various conventional and avant-garde observation platforms to evaluate and improve high-resolution simulations and forecasting.
8) Climate and Weather Interventions
Orals: Mon, 4 May, 08:30–12:30 | Room D1
This presentation will provide an overview of recent scientific developments at ECMWF, with focus on the upgrade of the Integrated Forecasting System (IFS) for cycle 50r1.
An upgrade to ECMWF's Integrated Forecasting System (IFS) is scheduled for operational implementation in early 2026. IFS Cycle 50r1 brings major advances in both the forecast model and the data assimilation system, marking a significant step forward in coupled Earth system prediction. The model upgrade includes the introduction of the NEMO4-SI3 ocean and sea ice model, improved wave-ice interactions, revised vertical diffusion and gravity wave drag in the stratosphere, changes to the convection scheme, a new glacier scheme, and a revisionof the SPP scheme introduced in 2023 that reduces excessive near-surface wind spread in the ensemble. On the data assimilation side, outer-loop coupling has been introduced between the atmosphere and ocean to provide balanced initial conditions for the coupled forecast model, while further enhancements of weak-constraint 4D-Var introduce time-varying model errors and has been extended to the boundary layer. Cost-efficiency improvements have been made through the use of single-precision trajectories, single-precision ocean model and from reducing the resolution of the first minimisation in the EDA. The system now allows humidity increments in the stratosphere, addressing longstanding issues in moisture analysis at these levels.
A major development of data-driven weather prediction models is the operational implementation (July 2025) of the Artificial Intelligence Forecasting System ensemble (AIFS-ENS). AIFS ENS is trained using a CRPS-based approach, which optimises the probabilistic scores of the ensemble forecasts. It delivers skilful weather forecasts with significantly improved speed and energy efficiency.
The 50r1 update of the ECMWF atmospheric composition forecast led to improved forecast of surface ozone and sulphur dioxide. A further addition is the routine assimilation of aerosol optical depth retrievals from VIIRS and Sentinel 3.
Besides the developments for the operational updates, the preparation of the production of 6th generation atmosphere and ocean reanalyses (ERA6/OCEAN6), the new version of the atmospheric composition reanalysis (EAC5), and the next seasonal prediction system (SEAS6) have progressed well.
How to cite: Flemming, J., English, S., and Pappenberger, F.: Recent progress and outlook for the ECMWF Integrated Forecasting System, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21470, https://doi.org/10.5194/egusphere-egu26-21470, 2026.
ACCORD is a consortium made up of 26 Meteorological Services (https://www.accord-nwp.org/ ). The primary objective is to provide the consortium's member services with state-of-the-art numerical weather prediction (NWP) limited-area model codes. A substantial part of the ACCORD codes are shared with IFS-ARPEGE.
During phase 1 of ACCORD (2021-2025), collaborative working methods have been drastically modernized. Code management has greatly benefited from the implementation of the ACCORD software forge (Github) as well as from the expanded use of a testing tool which enables component-wise testing of new code versions. The adaptation of the codes to new HPC architectures (CPU-GPU accelerators) has largely progressed in close collaboration with MF (ARPEGE) and ECMWF (IFS).
Research on the current (semi-implicit semi-Lagrangian spectral) dynamical kernel of ACCORD models continues, notably through an extensive reformulation of the semi-implicit operator. In addition, the alternative FVM (Finite Volume Model) code, initially developed at ECMWF, is being studied. SURFEX (https://www.umr-cnrm.fr/surfex/ ) has become the main code infrastructure for modeling surface processes and the surface-atmosphere interface. Efforts devoted to developing new options for very high-resolution modeling are increasing: dynamical kernel, 3D aspects of turbulence and radiation, refined surface characteristics, all for models at the hectometer scale. This trend is largely driven by user needs.
In data assimilation, a major advance is the near-operational status of flow-dependent algorithms (EnVar-type algorithms coded in the OOPS software framework). ACCORD has maintained first-rate expertise in preprocessing observations for assimilation. This applies both to satellite data (infrared or microwave, in polar orbit or geostationary orbit) and to ground-based networks of various types (radar, surface networks, citizen observations) or aircraft (Mode-S). In probabilistic forecasting and ensemble methods (EPS), scientific collaboration focuses (among other aspects) on ensemble perturbation methods. Several approaches for model perturbations have been studied, such as tendency perturbations (SPPT), model parameter perturbations (SPP or RP), and surface field perturbations.
A new scientific strategy was approved by the Assembly of Directors in December 2024 for the next Programme phase (2026-2030). The main objectives include:
-
Continue modernizing the collaborative working methods.
-
Continue adapting the codes for CPU-GPU architectures.
-
Continue research on the various model components with an increased focus on very high resolution and high-impact weather forecasting.
-
Develop a research infrastructure enabling process-oriented meteorological evaluation of models using specialized observations.
-
Continue to develop DA algorithms with flow dependence, leveraging observations from the coming years (MTG etc.).
-
Pursue and strengthen scientific collaboration on ensemble forecasting systems.
In connection with the rapid evolution of data driven forecast tools, ACCORD members want to be proactive in AI initiatives in Europe (within EUMETNET and with ECMWF). The scientific strategy foresees exploring hybrid "AI-physical NWP" solutions.
In the presentation, a few keynote features of the ACCORD scientific strategy will be further addressed.
How to cite: Fischer, C.: The ACCORD consortium and its scientific strategy, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1684, https://doi.org/10.5194/egusphere-egu26-1684, 2026.
Accurate medium-range weather guidance is essential, yet end-users lack clear, location-specific evidence on where forecasts are predictable and which freely available global systems perform best. We present a global evaluation of three conventional numerical weather prediction (NWP) models (ICON, IFS, GEFS) and an AI forecast model (AIFS) for 1-10-day lead times, focusing on four near-surface variables (2-m air temperature, precipitation, 10-m wind speed, and 2-m relative humidity). Using 00 UTC cycles over 1 September 2024 to 30 November 2025, we resampled forecasts to a 1° grid and assessed day-to-day variability (correlation and root mean square error), mean bias, variance bias, and lead-time dependence (drift) against multiple references (primarily JRA-3Q, with additional evaluation against ERA5, station data, and additionally IMERG-Late for precipitation). AIFS achieves the highest skill for temperature, precipitation, and wind at all lead times (relative humidity is unavailable from AIFS); at 3-day lead it explains, on average, 53\% more variance in daily precipitation globally than the next-best model (ICON), and 232\% more variance than GEFS in the tropics. Among the conventional systems, ICON is generally most skillful, while GEFS ranks lowest overall. Mean-bias drift is negligible across models, but variance drift is evident for several variables, most notably increasingly attenuated AIFS precipitation variability with lead time. Model correlation rankings are robust across reference datasets, although precipitation and humidity show greater reference sensitivity than temperature. We also map global predictability using 3-day lead daily temporal correlation of the locally best-performing model, showing highest predictability for temperature and wind in mid-to-high latitudes and markedly lower predictability for precipitation in the tropics. Our study provides actionable guidance on where global forecasts can be trusted and establishes a baseline for future AI and NWP model assessments.
How to cite: Puthiyaveettil, M., Beck, H., and Ma, J.: Where Is Weather Predictable and Which Models Get It Right? Global Assessment of Conventional and AI-Based ForecastModels, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7617, https://doi.org/10.5194/egusphere-egu26-7617, 2026.
AI-based methods have already proven their skill in weather prediction. The availability of high quality training data and ever more advanced model architectures opened the door for a new range of models providing predictions fast and at a low operational cost. Further development of such models, apart from seeking advancements in the architectural design and increased quality of the training datasets, might require a new approach. One still under-explored dimension is to tighten the link between the data-driven (often weather-system-agnostic) methods and the well established knowledge of atmospheric physics represented by NWP.
Physics-awareness in AI model development may guide training, reduce compute requirements and improve the consistency of the predicted variables. In this talk we discuss the possible approaches to physics-informed AI model design, ranging from physics-based terms in the cost function to hybrid physical-neural architectures. We show the impact these methods have on the training process and discuss possible improvements in the forecast skill and physical consistency.
How to cite: Rognvaldsson, O. and Stanislawska, K.: Physics-informed AI systems - how the constraints from NWP can support development of better AI models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18633, https://doi.org/10.5194/egusphere-egu26-18633, 2026.
Subseasonal forecasts, which predict weather patterns from weekly up to seasonal timescales, are crucial to minimize the adverse impacts of extreme weather events, such as heatwaves and droughts, on ecosystems and society. However, forecast skill at subseasonal lead times remains limited, as the chaotic nature of the atmosphere reduces the usefulness of the information contained in atmospheric initial conditions for increasing lead times. In contrast, land surface states, including soil moisture and vegetation anomalies, evolve more slowly and retain memory over weeks, allowing them to persist across subseasonal timescales and making them a potentially important source of predictability. Despite this, most operational weather forecast models represent only the mean seasonal cycle of land conditions, because accurately incorporating land surface anomalies remains challenging and can degrade model performance.
In order to address this situation, we develop a prototype of a hybrid weather prediction model to forecast near-surface temperature and surface soil moisture and related extremes. The model leverages the flexibility of deep learning to build on (i) satellite-based land surface observations, (ii) short-range forecasts from the Integrated Forecasting System to inform the model with physically consistent atmospheric evolution, and (iii) previous meteorological conditions sourced from reanalysis data. First results suggest that land surface anomalies exert a stronger influence during extreme conditions, when land memory persists, whereas under average conditions their influence is more evenly shared with atmospheric anomalies. Our study provides a benchmark for integrating land surface information into hybrid forecasting systems and highlights pathways to improve subseasonal prediction and early warning systems.
How to cite: Ruiz-Vásquez, M., Oh, S., Düben, P., and Orth, R.: Enhancing subseasonal forecasting skill with land observations and physics-informed deep learning, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15763, https://doi.org/10.5194/egusphere-egu26-15763, 2026.
The Earth System Modelling at the Weather Scale (ESM-W) project, a collaboration between the German Weather Service (DWD) and GeoInfoDienst BW, aims to develop a coupled ocean-atmosphere forecasting system. This system utilizes the ICON-O ocean model and the ICON-NWP atmospheric model.
In this presentation, we will showcase the advancements made in the ICON-based coupled global forecasting system. Notably, a near-real-time (NRT) system has been established, comprising a time-critical weakly coupled data assimilation cycle and coupled forecasts with lead times of up to ten days. The ocean component uses a 20km resolution while the atmospheric component uses the operational 13km resolution system. The ocean component assimilates ARGO buoy data and satellite-derived sea surface observations using 3D-VAR(-FGAT) methods, while the atmosphere employs an EnVAR method leveraging ensemble information from DWD’s operational routine. This system has been generating daily analyses and forecasts since May 2025 and is improved continuously.
We will present the coupled system’s developments and performance, including evaluations for both the data assimilation system and the model components. Additionally, we show verification against non-assimilated ocean datasets, such as fixed buoys.
How to cite: Krueger, D., Bevrnja, A., Hayrapetyan, A., He, Y., Hüther, T., Schenk, N., Sprengel, M., Potthast, R., Keller, J., Hollborn, S., Schlemmer, L., and Zängl, G.: The coupled ICON Ocean-Atmosphere System and its Use for Weather-Scale Forecasts, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16609, https://doi.org/10.5194/egusphere-egu26-16609, 2026.
Accurate prediction of tropical cyclones remains a major challenge for numerical weather prediction, particularly for storm intensity, structure, and track. Progress depends on effectively assimilating both passive and active microwave observations, which together provide complementary insights into atmospheric temperature and moisture, cloud microphysics, and precipitation. Passive microwave measurements from low-Earth orbiting satellites, including those in low-inclination orbits, provide frequent sampling that is particularly valuable for tuning temperature and water vapor initial conditions. Observing system experiments with NOAA’s Hurricane Analysis and Forecast System (HAFS) and NASA’s Global Earth Observing System (GEOS) show that assimilating these data leads to more realistic storm structures and measurable improvements in forecasts of intensity and track.
Spaceborne radar observations provide vertical detail on clouds and precipitation. Instruments such as the GPM Dual-frequency Precipitation Radar and EarthCare’s Cloud Profiling Radar are now supported in the Community Radiative Transfer Model (CRTM) through a new spaceborne radar forward model and a Discrete Dipole Approximation–based scattering database. These capabilities are being implemented and tested within the Joint Effort for Data assimilation Integration (JEDI) framework, developed collaboratively by JCSDA, NASA, NOAA, and international partners. Early results from experiments assimilating GPM-DPR observations within NOAA’s GEOS system and CloudSat CPR data within NOAA’s HAFS demonstrate improvements in the analysis of cloud and precipitation structures.
How to cite: Moradi, I., Zhu, Y., Kalluri, S., and Todling, R.: Advancing Assimilation of Microwave and Radar Observations in the NWP Models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1749, https://doi.org/10.5194/egusphere-egu26-1749, 2026.
The rapid development of AI weather foundation models, such as ECMWF’s AIFS-ENS, promises to revolutionize operational forecasting by delivering competitive skill at a fraction of the computational cost of traditional numerical weather prediction (NWP). However, a critical gap remains: these models currently lack native, robust mechanisms for assimilating real-time, novel observation types, particularly in data-sparse regions. We present a preliminary framework for the first integration of ensemble data assimilation with AIFS-ENS.
This study uses a unique observational capability from Sorcerer long-duration stratospheric balloons, the only platform currently capable of providing simultaneous, high-frequency vertical soundings and multi-day Lagrangian float trajectories from the upper troposphere (12–14 km). To quantify the unique value of this multi-modal data, we conduct a series of Observing System Experiments (OSEs) assimilating: (1) vertical "yo-yo" profiles only, (2) Lagrangian drift velocities only, and (3) a combined hybrid dataset.
We investigate the hypothesis that while vertical soundings constrain thermodynamic profiles, the assimilation of continuous Lagrangian drift data provides a superior constraint on the upper-level wind field and jet stream positioning. We present an assessment of the technical feasibility of assimilating these diverse geometries into an AI-based background and offer a preliminary evaluation of their relative impact on forecast spread and error reduction. This work represents a novel step toward "observation-adaptive" AI prediction, exploring how next-generation hardware and machine learning models can be coupled to close global observing gaps.
How to cite: Tian, X. and Tindle, A.: Assimilation of Long Duration Stratospheric Balloon Drift and Soundings in AI Weather Foundation Models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3090, https://doi.org/10.5194/egusphere-egu26-3090, 2026.
Satellite observations are critical for numerical weather prediction (NWP), yet the full potential of visible and near-infrared spectral data remains to be exploited. In recent years, ECMWF has been advancing efforts to incorporate visible satellite observations into the analysis and forecasting of clouds and aerosols within the Integrated Forecasting System (IFS). These developments are now reaching operational maturity. We present visible reflectance monitoring and assimilation experiments using observations from various satellite instruments: Spinning Enhanced Visible and Infrared Imager (SEVIRI), Flexible Combined Imager (FCI), Advanced Himawari Imager (AHI), Advanced Baseline Imager (ABI), and Ocean and Land Colour Instrument (OLCI). Our study assesses the first-ever successful experimental assimilation of visible (655 nm) all-sky satellite observations in the IFS. A comprehensive evaluation of these experiments demonstrates that visible reflectance assimilation can improve the model analysis of clouds by better fitting the model trajectory to observations in visible reflectance space. We will also discuss the remaining challenges related to the future operational assimilation of visible observations, including error modelling and biases. Our findings underscore the vast potential of visible spectral observations for operational numerical weather prediction and future re-analysis products.
How to cite: Lupu, C., Necker, T., Quesada Ruiz, S., Firat, V., and Benedetti, A.: Towards operational monitoring and assimilation of visible reflectances at ECMWF, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9826, https://doi.org/10.5194/egusphere-egu26-9826, 2026.
Atmospheric inversions rely on accurately modeled atmospheric conditions, especially wind speeds, often for days in the past. Most atmospheric model products - ECMWF, GFS, and HRRR in our case - report on a scale of 0.1 - 0.5 degrees, or ~10 - 50km. These products can work quite well for inversions at coarse mesoscale resolutions, that model winds for ~100 - 1000km and beyond. However, inversions working at finer scales face significant problems in trying to model winds over a single or even a few meteorological model grid cells.
The MethaneSAT mission attempts to run an atmospheric inversion at a scale of 4km x 4km through the CORE algorithm (Described by Knapp et al at EGU2026), and in trying to extract emissions from instrument observations, we sometimes see plumes that run 30 degrees or more off of model wind directions. Wind speed and especially direction errors at this scale often lead to failed inversions, accounting for a roughly 10% - 20% loss of scenes collected by MethaneSAT, second only to cloudiness.
Luckily, many atmospheric model products are distributed as both a single estimate and as an ensemble product, containing dozens of perturbed forecasts for every time step. To minimize the CORE inference error induced by error in wind estimates, we present a technique for quickly and efficiently analyzing the wind fields in comparison to the concentration Level3 maps in order to select the most likely ensemble member. We utilize a novel Total Variation approach to quantify the expected alignment of wind fields with measured methane gradients, and select the ensemble product whose winds are most likely to produce the observed concentration map.
We demonstrate this technique on both simulated and real MethaneSAT data, and discuss the effects this has on both success rate of the inversion and on residual errors from the inversion, though the approach is not specific to methane, and is broadly applicable in any case where winds are the primary drivers of transit. Special care is taken to identify pathological cases of wind error, and how these could be addressed in the future.
How to cite: Ayvazov, S., Veness, T., Russi, M., LoFaso, N., Knapp, M., Kyzivat, E., Benmergui, J., and Wofsy, S.: Automated Selection of Meteorological Ensemble Members for Inversions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15584, https://doi.org/10.5194/egusphere-egu26-15584, 2026.
The significant impact of Aeolus wind profiles on analyses and forecasts in the tropics underscores the importance of dynamical processes in shaping tropical-extratropical coupling and highlights the need for global wind profile observations. To better understand these processes, we conducted two observing system experiments using the ECMWF data assimilation system: one assimilating observations only within the 15°S-15°N belt, and another assimilating observations only outside this latitude range. Differences between these experiments and a reference experiment using global observational coverage reveal the extent to which the current observing system (GOS) constrains the analyses in the deep tropics and extratropics against influences from unobserved regions.
In the analyses produced by the experiment without tropical observations, only minor signals emerge from the tropics into the extratropics, consistent with a well-observed and well-analysed extratropical circulation in the current GOS. In contrast, we find that assimilating existing tropical observations within the GOS does not sufficiently constrain planetary- and synoptic-scale waves penetrating into the tropics from the extratropics. This echoes previously identified impacts of the Aeolus mission on large-scale tropical circulation. These uncertainties limit the reliability of large-scale tropical circulation in (re)analyses and our ability to predict tropical–extratropical coupling. In particular, uncertainty growth at large scales in medium-range forecasts is likely to dominate over the inverse cascade of effects associated with resolved small-scale processes in numerical models.
How to cite: Žagar, N., De Chiara, G., Healy, S., Sielmann, F., and Wang, C.: Global Observing System and Tropical-Extratropical Coupling, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19833, https://doi.org/10.5194/egusphere-egu26-19833, 2026.
Low Earth Orbit (LEO) observations are fundamental to global Numerical Weather Prediction (NWP), with the Joint Polar Satellite System (JPSS) serving as a critical pillar for monitoring extreme weather events such as wildfires, hurricanes, and floods. To ensure data continuity into the 2030s, NOAA is transitioning from the current JPSS era—supported by Suomi-NPP, NOAA-20, and NOAA-21—toward future missions, including JPSS-3, JPSS-4, and the innovative Near Earth Orbit Network (NEON) program.
Scientific experiments and formulation studies are critical to developing requirements for future sensor capabilities within the NEON architecture. The primary objective is to demonstrate how next-generation initiatives—such as the QuickSounder and the Series-1 microwave sounder missions—will mitigate data gaps and reduce systemic risks to NWP while enhancing Earth system prediction through technical innovation and commercial data integration. Series-1 will host the Sounder for Microwave-Based Applications (SMBA), a successor to the current Advanced Technology Microwave Sounder sensor on JPSS and QuickSounder missions. It features enhanced capabilities, such as hyperspectral measurements, to mitigate the potential influence of Radio Frequency Interference (RFI) while improving vertical resolution.
To develop a credible mission architecture responsive to program constraints, the LEO program sponsored several Observing System Experiments (OSEs) and Observing System Simulation Experiments (OSSEs). Data from various spaceborne platforms were assimilated alongside conventional observations to evaluate performance across diverse metrics, including root mean square errors and ensemble spread differences in atmospheric profiles for key forecast variables—such as temperature, water vapor, geopotential height, and wind fields—and Forecast Sensitivity-based Observation Impacts. Results from the OSSEs provide quantitative evidence of how various observations influence the accuracy of atmospheric profiling and NWP. Experiments also assessed the impact of microwave and infrared observations on tropical cyclone track and intensity prediction. Furthermore, OSSEs explored the complementarity of current satellite assets with a proposed "geostationary ring" of infrared sounders.
This presentation outlines NOAA’s strategic roadmap for evolving LEO capabilities. This roadmap emphasizes international collaboration, commercial satellite data, and the strategic deployment of advanced sensors to ensure a robust, high-fidelity Earth observation network for the coming decades. Finally, the presentation highlights the high impact of microwave and infrared soundings on NWP models.
How to cite: Kalluri, S.: Advancing Global Numerical Weather Prediction: Strategic Roadmaps and Experimental Evaluations of NOAA’s Next-Generation LEO Constellations, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5876, https://doi.org/10.5194/egusphere-egu26-5876, 2026.
Evaluation of short-range precipitation forecasts is sensitive to the spatial non-uniformity of surface observing networks. Gridded NWP precipitation forecasts are commonly verified against AWS/ASOS rain-gauge observations using point-based binary verification. However, this method is affected by a grid-to-point representativeness mismatch and an uneven station distribution, which can bias domain-aggregated verification scores. Consequently, domain-aggregated verification scores can be disproportionately influenced by observations from densely monitored areas.
Beyond sampling biases arising from network non-uniformity, point-based binary verification is also prone to the double-penalty effect. Displacement of precipitation features can be counted simultaneously as a miss at observation sites and a false alarm nearby. Collectively, these limitations motivate verification frameworks that better reflect the spatial nature of precipitation.
Here we assess how observation-network non-uniformity and verification configuration influence reported precipitation forecast skill over the Korean Peninsula. As a practical step to reduce network-density effects, we interpolated point-based precipitation into a gridded observation field and performed area-based verification in parallel with conventional point-based verification for an annual set of short-range forecasts. The results show that reported skill is highly sensitive to the verification framework; for some precipitation thresholds, the area-based approach yields Critical Success Index (CSI) values that are approximately 50% higher than those from point-based verification. The findings highlight that verification design can materially affect the interpretation of precipitation forecast performance in non-uniform observing networks and underscore the need for spatially representative verification frameworks.
How to cite: Oh, S. M., Kim, J. E., and Kang, H.-S.: Impact of observation network non-uniformity on precipitation forecast verification and skill scores, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3622, https://doi.org/10.5194/egusphere-egu26-3622, 2026.
Accuretaly representing the microphysical structure of precipitating systems remains a major challenge in Numerical Weather Prediction (NWP). Cloud and precipitation processes occur at spatial and temporal scales that are not explicitly resolved by current models and must therefore be described through simplified microphysical parameterizations. These parameterizations have a strong impact on key model outputs, such as precipitation intensity, and their improvement requires observations that are sensitive to the vertical structure and microphysical properties of hydrometeors.
Polarimetric Radio Occultations (PRO) provide a complementary observational capability to address this gap. As with standard GNSS Radio Occultations, PRO delivers high vertical resolution thermodynamic profiles under all-weather conditions. In addition, PRO is sensitive to the presence and vertical distribution of hydrometeors through the differential phase shift (ΔΦ), defined as the phase difference between horizontally and vertically polarized GNSS signals. When these signals propagate through non-spherical and/or oriented hydrometeors, differential propagation effects arise, leading to a positive differential phase shift. As a result, PRO measurements offer direct sensitivity to the microphysical structure of precipitating systems.
The accuracy of simulated PRO observables depends on the formulation of the forward operator, particularly on how the relationship between differential phase shift and hydrometeor water content is represented. Recent work has proposed a new forward operator based on a linear relationship that enables the inclusion of scattering-related information as a function of hydrometeor type. In this study, we evaluate the performance of this updated PRO forward operator under Atmospheric River (AR) conditions, which are characterized by intense moisture transport and strong precipitation. Simulations are compared with those produced using the offline forward operator currently implemented at ECMWF to assess whether the new formulation could serve as a viable replacement for operational applications.
Beyond this potential operational impact, the new forward operator enables the use of PRO as a constraint on cloud and precipitation microphysics. By exploiting the Atmospheric Radiative Transfer Simulator (ARTS) scattering database, we analyze the sensitivity of PRO observables to the scattering properties of different particle habits, providing insight into the extent to whether PRO measurements can discriminate between microphysical assumptions and improve the representation of precipitating systems in NWP models.
How to cite: Paz, A., Padullés, R., Cardellach, E., Lonitz, K., and Vidal, V.: Evaluating Polarimetric Radio Occultations for constraining precipitation microphysics in NWP, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19084, https://doi.org/10.5194/egusphere-egu26-19084, 2026.
Can we control the butterfly effect? This study addresses the fundamental question of whether we can control chaotic weather systems by taking advantage of their sensitivity to initial conditions. Specifically, we explore a theoretical framework to control the system beyond the predictability limit, where infinitesimal perturbations grow to alter the macroscopic trajectory. Based on the Control Simulation Experiment (CSE) framework, we focus on the Duality Principle, which posits that the control problem is mathematically dual to data assimilation (DA). In this view, adding interventions to nature for control is equivalent to adding analysis increments to correct the model forecast for DA. Therefore, controllability can be understood as the synchronization of the nature trajectory with a target model trajectory, analogous to filter convergence in DA. Using the Lorenz 63 model, we present a compelling case study that highlights an apparent paradox within this duality. Our previous paper showed that intervening only in the z-variable was effective for controlling the full system ("z-only intervention"). However, in the dual problem of DA, observing only the z-variable leads to filter divergence ("z-only observation"). Why does intervention succeed where observation fails, despite their theoretical duality? In this presentation, we address this asymmetry and discuss the underlying dynamics of the target trajectory. Based on the Duality Principle, we establish a theory for controlling chaotic systems beyond the predictability limit, opening new pathways for mitigating extreme weather events.
How to cite: Miyoshi, T.: Harnessing the Butterfly Effect: A Duality-Based Framework for the Efficient Control of Extreme Weather, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-2914, https://doi.org/10.5194/egusphere-egu26-2914, 2026.
Background and Objectives
As typhoons intensify due to climate change, typhoon modification technology has gained renewed attention. Japan's Moonshot Goal 8 program, launched in 2022, has accelerated technological development in this area. However, social implementation requires addressing not only technical feasibility but also ethical, legal, and social implications (ELSI). While previous studies have qualitatively organized the expectations and concerns held by people regarding typhoons and typhoon modification technologies through dialogue sessions, large-scale investigations of public cognitive structures remain limited. This study aimed to elucidate people’s perceptions regarding expected benefits, ELSI concerns, and relevant stakeholders through a nationwide questionnaire survey.
Methods
1,000 participants comprised across Japan, recruited through a web-based survey with quota sampling by gender, age, and residential region. The questionnaire assessed expected benefits (9 items), ELSI concerns (15 items), stakeholders (14 items), all rated on five-point scales. Cluster analysis was conducted for each domain, followed by two-way ANOVA with residential region (four prefectures most affected by typhoons vs. others) and cluster.
Results
Cluster analysis of expected benefits revealed three clusters: disaster risk reduction, social and broader benefits, and economic applications of technological development. Two-way ANOVA revealed a significant main effect of cluster (F = 270.87, p < .01, η² = .21), with disaster risk reduction rated highest. Neither regional main effect nor interaction was significant.
Cluster analysis of ELSI revealed six clusters: environmental and ecological risks, technological uncertainty and governance, economic costs, decreased disaster preparedness awareness, transformation of social structures and views of nature, and risks of misuse and international conflict. Both cluster (F = 120.13, p < .01, η² = .11) and regional main effects (F = 5.52, p < .05, η² = .01) were significant, with four prefectures showing heightened ELSI concerns. In addition, concerns regarding the cluster of economic costs were the highest.
Cluster analysis of stakeholders revealed five clusters: policy and technical experts, local practitioners and economic actors, general citizens and disaster victims, education and ethics professionals and foreign governments. The cluster main effect was significant (F = 363.12, p < .01, η² = .27), with policy and technical experts deemed most essential. A significant interaction (F = 2.97, p < .05) indicated that four prefectures prioritized consultation with foreign governments over input from education and ethics professionals.
Conclusions
The results of this study indicate that public recognizes both multifaceted expected benefits and ELSI regarding typhoon modification technology. Prefectures of typhoon-affected regions exhibit higher concern of ELSI. In addition, public emphasizes the involvement of stakeholders and recognizes the need for inclusive consensus-building that includes citizens and disaster victims, rather than leaving technological decision-making solely to experts. These findings highlight the importance of communication and consensus-building frameworks that take public awareness structures into account in societal decision-making related to typhoon modification technologies.
How to cite: Su, Y. and Matsuyama, M.: Public Perception of Typhoon Modification Technology: Examining Expected Benefits, ELSI, and Stakeholders, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6219, https://doi.org/10.5194/egusphere-egu26-6219, 2026.
Extreme weather events are intensifying under climate change, yet recent advances in weather prediction operate within a forecast-only paradigm that does not directly mitigate impacts once an extreme event is anticipated. Motivated by chaos control theory, we explore whether small, instability-aware perturbations can leverage intrinsic atmospheric sensitivity to influence extreme weather evolution within an AI-based forecasting framework. We use the Aurora foundation model and identify dynamically sensitive perturbation locations using Finite-Time Lyapunov Exponent (FTLE) diagnostics. To implement a physically interpretable intervention compatible with foundation models, we introduce an idealized cloud seeding based perturbation operator that mimics condensation-driven latent heat release applied in the lower–mid troposphere. In a case study, these upstream perturbations induce coherent downstream changes in integrated vapor transport, leading to reduced peak landfall intensity and slower precipitation accumulation. These results demonstrate that instability-aware perturbations within an AI foundation model can induce dynamically meaningful downstream impacts, providing a first step toward bridging chaos control concepts and data-driven weather prediction.
How to cite: Liu, M., Huang, Q., and Lall, U.: Instability-Aware Perturbations of Extreme Events in an AI Weather Foundation Model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22120, https://doi.org/10.5194/egusphere-egu26-22120, 2026.
Extreme heat is emerging as one of the deadliest weather-related hazards under global warming. In response, a growing range of weather and climate interventions has been proposed to locally mitigate extreme thermal stress. The spraying of water into the air is a well-established technique for direct evaporative cooling at small scales, such as in urban or industrial environments. In this study, we assess the potential for larger-scale sea water spraying as a localized extreme heat intervention.
We first quantify the maximum evaporative cooling potential of liquid water spraying and its dependence on air temperature and relative humidity. Under hot and dry conditions, theoretical cooling exceeding 20 °C can be achieved, providing an upper bound for realistic applications. We then employ Large Eddy Simulations (LES) to investigate kilometer-scale cooling of the lower atmosphere in coastal regions experiencing sea-breeze conditions in hot and dry climates. A key innovation of this study is the introduction of sea water spraying from wind turbines, which we compare with spraying from lower infrastructures such as platforms or boats.
The resulting impacts on near-surface temperature and human thermal comfort indices are evaluated, highlighting potential cooling benefits during extreme heat events. Finally, we examine how such cooling influences plume rise, a key process for Marine Cloud Brightening. Our results demonstrate that physically bounded, localized atmospheric interventions may offer a useful tool to mitigate extreme heat in vulnerable regions, while providing insight into the effectiveness and limitations of weather- and climate interventions.
How to cite: Broerze, A., de Roode, S., and Russchenberg, H.: Cooling the air: assessing the evaporative cooling potential of sea water spraying as an extreme heat intervention, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18836, https://doi.org/10.5194/egusphere-egu26-18836, 2026.
Posters on site: Tue, 5 May, 08:30–10:15 | Hall X5
A nocturnal surface-based temperature inversion refers to a phenomenon where air temperature increases with height. Such inversions are of interest because they play a significant role in enhancing the risk of frost events during the autumn–spring period and in the accumulation of air pollutants near the surface, with potential adverse impacts on human health.
The key conditions for nocturnal inversion formation are: the absence of cloud cover, which enhances radiative cooling, and weak winds or calm conditions, which minimize vertical air mixing and thus intensify near-surface cooling. Based on these criteria, we analyzed meteorological observation data for the period 2011–2025 and selected nights when the average cloud cover over Poland did not exceed 0.5 oktas and wind speed remained below 3 m/s. An additional requirement was sunset before 18 UTC (the forecast initialization time) to exclude the influence of incoming solar radiation. For the selected cases, we chose stations where cloud cover remained at 0 oktas and wind speed did not exceed 3 m/s throughout the modeling period.
For the modeling, we used the COSMO-2k8 ensemble prediction system with perturbations applied to the surface level soil temperature (T_SO) along with a separate deterministic run.
At the first stage, the 20 ensemble members were split into two groups: group 1 with perturbations in initial conditions only and group 2 with perturbations in both initial and boundary conditions.
In the near-surface layer, most cases show significant deviations between the modeled and observed temperature profiles, both positive and negative. For almost all simulated events, a characteristic feature is the rapid decrease in the amplitude of surface temperature among ensemble members. As a result, the amplitude of air temperature at 2 m and higher levels also decreases.
At the second stage, a new approach to introducing perturbations into the T_SO was implemented. A new 40-member ensemble with different methods of initialization of perturbations was generated to hold the spread level during the model forecast. The members were divided into four groups of 10:
- group 1: perturbations applied only to the initial conditions, with temperature deviations of ±0.5 and ±1.0 K and noise amplitude of 1–2 K during the first two time steps.
- group 2: perturbations applied to both the initial and boundary conditions, with gradual accumulation of shift and amplitudes from 0.05 to 1.0 K; due to the simulation time step, the maximum amplitude over 12 h reaches ±0.3 K.
- group 3: perturbations up to ±1.0 K, accumulated during the first ~100 minutes (240 steps).
- group 4: perturbations up to ±1.0 K, accumulated rapidly during the first ~17 minutes (40 steps).
The new perturbation scheme demonstrated not only the preservation of spread among individual ensemble members, but also its propagation up to a height of about 900 hPa.
How to cite: Mazur, A., Tabalchuk, T., and Wyszogrodzki, A.: Impact of different surface temperature perturbation schemes on the simulation of nocturnal temperature inversions in ensemble modeling , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-530, https://doi.org/10.5194/egusphere-egu26-530, 2026.
A new assimilation module for AGRI (Advanced Geostationary Radiation Imager) carried on Fengyun-4B (FY-4B) is developed within the WRFDA (Weather Research and Forecasting model’s Data Assimilation) system. The impacts of assimilating FY-4B AGRI clear-sky and all-sky data are evaluated with cycling assimilation experiments for the typhoon Doksuri (2023) and Talim (2023) forecast. For typhoon Doksuri, compared with the benchmark experiment, AGRI assimilation brings a better analysis and forecast for atmospheric variables (wind, temperature and humidity). Meanwhile, clear error reductions for typhoon track and intensity forecasts are achieved with clear-sky and all-sky AGRI assimilation. The positive impact on landing precipitation prediction is also obtained by verifying with GPM (Global Precipitation Measurement) precipitation data. Moreover, AGRI all-sky assimilation yields better typhoon forecasts than clear-sky assimilation. In addition, the channel selection sensitivity for AGRI assimilation is also assessed with group experiments. It is suggested that with the assimilation of a new water vapor channel at 7.42 μm, which is newly added in FY-4B, multiple-channel assimilation shows greater benefit for forecast than single-channel assimilation. The same positive impact of AGRI assimilation is also present in typhoon Talim (2023) forecast. Overall, the all-sky assimilation of FY-4B AGRI water vapor channels data is beneficial for the typhoon forecast.
How to cite: Yang, C., Zhong, T., and Min, J.: A preliminary study of FY-4B AGRI all-sky assimilation by WRFDA for tropical cyclones , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1848, https://doi.org/10.5194/egusphere-egu26-1848, 2026.
The impact of radiance data assimilation from Advanced Technology Microwave Sounder (ATMS) into the Center for Weather Forecasting and Climate Studies of the National Institute for Space Research (CPTEC/INPE), Brazil, is investigated and discussed in this work. The recent discontinuation of the Advanced Microwave Sounding Unit (AMSU-A) in the NOAA series satellites (NOAA-15, 18, and 19) gives the ATMS sensor greater relevance in the process of assimilating microwave radiance data, and this fact is the motivation of this study. The ATMS data used in this study are from the Joint Polar Satellite System (JPSS) series and the National Polar-orbiting Partnership (NPP) satellites. Therefore, understanding and studying the impact of assimilating ATMS sensor radiance channels becomes essential for developing a more realistic numerical weather forecast, especially in extreme atmospheric conditions such as storms, cyclones, and hurricanes. Thus, this work presents results on the impact of ATMS sensor data assimilation on the numerical forecast of Melissa hurricane, which reached category five on October 27, 2025, in the Caribbean Sea, Central America. The Numerical Modeling and Data Assimilation System (SMNA) is used in this study, which is composed of a Gridpoint Statistical Interpolation (GSI) System coupled to the Brazilian Global Atmospheric Model (BAM). Other types of data, such as radiance from AMSU-A data from MetOp (Meteorological Operational satellite programme) satellites, Atmospheric Motion Vectors (AMV) data from geostationary and polar orbit satellites, radio occultation GNSS (Global navigation Satellite System) data, dropsondes, radiosondes, and pilot balloons, were also used in the assimilation process. Different experiments were conducted to explore the radiance channels available in the ATMS sensor and assess their contribution to the improvement of the predictability of the Melissa hurricane. The results reported here are the first using the ATMS data in this center, and they are essential for establishing a consistent radiance database for the assimilation process in the new Brazilian climate prediction model being developed by CPTEC/INPE and partners, the Model for Ocean-laNd-Atmosphere PredictioN (MONAN), in phase of imprementation and test.
Keywords: Radiance, ATMS, Data assimilation, Numerical weather prediction, Melissa hurricane.
Acknowledgment: This work was supported by the National Council for Scientific and Technological Development - CNPq (Process No. 304388/2022-0).
How to cite: Viezel, C., Sapucci, L., and Ranieri, V.: Accessing the contribution of Advanced Technology Microwave Sounder (ATMS) in the Brazilian Numerical Modeling and Data Assimilation System, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-2151, https://doi.org/10.5194/egusphere-egu26-2151, 2026.
Flood inundation damage to the socioeconomic landscape caused by short-term extreme rainfall has been a primary concern for years, particularly regarding shifts in precipitation patterns driven by climate change. To mitigate these impacts, many studies have prioritized the enhancement of flood prevention infrastructure in high-risk areas. However, the construction of permanent facilities requires rigorous scientific assessment and significant investments of time and capital. As an alternative to traditional infrastructure, weather modification strategies are being explored to reduce societal impacts. These strategies include suppressing the formation of cloud systems by altering air currents through the construction of physical obstacles, as well as removing water vapor via proactive cloud seeding. To evaluate the efficacy of seeding in reducing flood risk, we utilized simulated rainfall data from the Weather Research and Forecasting (WRF) model, which incorporates a seeding core module to test rainfall control effectiveness within the Kurokawa River basin, Kyushu, Japan. The results demonstrate that a 10% reduction in 24-hour basin-averaged rainfall led to a 20% decrease in peak discharge (excluding overflow). Correspondingly, the inundation extent was reduced by 20% across different scenarios, with the most significant improvements observed in high-depth areas. To quantify the benefits for stakeholders and the government, this study evaluated potential economic losses. The reduction in inundation successfully mitigated agricultural and property economy losses by approximately 10%. Furthermore, we assessed threats to human life by proposing critical water depth thresholds specific to elderly populations (aged over 65 years old) and younger residents based on housing types and government data. These metrics confirm that rainfall control strategies effectively safeguard the community and the broader economy against climate-driven flood disasters.
How to cite: Chang, J. and Yorozu, K.: Impact of Rainfall Control on Socio-economic Flood Risk Assessment in the Kurokawa River Basin, Japan, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8827, https://doi.org/10.5194/egusphere-egu26-8827, 2026.
Reliable precipitation forecasts are crucial for water resource management and flood disaster early warning. However, numerical weather prediction (NWP) products often suffer from systematic biases, limiting their applicability across different regions. To address this issue, this study proposes a Two-step Time-dependent Enhanced Informer (TTEInformer) method for precipitation post-processing. This method employs a two-step classification and regression correction framework. It classifies precipitation into wet and dry days, then performs regression correction on samples identified as wet days, while dry day samples are maintained as zero values. To better capture temporal dependencies, TTEInformer augments the Informer model with a feature extraction module and a bidirectional gated recurrent unit (BiGRU) module. The study evaluates the proposed method over the Yalongjiang River basin upstream of the Yajiang hydrological station and compare it with multiple deep learning baselines. The results indicate that all corrected products substantially reduce forecast errors relative to the raw NWP precipitation. The proposed model demonstrates outstanding performance, achieving an R value of over 0.9 and significantly reducing forecast errors compared to other models. Moreover, the two-step correction framework effectively enhances model correction accuracy compared to traditional direct correction strategies, with notable improvements in correcting light precipitation events. The work provides a reliable post-processing method for hydrometeorological applications in precipitation forecasting.
How to cite: Xu, X., Qin, H., Yang, L., and Li, C.: A Two-step Time-dependent Enhanced Informer method for numerical weather prediction post-processing, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8864, https://doi.org/10.5194/egusphere-egu26-8864, 2026.
The Portable Model for multi-scale Atmospheric Prediction (PMAP) is a numerical model currently under development at ECMWF and ETH aimed at large-eddy simulation (LES) of atmospheric flows at the weather-system scale. It builds on a non-hydrostatic, locally conserving, finite-volume, 3D semi-implicit dynamical core coupled to state-of-the-art physics parametrizations. Written entirely in Python, it leverages the GT4Py domain-specific language to achieve high performance and portability – running straightforwardly on laptops and GPU-accelerated HPC systems alike. The systematic separation of concerns between domain science (physics, numerics) and performance engineering (parallelisation, hardware optimisation) provides new avenues for model development, setup, and refinement, which we present in two related contributions.
In this first contribution, we highlight PMAPs strengths as a numerical model framework to flexibly develop and refine numerical algorithms, as well as to implement and extend diagnostics to address research questions in atmospheric sciences. This is illustrated here with an LES tracer transport experiment over complex terrain. We first demonstrate PMAPs capability to perform decametre-scale LES using a height-based terrain-following vertical coordinate in steep terrain with slopes exceeding 80°. This is possible thanks to the locally conservative advection and the 3D semi-implicit integration scheme, which ensures stability of the integration and regularization of the flow. Moreover, we exemplarily show how the Python-based model formulation facilitates evaluating and improving various aspects of flux-form semi-Lagrangian tracer transport schemes in terms of directional splitting approaches and monotonic limiters, and how these impact simulated power spectra of the tracer fields. Lastly, we present how available model diagnostics can easily be extended to perform targeted analyses of sensitivities to implementation details.
How to cite: Papritz, L., Krieger, N., Kühnlein, C., Faghih-Naini, S., Ehrengruber, T., Ubbiali, S., Vollenweider, G., and Wernli, H.: The Portable Model for multi-scale Atmospheric Prediction (PMAP): Capabilities and development workflows , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9248, https://doi.org/10.5194/egusphere-egu26-9248, 2026.
The Portable Model for multi-scale Atmospheric Prediction (PMAP) is a numerical model currently under development at ECMWF and ETH aimed at large-eddy simulation (LES) of atmospheric flows at the weather-system scale. It builds on a non-hydrostatic, locally conserving, finite-volume, 3D semi-implicit dynamical core coupled to state-of-the-art physics parametrizations. Written entirely in Python, it leverages the GT4Py domain-specific language to achieve high performance and portability – running straightforwardly on laptops and GPU accelerated HPC systems alike. The systematic separation of concerns between domain science (physics, numerics) and performance engineering (parallelisation, hardware optimisation) provides new avenues for model development, setup, and refinement, which we present in two related contributions.
In this second contribution, we present two examples that demonstrate the benefits of resolving small-scale processes for simulating real weather and showcase how PMAP can flexibly be used as a research tool for atmospheric dynamics. (i) First, we demonstrate the benefits of sub-kilometer spatial resolution for accurately simulating the intensification of Hurricane Melissa in late October 2025 as compared to an operational km-scale model. (ii) Second, we present results from a process study of storm Éowyn, which brought record-strong winds to the British Isles in January 2025. LES of the low-level jet along the storm’s bent-back front not only accurately predicts peak wind speeds, but also resolves individual wind gusts in close agreement with observations. We highlight a range of diagnostic tools implemented in PMAP that make such analyses straightforward. Moreover, we demonstrate how the model can be employed to shed light on the atmospheric dynamical processes leading to the storm’s rapid intensification. Specifically, we show the importance of latent heat release by performing modified latent heating experiments, which in the PMAP framework are straightforward to set up, and quantitatively corroborate the crucial impact of cloud microphysical processes for the rapid intensification of Éowyn.
How to cite: Krieger, N., Papritz, L., Kühnlein, C., Hauser, S., Ubbiali, S., Vollenweider, G., and Wernli, H.: The Portable Model for multi-scale Atmospheric Prediction (PMAP): Sub-kilometre simulations of recent extreme weather, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9320, https://doi.org/10.5194/egusphere-egu26-9320, 2026.
Cloud-affected infrared satellites constitute a promising data source for numerical weather prediction models as they contain crucial information on atmospheric clouds and convective activity. Their sensitivity to both hydrometeor content and cloud top height, however, leads to a very non-Gaussian distribution of first-guess (FG) departures, which violates a fundamental assumption of current data assimilation schemes. To mitigate this issue, various cloud-dependent error models for normalizing the departures have been proposed (Geer and Bauer, 2011; Harnisch et al., 2016; Okamoto et al., 2014). In the current presentation, we revisit these error models and propose a refined approach that leads to a more Gaussian distribution.
We quantify the performance of these error methods in detail when applied to one-month infrared observations from the Spinning Enhanced Visible and Infrared Imager (SEVIRI) onboard the Meteosat Second Generation and simulations from the weather forecast model AROME (Application of Research to Operations at Mesoscale) over Austria. While all methods can successfully yield approximately Gaussian FG departure distributions when normalized by the observation error, an additional quality control and a minimum error threshold is necessary for some of them.
While previously published methods estimate the observation error using the average cloud effects from the model and observation spaces, we also introduce a new method that uses the maximum values from these two spaces for observation error calculation. Results show that the new method systematically outperforms the previous methods at no additional cost. Lastly, we analyze the performance of different practical implementation choices, such as using a linear or polynomial fit.
Reference:
Geer, A. J. and Bauer, P.: Observation errors in all-sky data assimilation, Quarterly Journal of the Royal Meteorological Society, 137, 2024–2037, https://doi.org/10.1002/qj.830, 2011.
Harnisch, F., Weissmann, M., and Periáñez, Á.: Error model for the assimilation of cloud-affected infrared satellite observations in an ensemble data assimilation system, Quarterly Journal of the Royal Meteorological Society, 142, 1797–1808, https://doi.org/10.1002/qj.2776, 2016.
Okamoto, K., McNally, A., and Bell, W.: Progress towards the assimilation of all-sky infrared radiances: An evaluation of cloud effects, Quarterly Journal of the Royal Meteorological Society, 140, 1603–1614, https://doi.org/10.1002/qj.2242, 2014.
How to cite: shi, B., Griewank, P., Meier, F., Min, J., and Weissmann, M.: Revisiting error models for the assimilation of infrared satellite radiances, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9670, https://doi.org/10.5194/egusphere-egu26-9670, 2026.
Cyclones are frequent meteorological phenomena in Brazil, commonly associated with intense rainfall and strong winds. Most of them are extratropical systems, although subtropical and, more rarely, tropical cyclones also occur. Due to the impacts they cause, understanding the intensity and dynamics of these systems is essential for the scientific community, particularly regarding the ability of numerical models to predict their formation, associated precipitation, and wind fields. In February 2024, Cyclone Akará developed near the southeastern coast of Brazil and, as the third system classified as a tropical cyclone in the South Atlantic, represented a unique opportunity for study. With the development of the MONAN model (Model for Ocean-laNd-Atmosphere predictioN), built through a multi-institutional effort, the need arose to evaluate its performance in forecasting extreme events, especially those associated with cyclones that directly affect the country. This study aimed to analyze MONAN’s ability to represent precipitation fields and mean sea level pressure (MSLP) associated with Cyclone Akará, comparing different versions of the model from the implementation/development process. To achieve this, an object-based approach was adopted using the MODE tool (Method for Object-Based Diagnostic Evaluation). ERA5 reanalysis data (ECMWF) were used as reference to compare the cyclone’s position and intensity, while precipitation was assessed using the MERGE product (CPTEC/INPE), a high-resolution, blended rainfall dataset that combines satellite information with in situ gauge measurements to create a homogeneous daily record for South America. The evaluation focused on the intensification phase of the cyclone (February 16), considering attributes such as centroid distance, area, and total interest of the objects, in addition to traditional metrics such as RMSE, BIAS, and CSI. The results showed that the updated version performed better in forecasting intense precipitation. The detailed assessment of this event offered valuable insights for the continuous improvement of the model, contributing to the strengthening of the Brazilian numerical weather prediction system.
Keywords: Cyclone Akará, MONAN model, Precipitation, Model Evaluation, MODE.
Acknowledgment: This study was supported by the São Paulo Research Foundation - FAPESP (Process No. 2025/06119-7) and National Council for Scientific and Technological Development - CNPq (Process No. 304388/2022-0).
How to cite: Antunes Ranieri, V., Fernando Sapucci, L., Couto de Souza, D., Leite da Silva Dias, P., Khamis, E., Viezel, C., and Freitas, S.: Performance of the MONAN Model in Forecasting Cyclone Akará and Associated Precipitation Using Object-Oriented Evaluation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10950, https://doi.org/10.5194/egusphere-egu26-10950, 2026.
From 17 to 22 July 2021, Henan Province in China experienced an exceptionally severe rainfall event (hereafter referred to as the “7·21” rainstorm). In particular, record-breaking local hourly precipitation occurred in Zhengzhou on 20 July, posing an unprecedented challenge to mesoscale numerical weather prediction (NWP) systems. Uncertainty is inherent in NWP, and ensemble forecasting has increasingly become the consensus approach for quantifying such uncertainty. The use of growing initial perturbations is essential for achieving high ensemble forecast skill. To investigate the influence of initial perturbations on forecast errors during this extreme rainfall event, this study applies the orthogonal conditional nonlinear optimal perturbation method (O-CNOP-Is) to construct an ensemble perturbation strategy tailored to the Henan rainstorm. The aim is to improve the representation of extreme precipitation and its associated forecast uncertainty, thereby providing new technical support for the prediction and early warning of similar high-impact events in the future.
The O-CNOP-Is represent a set of mutually orthogonal growing initial perturbations that satisfy prescribed physical constraints and exhibit the maximum nonlinear evolution at the forecast time. In this study, an O-CNOP-Is computation framework is established using the regional mesoscale Weather Research and Forecasting (WRF) model. The Global Ensemble Forecast System (GEFS) provided by the National Centers for Environmental Prediction is used as the source ensemble of initial perturbation samples. An ensemble projection algorithm is employed to derive a set of O-CNOP-Is perturbation fields specifically targeted at the “7·21” rainfall event, fully accounting for the nonlinear evolution of the model. These O-CNOP-Is are then superimposed onto the background field to generate an ensemble whose members embody the strongest features of uncertainty growth. The resulting ensemble forecasts are compared with those from the GEFS to assess the effectiveness of CNOP-type perturbations in ensemble forecasting of extreme precipitation.
The numerical results indicate that the initial perturbations provided by O-CNOP-Is are physically reasonable for regional ensemble prediction. The perturbation amplitudes increase with time, and their spatial structures effectively reflect the baroclinic instability characteristics of the evolving atmosphere. Compared with GEFS perturbations, O-CNOP-Is contain more uncertainty information at the initial time and exhibit stronger perturbation growth at the forecast time. Throughout the forecast period, the O-CNOP-Is ensemble displays larger ensemble spread that better matches the forecast errors. Moreover, the ensemble mean forecast shows notable improvements in reproducing both the location and peak intensity of extreme precipitation centers.
Overall, the results demonstrate that the conditional nonlinear optimal perturbation approach is a highly promising method for capturing the dominant error growth modes in the Henan rainstorm. It effectively enhances ensemble forecast skill for high-impact, strongly nonlinear extreme weather events such as the “7·21” Henan rainfall, and provides a solid scientific basis and practical foundation for the development of regional mesoscale ensemble forecasting systems.
How to cite: liu, Y. and liu, J.: The Role of Orthogonal Conditional Nonlinear Optimal Perturbations (O-CNOPs) in Ensemble Forecasting of Extreme Rainfall: Improvement and Physical Perturbation Mechanisms, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12835, https://doi.org/10.5194/egusphere-egu26-12835, 2026.
Coupled data assimilation (CDA) is a core method for achieving seamless forecasting with Earth system models (ESMs). It provides high-quality initial conditions for coupled models, minimizes inter-component imbalances to the greatest extent, and better utilizes observational networks. Based on whether cross-component information transfer is achieved, CDA can be classified into weakly coupled data assimilation (WCDA) and strongly coupled data assimilation (SCDA). SCDA, which incorporates cross-component information transfer, can produce more balanced and consistent analysis fields, thereby improving forecast skill. According to ECMWF (refer to the literature Coupled data assimilation at ECMWF: current status, challenges and future developments), strongly coupled data assimilation enables the transfer of observational information across different Earth system components during the assimilation phase.
This study employs the high-resolution version of the global sea-land-air-ice system model FGOALS-g3, developed by the State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics (LASG) at the Institute of Atmospheric Physics, Chinese Academy of Sciences, along with the Dimension-Reduced Projection Four-Dimensional Variational (DRP-4DVar) system, to conduct single-point coupled assimilation experiments. The aim is to investigate how cross-component covariance specifically regulates the transfer of observational information between Earth system components. The experimental results show that within the coupled framework, assimilating only a single-point surface pressure observation in the atmosphere not only generates analysis increments for atmospheric wind and temperature but also influences the state of the ocean surface layer. Similarly, assimilating only a single-point sea surface temperature observation not only produces analysis increments for sea surface height but also induces responses in the lower atmospheric state.
How to cite: Fu, J. and Liu, J.: Multi-Sphere Covariance Analysis for Coupled Assimilation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12842, https://doi.org/10.5194/egusphere-egu26-12842, 2026.
Starting in the early 2020s, artificial intelligence-based models (Machine Learning Weather
Prediction, MLWP) play an increasingly important role in weather forecasting. Their widely known
advantages make the use of MLWP models desirable in the future; however, it is important to have
a thorough understanding of the nature of their prediction fields and their strengths and weaknesses.
Most deterministic MLWP models have been trained using a Mean Squared Error (MSE)-
based cost function. The effective resolution of these models decreases as the forecast time
increases to avoid the double penalty occurring in the evaluation of sharp precipitation fields. It has
been shown that the effective resolution of a MLWP model forecast field is equivalent to the
ensemble average for forecast time period which had been explicitly included in the model cost
function. Since this similarity to the ensemble mean provides a strong basis for the operational
interpretation of the MLWP forecasts, it is important to verify it using model data for different
atmospheric variables.
In our work, we examined the effective resolution of a particularly important variable in
forecasting practice, the 6-hour precipitation field, by comparing the ECMWF-AIFS MLWP model,
the ECMWF HRES ensemble average, and the ERA5 reanalysis data. Effective resolution is usually
determined by examining the spectrum of model fields. However, since precipitation is a highly
localized and non-continuous atmospheric variable, we considered it appropriate to quantitatively
examine the smoothing of the fields in addition to spectral analyses. The aim of our calculations
was to determine how long the smoothing of ECMWF-AIFS precipitation forecasts follows the
ensemble average and how much they deviate from reality as represented by ERA5. We defined
smoothing as the extent to which the ERA5 precipitation field, which most accurately represents the
real field, needs to be smoothed using Gaussian filters in order for its sharpness indicators to match
those of the given forecast. Since smoothing in the precipitation field of MLWP models occurs not
only in the spatial structure but also in the reduction of extremes, we have developed a separate
method for examining sharpness that focuses exclusively on extreme values, based on the
examination of the exponential approximation of the global distribution of 6-hour precipitation
intensity.
Our results show that the agreement between the MLWP model and the ensemble mean
smoothing over the training period is only achieved in the case of extratropical precipitation
dominated by synoptic-scale processes. We did not observe such an agreement at all in tropical
areas with predominantly convective precipitation. We obtained similar results using a sharpness
metric developed for extreme values, i.e., the MLWP model's smoothing of extreme values only
matched the ensemble mean in the extratropics These results indicate that MLWP models can only
successfully predict processes on a sufficiently large scale and smooth small-scale processes such as
convection.
How to cite: András, C., Ádám, L., and Ákos, V.: Quantification of 6-hour precipitation field smoothing in deterministic Machine Learning-based Weather Forecasts, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13006, https://doi.org/10.5194/egusphere-egu26-13006, 2026.
How to cite: Najari, N., Potthast, R., Keller, J., Hollborn, S., and Deppisch, T.: Improving ICON-DREAM Cloud Information through AI-driven Data Assimilation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15804, https://doi.org/10.5194/egusphere-egu26-15804, 2026.
As the demand for high-resolution weather forecasts continues to increase, operational numerical weather prediction centers face challenges in maintaining model consistency across horizontal scales while accurately representing region-specific physical processes. This study evaluates the KIM-Regional/Local system, a unified modeling framework employing the Korea Physics Parameterization Package (KPPP) optimized for the Korean Peninsula. The system operates on two nested scales: a 3-km regional domain covering East Asia (5-day forecasts) and a 1-km local domain covering the Korean Peninsula (2-day forecasts). A key distinction in their configuration is the treatment of convection: cumulus parameterization is applied in the 3-km regional model, whereas the 1-km local model is configured as convection-permitting, explicitly resolving deep convection without a cumulus scheme. Both domains share identical dynamical cores and other KPPP components.
KPPP introduces several key advancements optimized to regional environmental conditions. These developments include refinements to microphysical processes, along with enhanced radiative parameterizations that account for all-sky radiation and topographic slope and shading effects. The land surface scheme includes observation-based refinements to tree and canopy height, which are expected to improve the representation of surface fluxes. To enhance initial and boundary conditions, an oceanic mixed-layer model is activated, and spatially and temporally varying Charnock coefficients are introduced for sea surface roughness calculations, replacing the previously used constant values. Together, these developments are introduced to better represent key physical processes in the model, while accounting for computational efficiency required for operational application.
Forecast performance was evaluated for representative summer and winter periods through comprehensive verification of surface and upper-air variables against observations, alongside quantitative precipitation forecasts assessment. The most pronounced improvements are found in surface verification results, particularly in surface wind speed, where the Root Mean Square Error (RMSE) is substantially reduced compared to the current operational system. Overall enhancements are also evident in precipitation performance, with reduced biases across forecast ranges in both the 3-km and 1-km domains. These results indicate that region-specific physics development provides a robust pathway toward operationally reliable high-resolution prediction systems over regions with complex terrain.
How to cite: Jeong, S., Lee, J., Lee, E., and Lee, Y.-H.: Evaluation of KIM-Regional/Local Forecast Model Performance Based on the Korea Physics Parameterization Package, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15894, https://doi.org/10.5194/egusphere-egu26-15894, 2026.
The forecast performance of numerical weather prediction models strongly depends on the accuracy of the initial conditions, which is largely determined by the quality of observations used in data assimilation. Aircraft observations provide three-dimensional atmospheric information over data-sparse regions, such as the upper level and oceans, thereby complementing the spatial inhomogeneity of the global observations and contributing to improvements in initial conditions. However, aircraft temperature observations are known to exhibit positive biases compared to radiosonde observations, and correcting these biases is an important challenge for improving forecast performance.
The Korea Meteorological Administration (KMA) has been operating the latest version of the Korean Integrated Model (KIM) v4.0 operationally since May 2025, but aircraft temperature bias correction has not yet been applied. In this study, two experiments were conducted in July 2025 to quantitatively evaluate the impact of aircraft temperature bias correction on forecast performance using KIM v4.0 at an approximately 25 km horizontal resolution: one experiment applied bias correction globally, while the other applied it selectively over regions north of 30°N. This latitude threshold was determined based on the spatial distribution characteristics of temperature biases identified in the KIM v4.0. Aircraft observations were stratified into three vertical layers (lower: 1050–700 hPa, middle: 700–300 hPa, upper: 300–150 hPa), and aircraft ID temperature bias correction coefficients were derived and applied during the observation preprocessing step.
The experiment applying bias correction north of 30°N showed overall improved forecast performance of temperature and geopotential height over the Northern Hemisphere and North America compared to the globally applied experiment. Additionally, performance improvements were observed in East Asia during the later forecast periods (days 4–5), with lower-level specific humidity and temperature showing improvements of 1.18% and 1.55%, respectively. These results demonstrate that selective temperature bias correction considering the spatial characteristics of aircraft observations can contribute to improving forecast performance in numerical weather prediction models.
How to cite: Lee, Y., Ha, J.-H., and Lee, Y.: Impact of Aircraft Temperature Bias Correction in the Korean Integrated Model (KIM) , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15923, https://doi.org/10.5194/egusphere-egu26-15923, 2026.
Data assimilation (DA) seeks to provide the most likely estimate of the true state of the atmosphere or ocean by combining a background estimate from a numerical model with observations, each weighted by their respective error covariances. For computational reasons, diagonal covariances together with variance inflation have traditionally been favoured for the observational error covariance. However, recent studies have shown that variance inflation can degrade DA performance when the correlation length scales of observational errors are comparable to, or exceed, those of the background errors—a situation frequently encountered when assimilating wind vectors derived from Atmospheric Motion Vectors. These observations are assimilated, alongside many others, in the ensemble DA system at the European Centre for Medium-Range Weather Forecasts (ECMWF). In this system, each ensemble member undergoes an independent 4D-Var minimisation after perturbing its observations and model parameters to represent observational and model uncertainties. This work presents results from efforts to explicitly account for spatial correlations in observational errors within the ensemble DA framework. In particular, it demonstrates the positive impact of introducing spatially correlated perturbations to assimilated observations on ensemble spread, offering a pathway to improved representation of uncertainty in operational forecasting.
How to cite: Pasmans, I., Holm, E., Bonavita, M., Dance, S., and Bhatt, R.: Reaching far and wide: accounting for spatially correlated observational errors in an ensemble of 4D-Vars system, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22255, https://doi.org/10.5194/egusphere-egu26-22255, 2026.
A numerical weather prediction (NWP) model is a computer program that follows a numerical recipe to discretize the governing equations of atmospheric dynamics for numerical solutions. The center of these governing equations is the Navier-Stokes (NS) equations of fluid motion. The closure paradox theorem (Guermond et al., 2004, J. Math. Fluid Mech.) for the numerically discretized (i.e., filtered) NS equations states that discretization (i.e., filtering) and exact subgrid closure are mutually exclusive in practice. To feasibly solve the discretized NS equations, the subgrid closure must be inexact. All physics parameterization schemes in an NWP model serve as a closure for the discretized governing equations of atmospheric dynamics. Even though these schemes vary in complexity, the closure paradox theorem implies that none of them can be exact if the objective is to efficiently produce useful forecasts at the NWP model's grid resolution. Therefore, empirical adjustment, i.e., constraining its behavior using available observations, is inevitable for any physics parameterization scheme to be feasibly used in an NWP model.
In this presentation, we will use the land-surface parameterization scheme as an example to discuss what the complexity of a physics parameterization scheme actually means, since it cannot be exact. We will argue that the choice of complexity in a scheme is a trade-off between realism and simplicity. We will show that a simple land-surface scheme may meet performance constraints with modest observational requirements and is computationally inexpensive enough to be practically useful. In contrast, a more complex land-surface scheme with sounder physical foundations will yield forecasts that are acceptably more accurate only if enough observations are available to constrain its behavior. When there are insufficient observations to constrain the complex scheme, the simple scheme should be used so that the scheme's behavior can be effectively constrained using available observations.
How to cite: Bao, J.-W., Michelson, S., and Grell, E.: What does the complexity of physics parameterization mean, since no parameterization can be exact?, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22510, https://doi.org/10.5194/egusphere-egu26-22510, 2026.
Tropical cyclone (TC) ensemble forecasting faces a fundamental dilemma. Parameter-based approaches provide accurate track and intensity estimates but lack the continuous spatial fields needed for impact assessment, whereas ensemble-mean approaches offer complete meteorological patterns yet generate unphysical artifacts such as multiple eyes and weakened intensity due to spatial misalignment. Here we present TC-SuperEns, a two-stage framework that resolves this issue through machine learning optimization and physics-based reconstruction. The first stage learns adaptive weights for key TC parameters from historical forecast errors across seven models, while the second stage uses these parameters as dynamical constraints to align ensemble members and reconstruct physically consistent fields. Validation for 2023-2024 Northwest Pacific TCs shows 15-25% improvement in 72-hour track accuracy compared with operational models, along with notable gains in intensity relative to ECMWF's IFS. By unifying discrete parameters and continuous fields into one coherent product, the framework enhances forecast realism and interpretability for effective warning and response.
How to cite: Zhang, X., Chen, X., Liang, Y., Lin, S.-J., Liang, Z., and Song, Q.: High-fidelity tropical cyclone prediction improves public risk communication and disaster mitigation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4651, https://doi.org/10.5194/egusphere-egu26-4651, 2026.
Wintertime heavy precipitation and flooding events along the US West Coast are associated with intense onshore water vapor transport by atmospheric rivers (ARs). Although it has been widely recognized that the uncertainty in AR forecasts is a major contributor to the uncertainty in quantitative precipitation forecast (QPF) along the US West Coast, from the perspective of the atmospheric general circulation, there are multiscale contributors to the QPF uncertainty, depending on the forecast lead time and the forecast model domain size. It remains a major forecasting and risk-management challenge to understand and quantify the multiscale interactions between uncertainties in forecasts of the upper-level jet in the North Pacific and the genesis and evolution of extratropical cyclones, whose AR-induced moisture transport directly contributes to heavy precipitation events.
In this presentation, we leverage an ongoing effort to evaluate NOAA's newly developed AR forecast system to untangle the interactions among the multiscale controllers that contribute to the QPF uncertainty along the US West Coast. We will show using precipitation and atmospheric analysis datasets that the QPF uncertainty along the US West Coast, dependent on the forecast lead time and the model's forecast domain size, can be linked to the uncertainties in the forecasts of convective activities in the tropical Pacific, the interaction between tropical convection and upper-level jet in the North Pacific, and the genesis and evolution of extratropical cyclones associated with the upper-level jet.
How to cite: Michelson, S., Grell, E., and Bao, J.-W.: On the Multiscale contributors to quantitative precipitation forecast uncertainty in the US West Coast, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14986, https://doi.org/10.5194/egusphere-egu26-14986, 2026.
An atmospheric river analysis and forecast system (AR-AFS) is being developed by NOAA’s Environmental Modeling Center to better understand and predict the extreme precipitation events induced by atmospheric rivers (ARs). This system is a limited-area version of NOAA’s Unified Forecast System, with 3-km horizontal resolution. As part of a community effort to optimize the system’s performance, we are currently evaluating the impact of different physics parameterizations on the system’s quantitative precipitation forecast (QPF) along the US West Coast.
To a large degree, the accuracy of the precipitation forecast for a landfalling AR is determined by synoptic-scale, dynamical forcing; however, the parameterized physical processes in the model also play an important role. The factors contributing to errors in QPF are multiscale in nature, and vary in their sensitivity to the model representation of both dynamical and physical processes. Using precipitation observations as well as meteorological analyses, we present an evaluation of the impact of different physics parameterizations on the model performance.
How to cite: Grell, E., Bao, J.-W., Michelson, S., Bengtsson, L., and Swenson, L.: Testing and Evaluation of an Atmospheric River Prediction Model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14992, https://doi.org/10.5194/egusphere-egu26-14992, 2026.
Control simulation experiments (CSEs) aim to steer model predictions toward a desired target by intervening in the model inputs. This concept is mathematically analogous to data assimilation, in which model states are adjusted to align with observations. In recent machine-learning models, automatic differentiation is used to compute gradients of a loss function and to optimize model parameters to minimize the difference between predictions and targets (e.g., reanalysis data). By optimizing input variables (initial conditions) instead of model parameters, CSEs can be formulated for differentiable models. In this study, we conduct CSEs for Typhoon Jebi (2018) using NeuralGCM, a fully differentiable global climate model. Tropospheric temperature perturbations are optimized to minimize the mismatch of 500-hPa geopotential height (Z500) over 130°E–160°E and 10°N–40°N. The results demonstrate that the predicted fields converge toward the target, indicating that CSEs can be successfully implemented and executed in a differentiable GCM framework.
How to cite: Nakano, M., Nasuno, T., and Kodama, C.: Control simulation experiments using a differentiable global climate model: A case study of Typhoon Jebi (2018), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15380, https://doi.org/10.5194/egusphere-egu26-15380, 2026.
The Weather Research and Forecasting (WRF) model has been widely utilized for regional weather and climate prediction. However, variable-resolution models like the Model for Prediction Across Scales (MPAS) offer promising alternatives due to computational advantages, particularly their ability to achieve higher resolution in regions of interest without the nested domains required by WRF. This study evaluates short-term (12-36 hour) surface wind speed predictions from regional MPAS and WRF configurations against 10 m wind observations from 510 meteorological stations across Turkey. Both models employ nearly identical physics parameterizations and use 00 UTC initialization of 0.25° Global Forecast System (GFS) data for lateral boundary conditions. WRF simulations use one-way and two-way nested configurations (12 km and 4 km domains), while MPAS employs variable-resolution meshes with quasi-uniform 4 km resolution at the elliptical core. The high-resolution regions of both models cover approximately the same area, covering all of Türkiye. Model performance is assessed for individual stations using hourly means over two months (January and June 2023) and stratified by three terrain complexity categories (low, medium, high) defined by the Terrain Ruggedness Index (TRI). Results demonstrate that MPAS generally outperforms WRF across all metrics, including mean error (ME), root mean square error (RMSE), and correlation coefficient. Both models overpredict wind speeds, with mean errors of 0.76 m/s (MPAS) and 0.99 m/s (WRF). Overall correlation coefficients are approximately 0.54 for both models. As expected, all metrics deteriorate with increasing terrain complexity. However, MPAS significantly reduces bias, particularly at high-complexity sites, showing 38% lower mean error and 10% lower RMSE compared to WRF's 4 km predictions. Two-way nesting provides limited improvement in WRF's fine-resolution domain, while unexpectedly, the coarse domain (12 km) achieves lower bias and RMSE than the fine domain. These findings suggest that MPAS's unstructured spatial discretization is well-suited for simulating surface winds over complex terrain. Further investigation is needed to understand the physical mechanisms underlying MPAS's superior performance and to assess whether high resolution is necessary over regions with low terrain complexity.
How to cite: Yalcin, R. D., Yilmaz, M. T., and Yucel, I.: Comparative Assessment of Short-Term Near-Surface Wind Prediction of Variable-Resolution MPAS and Nested WRF Models Across Varying Terrain Complexity, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16666, https://doi.org/10.5194/egusphere-egu26-16666, 2026.
Extreme weather events, e.g., droughts, floods, heatwaves, freezes, increasingly challenge physical, financial, and social infrastructure as population and economic growth increase exposure and vulnerability. We propose supplementing conventional disaster risk management strategies with Weather Jiu-Jitsu, an approach that leverages the chaotic dynamics of weather systems to redirect or dissipate destructive trajectories through targeted, low-energy perturbations. Coupled with deep learning models, this framework could serve as a form of nature-assisted global infrastructure to reduce catastrophic climate-extreme impacts in the 21st century. We demonstrate the potential of this strategy through successful perturbation experiments applied to tropical cyclones, atmospheric rivers, freezes, and other high-impact events.
How to cite: Huang, Q., Liu, M., and Lall, U.: Weather Jiu-Jitsu: Prospects for Atmospheric Nudging to Defuse the Impact of Catastrophic Weather Extremes, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22129, https://doi.org/10.5194/egusphere-egu26-22129, 2026.
Aviation emissions contribute to climate change, one of the key contributors being contrail cirrus clouds. The importance of the impact is strongly influenced by the conditions in which they form and evolve.
A condensation trail - or contrail - is composed of ice crystals which form behind the aircraft engine exhaust at high altitudes when local weather conditions are favorable. The formation is also influenced by the engine technology and operating conditions, and by the fuel type. The contrail persists and evolves as long as it remains in an ice supersaturated region - or ISSR-, a local atmospheric air mass characterized by a low temperature and a humidity level that is saturated versus ice. Only persistent contrails are considered as having a potential climate effect.
As part of the SESAR CICONIA project and in order to help the forecasting of ISSRs and hence persistent contrail regions, Météo France has implemented a modification to the cloud scheme of the ARPEGE (Action de Recherche Petite Echelle Grande Echelle) operational numerical weather prediction (NWP) model to enable the representation of ISSRs at cruise altitude. As part of the CICONIA project, Météo France provided Airbus with access to this modified version of ARPEGE to use operationally in forecasting areas where persistent contrails could be formed in flight tests.
During October and November 2025 the temperature and humidity from the modified ARPEGE model was used to forecast areas of potential persistent contrail formation and the test flights were performed in these identified areas. In previous test flights the Global Forecast System (GFS) had been used and was again used for these flight tests as a comparison.
In-flight humidity and temperature measurements are compared to the ARPEGE forecast by interpolating the weather data along the flight trajectories. The observation of persistent contrails is compared to their simulation along the trajectories using the Airbus in house model. These results support the verification and validation of the data from the modified ARPEGE model.
In addition, for a particular day, time and area where persistent contrail coverage was forecast, the in-flight measurements from IAGOS aircraft have also been analysed to confirm where the flights were in ISSRs and if persistent contrails were formed. These results and the associated meteorological parameters were compared to the ARPEGE and GFS forecasts and the ERA5 reanalysis datasets.
The stability of the forecast, which was provided as an hourly forecast for the first 48 hours and then 3 hourly up to 72 hours will be discussed.
The changes to the ARPEGE model to improve the ISSR forecasts, a short description of the studies and analyses of the results for the selected flights will be presented.
How to cite: Mackay, C., Crispel, P., and Arriolabengoa Zazo, S.: The prediction of Ice SuperSaturated Regions and persistent contrail formation using the modified ARPEGE weather forecast and comparison with in-flight measurements and observations., EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7933, https://doi.org/10.5194/egusphere-egu26-7933, 2026.
Extreme precipitation events in India are becoming more frequent and intense, increasing the need for reliable ensemble precipitation forecasts to support early warning systems and disaster preparedness. However, current Numerical Weather Prediction models often underestimate extreme precipitation, and their forecast skill is constrained by errors in initial conditions, numerical approximations, inadequate representation of sub-grid convective processes, and coarse spatial resolution. Additionally, systematic biases in ensemble forecast distributions, such as deviations in central tendency and under- or over-dispersion, further limit the accuracy of probabilistic forecasts. Ensemble Model Output Statistics (EMOS) can reduce some of these limitations by correcting systematic biases in the ensemble mean and spread, and by partly adjusting the predicted overall distribution. Classical EMOS relies on linear transformations, limiting the ability to capture non-linear relationships between the original forecast and the corrected ensemble, and to correct asymmetric distribution errors. Moreover, it derives the corrected distribution at a given location only from the original forecast ensemble for this location. Deep learning-based distributional regression methods, such as U-Net architectures, can generalise classical EMOS by linking the original full spatial field of ensemble forecasts in a complex way to the corrected ensemble forecasts.
This study presents a U-Net based distributional regression (DRU) for daily rainfall forecasts over India, that predicts parametric marginal distributions at each forecast grid cell from the statistics of the original ensemble forecast at all grid cells. It minimises the area mean Continuous Ranked Probability Score (CRPS), while classical EMOS minimises the CRPS individually for each location. DRU is applied to postprocess forecasts with one day lead time for daily precipitation at 12-km resolution from 11 members of the National Centre for Medium Range Weather Forecasting (NCMRWF) global ensemble prediction system for the period 2018-2024. The observations for U-Net training are gridded precipitation data at 0.25° resolution from the Indian Meteorological Department and the NCMRWF forecasts were regridded to this resolution prior to DRU training. Over most of India, DRU improves local precipitation distributions, including for higher quantiles, and corrects under- or overdispersion. The forecast skill in terms of Continuous Ranked Probability Skill Score increases over large areas, particularly northern and western India, while in central and northeastern regions, there are locations where the skill decreases. For some of these, the marginal distributions are also not improved. Additionally, DRU improves the reliability for predicting the exceedance probability of various precipitation thresholds. Future endeavours will focus on optimizing DRUs for postprocessing heavy precipitation events and evaluating the forecast skill using the Brier Score.
How to cite: Thakur, C., Widmann, M., Ashrit, R., Orr, A., C. Leckebusch, G., and Geen, R.: Improving precipitation ensemble forecasts over India using a convolutional distributional regression framework, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9569, https://doi.org/10.5194/egusphere-egu26-9569, 2026.
Since 1972, the US Weather Modification Reporting Act has required federal reporting of any cloud seeding activities taking place in the United States to NOAA. Leveraging these historical records, we used OpenAI’s o3 large language model to extract information about the project name, year, season, state, operator, seeding agent, apparatus used for deployment, stated purpose, target area, control area, start date and end data of all publicly reported cloud seeding activities (Donohue and Lamb, 2025). This method was validated through the performance of the data extraction pipeline on a manually labeled subset of 200 records, achieving an average accuracy of 98.38% across all fields. This structured data set, encompassing 832 distinct operational periods from 2000 – 2025, represents the first large-scale, publicly available data set of cloud seeding activities for the US that can facilitate large-scale historical analysis.
Using this data set, we use synthetic control, a statistical method to estimate the effect of an intervention, to understand whether cloud seeding is an effective strategy for augmenting snowpack and precipitation from a climatological perspective. Using the reported locations and times of historical cloud seeding operations, along with historical datasets of precipitation and snowpack, we analyze whether cloud seeding can have a climatologically significant impact in augmenting precipitation and snowpack regionally. Our approach makes it possible to rigorously evaluate cloud seeding effectiveness on a climatological scale across the US for the first time.
How to cite: Lamb, K. and Donohue, J.: Using Synthetic Control to Assess the Climatological Significance of Cloud Seeding in the Western US with a Structured Data Set of Reported Activities from 2000 – 2025, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15743, https://doi.org/10.5194/egusphere-egu26-15743, 2026.
This study explores how different integration strategies for infrared (IR) and microwave (MW) brightness temperatures (TBs) impact precipitation forecasting within a 3D-Var all-sky radiance data assimilation (DA) framework. To improve heavy rainfall forecasts over the Korean Peninsula, we propose an asynchronous assimilation strategy. In this approach, when IR and MW observations overlap, we prioritize MW TBs and intentionally exclude IR TBs to minimize potential redundancies and physical inconsistencies in cloud representation. We compared this proposed method against two common synchronous strategies: one assimilating clear-sky IR with all-sky MW, and another integrating both IR and MW under all-sky conditions. Using a heavy precipitation case over Korea as a testbed, we assimilated AHI (Himawari-8), GMI (GPM), and AMSR2 (GCOM-W) data to evaluate their impacts on hydrometeor analysis and subsequent forecast accuracy. Our results indicate that the asynchronous strategy leads to a more balanced vertical distribution of solid and liquid hydrometeors, resulting in the most reliable precipitation forecasts. In contrast, the all-sky IR+MW strategy tended to overemphasize upper-level clouds while reducing low-level moisture, leading to biased localized rainfall. Meanwhile, the clear-sky IR+MW approach failed to adequately capture upper-level stratiform structures. These findings suggest that an optimized, sensor-specific integration strategy is essential for maximizing the benefits of multi-platform satellite data assimilation.
How to cite: Hwang, J. and Cha, D.-H.: Optimizing Synergistic Strategies of IR and MW Radiance in All-sky Data Assimilation for Heavy Precipitation Forecasting, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16528, https://doi.org/10.5194/egusphere-egu26-16528, 2026.
Accurate prediction of tropical cyclogenesis is fundamental to enhancing typhoon forecasting and disaster mitigation. This study investigates the impacts of Global Navigation Satellite System (GNSS) observations on cyclogenesis predictions by integrating a multi-case statistical evaluation with a targeted case study. Initially, the impact of GNSS Radio Occultation (RO) data assimilation (DA) was assessed by assimilating both conventional observations and RO data in ten tropical cyclone cases in the Northwestern Pacific from 2020 to 2022. Utilizing the WRF hybrid-3DEnVar system, the results demonstrate that incorporating RO data with a nonlocal excess phase operator improves the accuracy of cyclogenesis localization and timing, a finding further corroborated by ensemble forecasts.
Beyond RO, this research explores the potential of integrating GNSS Reflectometry (GNSS-R) data, which provides sea surface wind speed information to better capture the near-surface dynamical environment during the early stages of cyclogenesis. In a case study of Typhoon Gaemi (2024), RO data assimilation successfully captured the cyclogenesis signal that was missed in experiments without RO. Furthermore, jointly assimilating RO and GNSS-R observations (RO+R) refined the predicted genesis timing compared with RO-based experiments. These findings suggest the potential to use multi-source GNSS observations to improve the precision of tropical cyclogenesis forecasting.
How to cite: Chen, S.-Y., Pham Xuan, Q., Huang, C.-Y., Kuo, Y.-H., and Yang, S.-C.: Impact of GNSS Radio Occultation and Reflectometry Data Assimilation on Tropical Cyclogenesis Prediction, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15881, https://doi.org/10.5194/egusphere-egu26-15881, 2026.
Posters virtual: Mon, 4 May, 14:00–18:00 | vPoster spot 5
EGU26-8881 | ECS | Posters virtual | VPS2
Effect of decadal land use change on WRF model-simulated surface meteorological parameters over the Indian regionMon, 04 May, 14:00–14:03 (CEST) vPoster spot 5
Variations in land surface characteristics directly alter surface biophysical properties like albedo, roughness length, and canopy resistance, leading to changes in surface radiative and turbulent fluxes. These changes influence sensible and latent heat fluxes, which can further regulate surface temperature, evapotranspiration, and near-surface moisture transport. Thus, variations in surface fluxes associated with changes in land-surface properties can regulate convective instability, moisture convergence, and can modulate short-range rainfall characteristics and their predictability. Such land–atmosphere feedbacks are particularly important over the Indian region, where strong seasonal contrasts and heterogeneous land surfaces play a critical role in shaping rainfall variability. Hence, this study investigates the sensitivity of short-range precipitation forecasts over India to decadal changes in land use and vegetation during the pre-monsoon and monsoon periods. USGS 24-category land use data based on the 1994 landscape is used as the control run. Seven different simulation experiments are conducted using WRF model with various land use and vegetation data from MODIS and ISRO; ie: Experiment1 (MODIS 2001, USGS LAI and VF default), Experiment 2 (MODIS 2001, LAI and VF default), Experiment 3 (MODIS 2019, LAI and VF default), Experiment 4 (MODIS 2001, with Urban Class of 2019, LAI, and VF default), Experiment 5 (MODIS 2001, with water bodies of 2019, LAI, and VF default), Experiment 6 (MODIS 2019, LAI and VF of 2019), Experiment 7 (ISRO 2018-2019, VF and LAI default). A comprehensive assessment based on quantitative error metrics and categorical forecast skill scores demonstrates statistically significant improvements in rainfall forecast performance following the assimilation of updated land-use and vegetation datasets. These statistically robust improvements indicate that realistic representation of land-surface conditions contributes meaningfully to enhanced short-range precipitation predictability. The computed Extreme Dependency Index (EDI) values indicate an enhanced ability of the model to capture rare extreme rainfall events following the incorporation of updated land-use information. The incorporation of realistic land-use classifications derived from MODIS and ISRO datasets led to improved simulations of surface meteorological variables, including temperature, wind speed, relative humidity, surface pressure, and surface fluxes. Corresponding improvements were also observed in the vertical atmospheric profiles for wind, temperature, and specific humidity profiles. These enhancements indicate a more realistic depiction of land–atmosphere interactions and boundary-layer processes in the model.
How to cite: Putatunda, I., Vasudevan, R., and Singh, R.: Effect of decadal land use change on WRF model-simulated surface meteorological parameters over the Indian region, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8881, https://doi.org/10.5194/egusphere-egu26-8881, 2026.
EGU26-15265 | ECS | Posters virtual | VPS2
Is ERA5 Fit for Purpose? A Global Multi-Variable Evaluation of Reanalysis Strengths and WeaknessesMon, 04 May, 14:03–14:06 (CEST) vPoster spot 5
Reanalysis products, which blend weather model output with observations are commonly used as substitutes for observations to assess numerical weather prediction model forecast skill, the accuracy of climate model historical realizations, and AI training. However, the quality of reanalysis output is not uniform across all variables, times of day, seasons, or geographic settings. This study evaluates the strengths and weaknesses of ERA5 reanalysis (0.25° grid) over a multi-year period by comparing them to worldwide hourly surface observations from over 1200 stations, buoys, and radiosonde vertical profiles.
Our analysis focuses on several metrics across the diurnal cycle (7 AM and 3 PM local time) and during temperature outlier events (< 10th percentile and > 90th percentiles for the 30-year climatology). Results indicate that ERA5 provides reliable 2-meter air temperatures in most regions, but shows a frequent dry bias in dewpoint of greater than 3 ℃ more than 5% of the time for many stations in the Dry and Mediterranean climate zones. ERA5 often underestimates warm events (> 90th percentile), with the largest cold biases, less than -3 ℃ occurring more than 11% of the time in the Mediterranean climate zone. Temperature and dewpoint biases are amplified in complex terrain, and dewpoint biases tend to be larger near coastal locations. To investigate whether higher spatial resolution mitigates these issues, we also examine the performance of the ERA5-Land product (0.1° grid). These findings emphasize the importance of evaluating the adequacy of purpose when using reanalysis for specific applications, since performance can vary significantly by variable, time of day, season, and climate zone.
How to cite: Lewis, W., Yuter, S., and Miller, M.: Is ERA5 Fit for Purpose? A Global Multi-Variable Evaluation of Reanalysis Strengths and Weaknesses, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15265, https://doi.org/10.5194/egusphere-egu26-15265, 2026.
