We welcome contributions addressing both traditional and machine learning–based approaches to data assimilation, ensemble generation, and ensemble utilization. Of particular interest are studies that explore how these techniques evolve with forecast lead time, including the transition from short-range to medium-range, extended-range, and seasonal forecasting systems. Contributions highlighting the perspective of operational forecasters and the practical use of ensembles in forecasting and decision-making, especially for extreme events, are strongly encouraged.
The conveners invite papers on a broad range of topics related to Data Assimilation and Ensemble Forecasting for weather and climate prediction, including (but not limited to):
- intercomparison and assessment of the complementarity between different data assimilation techniques, such as Kalman filtering, variational methods, hybrid approaches, and nudging techniques for frequent or rapid update analysis cycles;
- variational data assimilation with extended assimilation windows, including weak-constraint formulations that allow for the explicit representation of model error;
- ensemble data assimilation systems and flow-dependent estimation of background, observation, and model error statistics;
- representation of uncertainties in initial conditions, model physics, boundary conditions, and coupling strategies in global and limited-area ensemble prediction systems, across time scales from short-range to seasonal;
- strategies for bridging weather and climate prediction, including ensemble methods tailored for subseasonal to seasonal forecasting;
- verification, calibration, and post-processing methods for ensemble prediction systems, with attention to scale-dependent performance;
- use of multi-model and ensemble databases;
- applications of ensemble forecasts across different sectors, including energy, health, transport, agriculture, insurance, finance, and climate-sensitive decision support.
S., arXiv, (2025).
How to cite: Grenzi, M., Gastaldo, T., Poli, V., Marsigli, C., Janjic, T., and Carrassi, A.: Improving satellite radiance assimilation at convective-scale: from operational numerical weather predictions towards machine learning limited area models, EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-447, https://doi.org/10.5194/ems2026-447, 2026.
To improve the forecast quality of numerical weather prediction (NWP), the German Meteorological Service (Deutscher Wetterdienst, DWD) has initiated a project aimed at assessing data quality and assimilation of profile observations from ground-based remote sensing instruments that have not yet been exploited operationally.
The objective of this initiative is to fill the observational gap in the atmospheric boundary layer, especially with respect to short time scales, by providing continuous, high-temporal-resolution profiles of thermodynamic variables, wind, and cloud properties. These observations are expected to be especially beneficial for weather forecasting applications. The DWD is evaluating various remote sensing systems for their ability to provide continuous operational feasibility and impact on NWP.
In this contribution, we present results of the assimilation of two ground-based remote sensing instruments into the kilometre-scale ensemble data assimilation system (KENDA): water vapour mixing ratio profiles from a Differential Absorption Lidar (DIAL) and radar reflectivity profiles from a cloud radar. For the integration of the DIAL observations into the data assimilation code environment, only small adjustments were necessary. In contrast, the cloud radar data required an adaptation of the complex forward operator EMVORADO (Efficient Modular Volume scan Radar Operator), which was originally developed and previously used only for precipitation radars.
In an initial step, single observation data assimilation experiments and the corresponding observation minus first guess statistics showed promising results. To assess the impact in an operational setting, we performed dedicated data assimilation experiments with and without these additional observations. We considered both summer and winter periods, as well as different observation error specifications for the DIAL measurements. Based on the successful data assimilation cycling experiments, we conducted first forecast experiments, including DIAL water vapour mixing ratio observations. The results indicate a positive impact on humidity and temperature forecasts. We are currently investigating the impact of cloud radar reflectivity data in such experiments. Preliminary results show a neutral to slightly positive impact on the humidity first guess.
Our findings suggest that ground-based remote sensing data can provide valuable additional information for convective-scale data assimilation and justify more extensive impact studies in the context of NWP.
How to cite: Pruschke, J., Schomburg, A., Mendrok, J., Stephan, K., Görsdorf, U., Löffler, M., Knist, C., and Schraff, C.: Data Assimilation of Differential Absorption Lidar Data and Cloud Radar Data, EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-592, https://doi.org/10.5194/ems2026-592, 2026.
Extreme weather and climate events impose substantial societal and economic costs under continued global warming, but their prediction remains a challenge in meteorology and other geosciences. Their emergence and development are results of nonlinear multivariate interactions within the earth system at a wide range of spatial and temporal scales. A favorable initial state is essential for triggering the evolution of extreme weather and climate events, except for large‐scale drivers, positive feedbacks and stochastic processes. Thus, accurate initialization of coupled systems poses a fundamental challenge in extreme event prediction. Several operational weather forecasting centers have successfully established their own covariance-based data assimilation (DA) systems to address this challenge. However, the conventional assimilation approaches, such as the ensemble Kalman filter (EnKF), tend to underestimate extreme events due to their inability to capture these nonlinear coupling features, given their reliance on linear background error covariance estimation. Thus, in this study, we aim to consider the complex coupling features in state estimation by leveraging the capabilities of machine learning (ML) algorithms in nonlinear representation. The novel ML-based assimilation method effectively and nonlinearly projects the observational information to the prior predictions, generating reliable analysis for extreme phenomena. This data driven approach effectively characterizes the time-variant and complex multivariate relationships, thereby nonlinearly projecting the innovation onto the ensemble subspace. This significant improvement enables the ML-based approach to increase the analysis accuracy for extreme phenomena by up to 66% over EnKF, and its ensemble increment distribution is well aligned with that of the target increments, showing the potential of data driven assimilation approach for advancing the capabilities of capturing and triggering the extreme events.
How to cite: Zheng, F.: Improving the Assimilation Ability for the Extreme Eventsby Proposing a Nonlinear Machine Learning DataAssimilation Approach, EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-64, https://doi.org/10.5194/ems2026-64, 2026.
This study examines the impact of radar data assimilation with Large-Scale Analysis Constraints (LSAC) under different monsoon synoptic conditions, with special focus on vertical cloud structure and associated convective processes. The role of LSAC in reducing convective scale imbalances is also evaluated. Two extreme monsoon rainfall events during August 2018 and 2019 were considered to represent contrasting synoptic environments, with the 2019 event characterized by strong localized convection. In this study, C-band radar reflectivity is assimilated indirectly, while radial wind is assimilated directly, and their combined impact on the forecast of these extreme events is assessed. The experiments are carried out with and without the application of Large-Scale Analysis Constraints, referred to as LSAC and noLSAC, respectively. It is hypothesized that the imbalance between large-scale and convective-scale processes introduced during high resolution radar assimilation can be reduced by incorporating LSAC. The results show clear improvement in minimizing convective-scale imbalances, particularly for the August 2019 event, where strong localized convection was present. Rainfall verification indicates that inclusion of LSAC improves the location, spatial pattern, and amount of precipitation in both cases. Cloud top heights are also better represented in LSAC experiments, whereas high resolution radar assimilation without LSAC leads to spurious cloud top development. Analysis of hydrometeor profiles shows that radar assimilation with LSAC consistently reduces the overestimation of hydrometeor condensate in the entire vertical column for 2018 event. For 2019, overestimated hydrometeor condensates above the melting layer in minimized. This is supported by radar reflectivity comparisons, which indicate that unrealistically strong reflectivity signatures above the melting layer are minimized in LSAC experiments for 2019 event. More organized and consolidated convective echoes are also evident with LSAC. Thermodynamic analysis further supports these findings, showing that excessive convective potential generated during radar assimilation above the melting layer is moderated with large-scale constraints in the presence of strong and localized convective environment. This concludes that radar data assimilation with large-scale constraints behaves differently for the above mentioned monsoon extreme events depending on atmospheric characteristics. In general, incorporation of LSAC reduces biases in hydrometeor formation and improves cloud structure and precipitation forecasts.
How to cite: Babu, F. and Kutty, G.: Impact of Large-Scale Analysis Constraints on C-Band Radar Data Assimilation during the Simulation of Two Contrasting Monsoon Extreme Events over Southern Peninsular India , EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-183, https://doi.org/10.5194/ems2026-183, 2026.
Land-atmosphere interactions play a key role in the predictive skill of numerical weather prediction (NWP) models. The skin temperature is an important variable in this context, being at the interface between the land surface and the atmosphere, and modulating exchanges of heat, moisture, and momentum. Despite the abundance of high-resolution satellite data, assimilating infrared radiances over land has historically been challenging due to complex, heterogeneous surface emissivity and the highly dynamic diurnal variations of skin temperature.
This presentation highlights recent developments within the coupled data assimilation (DA) system at the European Centre for Medium-Range Weather Forecasts (ECMWF). For the first time, window channel (10.8 μm) infrared brightness temperatures from the Spinning Enhanced Visible and InfraRed Imager (SEVIRI) are activated in the four-dimensional variational analysis (4D-Var) system over land. To simulate the corresponding brightness temperatures in observation space, the forward operator - the RTTOV radiative transfer model – requires accurate representations of land surface emissivity, provided by the CAMEL Broadband Emissivity Climatology Version 3, and the land surface skin temperature. Up until now, this skin temperature acted as a sink variable in the 4D-Var.
Our new approach utilises an extended control vector, known as the TOVSCV, to explicitly assimilate the skin temperature in the Land Data Assimilation System (LDAS). Within each 12-hour DA window of the coupled land-atmosphere system, the analysed skin temperature from 4D-Var is included in the LDAS simplified extended Kalman filter (SEKF) observation vector as a pseudo-observation to constrain the soil temperature analysis.
This coupled DA approach bridges the gap between atmospheric radiance assimilation and the estimation of the land surface state. We will discuss the technical implementation of the TOVSCV extended control vector, the updated SEKF formulation, and the impact of this coupled methodology on the Earth system analysis. We will present results comparing the performance of these recent developments with a control experiment.
The work in this study is performed as part of the Data Assimilation and Numerical Testing for Copernicus Expansion Missions (DANTEX) project in collaboration with the European Space Agency (ESA).
How to cite: Heyvaert, Z., de Rosnay, P., Benedetti, A., English, S., Herbert, C., Villeneuve, E., Weston, P., and Catapano, F.: Advancing coupled land-atmosphere data assimilation through the integration of thermal infrared observations, EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-374, https://doi.org/10.5194/ems2026-374, 2026.
The multiscale interactions of initial condition (IC) perturbations for 12-h convection-permitting ensemble forecasting were explored in this study. The heavy-rainfall events occurring in South China in the rainy season during 2013–2020 were focused on and classified into three types: the weak-forcing, strong-forcing, and tropical cyclone (TC) cases. The impacts of both large- and small-scale IC perturbations on the multiscale characteristics of forecast perturbations and the forecast performance were investigated. Both the upscale growth of small-scale IC perturbations and the downscale propagation of large-scale IC perturbations favored perturbation growth, with damping impacts of both processes beyond the first few hours. Compared with upscale growth, downscale propagation showed greater impacts on forecast perturbations in both magnitude and location and more evident variable-dependent impacts. The small-scale IC perturbations spread the high-probability rainfall downstream, while the large-scale IC perturbations spread the high-probability rainfall in all directions to the low-probability areas. Both of these two types of rainfall spreading led to increasing locational perturbations. Probabilistic forecasts of heavy short-duration rainfall benefited from both the small- and large-scale IC perturbations, leading to performance improvements of the multiscale IC perturbations over the single-scale IC perturbations. In particular, the small- and large-scale IC perturbations more benefited the forecasts of weak-forcing heavy rainfall with larger spatial forecast errors and at earlier lead times, respectively. The flow regimes with large forecast uncertainties, such as the coastal warm-sector rainband of strong-forcing cases, which were determined by the complicated interactions between synoptic-scale forcing and moist convection, directly boosted both upscale growth and downscale propagation.
How to cite: Zhang, X.: Multiscale interactions of Initial Condition Perturbations for Convection-Permitting Ensemble Forecasting over South China during the Rainy Season, EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-229, https://doi.org/10.5194/ems2026-229, 2026.
Subseasonal-to-seasonal (S2S) ensemble forecasts are critical for sectoral risk management, yet they are frequently constrained by coarse spatial resolutions and systematic biases. While post-processing is essential, conventional statistical methods (such as Bias Correction Spatial Disaggregation, BCSD) and deterministic deep learning approaches often fail to preserve physical consistency or inadvertently collapse the ensemble dispersion essential for quantifying uncertainty. This suppression limits their utility in representing high-impact extreme weather events.
To address these challenges within the context of next-generation machine learning post-processing, we present MSSDiff, a member-wise super-resolution diffusion framework designed to bridge the gap between coarse global ensemble outputs and local impacts. Instead of injecting stochastic noise that disrupts ensemble consistency, MSSDiff processes individual ensemble members using a deterministic probability-flow ODE sampling strategy. This approach simultaneously corrects biases and enhances resolution without disrupting the inherent spatiotemporal coherence or aleatoric uncertainty of the original dynamical ensemble.
The architecture features a Wavelet-Coupled Upsampling Module (WCUM) to explicitly recover high-frequency textures, such as precipitation extremes, and a Latent Space Attention (LSA) mechanism to capture large-scale teleconnections vital at the extended range. By jointly modeling temperature and precipitation, the framework leverages thermodynamic constraints to further improve physical reliability.
Validated on a newly constructed multi-model S2S benchmark (S2S-SR), MSSDiff is compared extensively against both traditional statistical methods and cutting-edge deterministic AI models. MSSDiff significantly improves the Anomaly Correlation Coefficient (ACC) for precipitation by over 19% compared to operational BCSD baselines. Crucially for ensemble forecasting, MSSDiff achieves the lowest Continuous Ranked Probability Score (CRPS) and produces quasi-uniform rank histograms. This demonstrates superior probabilistic calibration and the successful preservation of a physically plausible ensemble spread, offering a robust machine learning alternative to traditional calibration methods for extended-range extreme event forecasting.
How to cite: Meng, F., Li, W., Li, Y., and Wu, J.: Bridging the Gap in Extended-Range Ensemble Forecasting: A Machine Learning Diffusion Framework for High-Resolution Downscaling and Probabilistic Calibration , EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-19, https://doi.org/10.5194/ems2026-19, 2026.
Tropical cyclones (TCs) often bring destructive winds, heavy rainfall, high waves, and storm surges, causing severe losses. Accurate TC forecasting is therefore vital for disaster mitigation. However, the chaotic nature of TCs means that small initial errors can rapidly grow, leading to large forecast uncertainties. To address this, ensemble forecasting is employed to generate multiple scenarios, quantify uncertainty, and enhance forecast reliability for disaster prevention. Ensemble forecasting has greatly improved TC predictions at major centers like ECWMF and NECP, etc. , but it requires substantial computational resources due to complex physics-based models. This study would address this challenge by developing an AI-driven optimized ensemble forecast system using Orthogonal Conditional Nonlinear Optimal Perturbations (O-CNOPs). The system bridges the gap between computational efficiency and dynamic consistency in TC forecasting. Unlike conventional ensembles limited by computational costs or AI ensembles constrained by inadequate perturbation methods, O-CNOPs generate dynamically optimized perturbations that capture fast-growing errors of FuXi model while maintaining plausibility. The key innovation lies in producing orthogonal perturbations that respect FuXi’s nonlinear dynamics, yielding structures reflecting dominant dynamical controls and physically interpretable probabilistic forecasts. The study demonstrates generally-superior deterministic and probabilistic skills over the operational Integrated Forecasting System Ensemble Prediction System, establishing a new paradigm combining AI’s computational advantages with rigorous dynamical constraints. Success in TC track forecasting paves the way for reliable ensemble forecasts of other high-impact weather systems, marking a major step toward operational AI-based ensemble forecasting. It is expected that such AI-driven optimized ensemble forecast system can be effectively applied to operational forecasts.
How to cite: Duan, W.: A Synergistic Approach: Dynamics-AI Ensemble in Tropical Cyclone Forecasting , EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-6, https://doi.org/10.5194/ems2026-6, 2026.
Climate variability is a key factor influencing tourism activity, as weather and climatic conditions largely determine the attractiveness of a destination, the level of comfort experienced by visitors, and the feasibility of outdoor activities. Over recent decades, the relationship between climate and tourism has emerged as a rapidly developing field of research, leading to the development of various approaches and indices aimed at quantifying the impact of climatic conditions on tourism. At the same time, advances in seasonal forecasting have created new opportunities for predicting meteorological and climatic conditions over lead times of several months. Despite this progress, the application of such forecasts in the tourism sector remains relatively limited.
The present study aims to explore the potential of seasonal forecasts for assessing climate suitability for tourism activities, with a particular focus on evaluating the predictive skill of numerical forecasting models in estimating tourism-related climate indices on a seasonal timescale. To this end, hindcast simulations produced by the Weather Research and Forecasting (WRF) model are utilized to derive key climatic variables. Based on these variables, a range of climate-based tourism suitability indices are calculated and analyzed, including widely used metrics such as the Tourism Climate Index (TCI), the Holiday Climate Index (HCI), and the Physiological Equivalent Temperature (PET), as well as indices tailored to specific types of tourism, such as coastal, cultural, and winter tourism.
The analysis focuses on assessing the ability of seasonal forecasts to reproduce the variability of these indices and to provide reliable predictions over lead times of three to six months. By evaluating the performance and limitations of these forecasting tools, the study seeks to determine the extent to which seasonal climate information can support decision-making in the tourism sector. Ultimately, the findings are expected to contribute to a deeper understanding of the relationship between climate variability and tourism comfort, while highlighting the potential of seasonal forecasting as a valuable tool for tourism planning and management.
Acknowledgements: This research was supported by the PREVENT project that has received funding from the EU Horizon Europe framework programme (grant no. 101081276)
How to cite: Velikou, K. and Anagnostopoulou, C.: From Climate Variability to Tourism Comfort: The Role of Seasonal Forecasting, EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-574, https://doi.org/10.5194/ems2026-574, 2026.
This study assesses the performance of the third generation operational climate prediction system developed by the China Meteorological Administration (CMA-CPSv3) in predicting the Asian summer monsoon on seasonal time scales. The evaluation is carried out using a 20-year set of ensemble hindcast data, which provides a solid foundation for comprehensively examining the model’s predictive capability and reliability. Results from the assessment demonstrate that CMA-CPSv3 has higher predictive skill for key components of the Asian summer monsoon system, covering a wide range of crucial climatic elements. Specifically, the model performs well in predicting the location of the summer rain belt, maximum rainfall intensity and distribution, large-scale atmospheric circulation patterns, the monsoon onset progression, as well as the interannual variability of dynamic summer monsoon indices that reflect the intensity and fluctuation of the monsoon system.
Notably, the model can realistically capture the interannual variability of the western North Pacific subtropical high, a pivotal atmospheric circulation system closely associated with the position and movement of the summer rain belt over eastern China. Accurate representation of this variability lays a solid foundation for improving regional rainfall predictions. When compared with its previous version, CMA-CPSv2, the upgraded CMA-CPSv3 exhibits substantial and widespread improvements in summer precipitation prediction across the Asian continent, with particularly remarkable enhancements over eastern China, a region deeply affected by the Asian summer monsoon. Further analysis suggests that these improvements are mainly attributed to the optimized simulations of sea surface temperatures in the tropical Pacific Ocean and Indian Ocean, as well as the strengthened and more realistic ocean–atmosphere coupling processes linked to these tropical sea areas. The refined air-sea interactions enable the model to better depict the remote impacts of tropical oceans on the Asian summer monsoon system, thus elevating the overall accuracy and stability of seasonal climate predictions.
How to cite: Liang, X., Li, Q., and Wu, T.: Dynamical Seasonal Prediction of the Asian Summer Monsoon in CMA-CPSv3, EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-281, https://doi.org/10.5194/ems2026-281, 2026.
In the context of climate change and the increasing frequency of unprecedented weather and climate events, seasonal forecasts are becoming an essential tool for adaptation, supporting decision-making and preparedness across strategic sectors and society at large. However, in mid-latitudes, seasonal prediction systems still exhibit limited skill and typically provide climate information averaged over the entire season, which may not offer the resolution and specificity required for some practical applications. This limitation is particularly relevant for extreme events, such as heavy precipitation, droughts, or heat waves, which can have substantial impacts on agriculture, hydrology, health, and energy sectors.
Post-processing techniques offer a promising pathway to enhance the relevance and applicability of seasonal forecasts. This work presents a set of post-processing approaches currently being developed and tested at AEMET, aimed at generating climate information better suited for decision-making. Statistical downscaling methods are applied to improve forecast skill by relating large-scale model outputs to observed local variations, including the effects of orography and local-scale climate features. Additionally, the availability of refined, higher-resolution information enables the development and evaluation of impact-based indices, including those focused on extreme precipitation events, droughts, and heat-stress indicators, tailored to support adaptation strategies across multiple sectors.
This work assesses the extent to which these post-processing strategies enhance the capability of seasonal forecast systems to provide skillful, reliable, and decision-relevant information. Special attention is given to the added value in terms of spatial detail, forecast reliability, and representation of climate-related risks, including high-impact events. The results highlight both the potential and limitations of current approaches, contributing to ongoing efforts to bridge the gap between seasonal climate prediction and its practical use in climate services.
How to cite: Hernández Calleja, M., Domínguez Alonso, M., Sanfiz, S., González Alemán, J. J., and Rodríguez Guisado, E.: Enhancing the Value of Seasonal Forecasts through Post-Processing at AEMET: From Seasonal Systems to User-Oriented Information, EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-475, https://doi.org/10.5194/ems2026-475, 2026.
Predicting extreme climate events at seasonal timescales remains a significant scientific and operational challenge, particularly in subtropical regions, where variability is strongly influenced by large-scale circulation patterns and where impacts on water resources, ecosystems, and human health are especially critical. These regions are often characterized by pronounced climate variability and exposure to high-impact events, making the reliable prediction of extremes a key priority. While dynamical forecasting systems generally exhibit skill in representing large-scale modes of variability, they often struggle to accurately translate these signals into consistent and reliable regional-scale responses. This mismatch, combined with a low signal-to-noise ratio, limits the direct usability of raw model outputs for decision-making.
This work presents an exploratory assessment of post-processing strategies aimed at enhancing the representation and predictability of temperature and precipitation extremes in subtropical areas. Using ensemble outputs from dynamical seasonal prediction systems, we investigate the extent to which statistical post-processing can improve the characterization of extremes, defined using percentile-based indices. The analysis adopts as a flexible and data-driven framework, focusing on identifying relationships between large-scale predictors and regional-scale extreme responses, as well as on improving the consistency and robustness of probabilistic forecasts.
Preliminary results suggest that even relatively simple post-processing approaches can add value to the prediction of seasonal extremes, particularly by better exploiting the large-scale information already captured by dynamical models. These findings highlight the potential of post-processing as a complementary tool for improving forecast usability and supporting climate services and risk-informed decision-making in vulnerable subtropical regions.
How to cite: González-Alemán, J. J., Quintana-Doña, C., Fernández-Jerez, A., Hernández-Calleja, M., and Rodríguez-Guisado, E.: Exploring Post-Processing for Seasonal Forecasts of Extremes in the Subtropics, EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-612, https://doi.org/10.5194/ems2026-612, 2026.
How to cite: Ward, L., O'Loughlin, F., and Sweeney, C.: Evaluation of ECMWF subseasonal-to-seasonal forecast skill over Ireland for flood and drought events, EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-397, https://doi.org/10.5194/ems2026-397, 2026.
The standard strong-constraint four-dimensional variational data assimilation (SC-4DVar) approach was designed to correct random, zero-mean errors in both model forecasts and observations in the China Meteorological Administration Global Forecast System (CMA-GFS). However, global numerical weather prediction (NWP) models often exhibit significant systematic biases within the assimilation windows. To address this limitation, a weak-constraint 4DVar (WC-4DVar) approach, which explicitly accounts for model biases, has been proposed as a method for reducing the impact of these biases. In this study, the WC-4DVar approach is integrated into the CMA-GFS, and a one-month cycling assimilation and forecasting experiment is conducted to evaluate its effectiveness. The WC-4DVar approach incorporates a model bias term into the cost function, with the control variables and covariance matrix estimated from 240 samples collected over a one-year period from the CMA ensemble data assimilation (EDA) trial. The results show that WC-4DVar significantly reduces the model biases, particularly in regions with larger biases. Specifically, the analysis fields are improved, as evidenced by the negative root-mean-squared error reduction ratios obtained for the temperature and wind components. WC-4DVar exhibits an enhanced ability to forecast the geopotential height, temperature, and wind fields in the northern hemisphere within the first 192 hours. In the southern hemisphere, significant enhancements are observed in the upper atmosphere during the first 96 hours. Additionally, WC-4DVar improves its forecasting ability at the top of the model in tropical regions. These findings highlight the potential of WC-4DVar to serve as a valuable tool for improving operational NWP systems, particularly in regions where systematic biases are most pronounced.
How to cite: Wang, L. and Liu, Y.: Development and impact of weak-constraint 4DVar with model bias consideration in the CMA-GFS , EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-182, https://doi.org/10.5194/ems2026-182, 2026.
Accurate forecasting of extreme precipitation using mesoscale models remains a major challenge, particularly for warm-sector heavy rainfall events driven by complex cloud-microphysical processes that are poorly constrained by conventional observations. Satellite microwave radiances offer critical thermodynamic and hydrometeor information, yet their assimilation under cloudy and precipitating conditions requires careful treatment of radiative transfer errors. This study develops a physically constrained assimilation method for all-sky radiances from microwave temperature and humidity sounders (MWTS and MWHS) aboard the new-generation Fengyun-3 (FY-3) satellites, integrated into the China Meteorological Administration (CMA) MESO system. Within the CMA MESO 3DVAR system, the Advanced Radiative Transfer Modeling System (ARMS) is employed as the fast satellite observation operator, incorporating three key components: (1) a dynamically adaptive land emissivity parameterization, (2) a microphysics-consistent adaptive formulation for hydrometeor effective radius, and (3) a new delta-M multiple-scattering scheme. Three experiments were conducted for a record-breaking rainfall event over Hunan Province, China: a control run without all-sky radiance assimilation (CTRL), an all-sky configuration (EXPR1) utilizing the baseline ARMS, and an enhanced configuration (EXPR2) with an improved ARMS scattering module. Results show that EXPR2 reduces systematic biases in moisture-sensitive channels by approximately 7 K and decreases random errors by nearly 50% relative to CTRL, indicating substantially improved observation-minus-background statistics in cloudy scenes. This improvement further enables more realistic storm predictions, accurately reproducing the observed quasi-stationary convective core (> 45 dBZ) and the maximum rainfall exceeding 400 mm. Spatial verification using the Fractional Skill Score (FSS) demonstrates robust and statistically meaningful skill gains across multiple precipitation thresholds and spatial scales. At the extreme 100 mm threshold, EXPR2 is the only configuration that achieves useful skill at 50–100 km scales, whereas CTRL exhibits negligible skill. These findings underscore the value of physically constrained all-sky operator design for enhancing the prediction of high-impact precipitation events, with implications for operational convection-allowing data assimilation systems.
How to cite: chen, S., xie, Y., and weng, F.: Physically-Constrained Assimilation of All-Sky Radiances from FY-3 Microwave Sounders for Extreme Precipitation Forecast, EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-227, https://doi.org/10.5194/ems2026-227, 2026.
Seasonal climate forecasts are increasingly integrated into climate services and risk-based decision-making across sectors such as energy, agriculture, and water management. Their value lies not only in predicting mean conditions but in providing probabilistic information about anomalies and extremes on seasonal timescales. However, their skill remains highly region- and variable-dependent. In Central Europe, predictive skill is generally modest and particularly limited for precipitation, which is strongly influenced by internal atmospheric variability and small-scale processes that are not fully resolved by global models. These limitations highlight the need for advanced postprocessing techniques to extract usable local-scale information.
This study investigates the potential of neural network–based statistical postprocessing to improve seasonal forecasts of near-surface air temperature and precipitation in the Czech Republic. The approach is based on empirical relationships between local observations and large-scale predictors derived from global seasonal forecast systems.
We use hindcast data from the Copernicus Climate Change Service (C3S), focusing on four forecast systems: ECMWF, Météo-France, DWD, and CMCC. Predictor variables include near-surface air temperature, sea level pressure, and selected large-scale circulation fields relevant for precipitation variability. Observational gridded datasets derived from station measurements serve as the reference for both temperature and precipitation.
The analysis covers the common hindcast period 1993–2016 over the Czech Republic. Neural networks are applied as a nonlinear postprocessing tool to capture complex relationships between predictors and local climate variables. Forecast performance is evaluated using categorical verification (above-normal, normal, below-normal conditions) for both temperature and precipitation.
Results indicate that neural network postprocessing improves forecast skill for temperature, particularly in winter months and at shorter lead times, while improvements for precipitation remain limited. These results demonstrate the potential of neural networks to enhance seasonal prediction skill in Central Europe, while also highlighting their limitations.
How to cite: Kliegrová, S., Metelka, L., Solánská, J., and Štěpánek, P.: Neural Network Postprocessing of Long-Range Forecasts of Temperature and Precipitation in the Czech Republic, EMS Annual Meeting 2026, Utrecht, Netherlands, 6–11 Sep 2026, EMS2026-261, https://doi.org/10.5194/ems2026-261, 2026.
Supporters & sponsors
