This presentation will provide an overview of recent scientific and technological developments at ECMWF, including the 2024 upgrade of the Integrated Forecasting System (IFS).
IFS Cycle 49r1 was implemented operationally on 12 November 2024 and substantially improved near surface wind and temperature predictions, especially in the Northern Hemisphere winter season. Key changes to the forecast model in Cycle 49r1 included the introduction of the Stochastically Perturbed Parametrisations (SPP) scheme for model uncertainty, improvements in wave and convective physics, and updates to land surface and atmospheric composition models.
Updates to data assimilation and observation usage in Cycle 49r1 included the assimilation of 2m temperature observations, assimilation of additional satellite data over sea ice, improved modelling of ocean emission and reflection and all-sky assimilation of AMSU-A, several changes to the land data assimilation system, and the introduction of version 13.2 of the radiative transfer model, RTTOV. The grid spacing of the Ensemble of Data Assimilations (EDA) was also reduced from 18 km to 9 km, with the inner loop minimisation grid reduced from 100 km to 40 km.
The impacts of Cycle 49r1 include substantial improvements to 2m temperature and 10m wind speed forecasts, increased spread for tropical cyclones intensity forecasts, a slight reduction of extreme wind forecast errors, and changes to representation of snow cover and snow density. The land data assimilation and model changes lead to a systematic reduction of soil moisture and higher spatial variability in soil moisture levels. At sub-seasonal lead times, Cycle 49r1 has small but statistically robust impacts on ensemble spread, which are driven by the switch to SPP scheme for model uncertainty. These changes are most evident in the tropics, where ensemble spread in the free atmosphere is reduced by several per cent, which represents a slight improvement in ensemble reliability relative to Cycle 48r1. In addition, Cycle 49r1 slightly improves the skill of Madden–Julian Oscillation (MJO) forecasts during weeks 3-4.
Cycle 49r1 is also the foundation for Cycle 49r2, a non-operational IFS cycle that will introduce new versions of the NEMO4/SI3 ocean and sea-ice model and underpin the 6th generation atmosphere and ocean reanalyses (ERA6/OCEAN6), the new version of the atmospheric composition reanalysis (EAC5), and the next seasonal prediction system (SEAS6).
In parallel to ongoing development of the IFS, ECMWF has developed the Anemoi machine learning toolbox to facilitate the development of data-driven weather prediction models, including deterministic and ensemble variants of the ECMWF AIFS. Real-time evaluation of pre-operational AIFS configurations has demonstrated that they are capable of very skilful medium-range forecasts for a range of upper-atmosphere variables, surface weather variables, and tropical cyclone tracks. The first operational version of the AIFS will be implemented later this year.
Finally, higher-resolution modelling capabilities are being accelerated by Digital Twin developments under the European Commission Destination Earth programme, which is building km-scale capability for a range of potential future HPC architectures. Major efforts are being invested in the code scalability of the Integrated Forecasting System to be able to run on GPUs and investigate alternative dynamical core options.
How to cite: Roberts, C., Rabier, F., Flemming, J., and English, S.: Recent progress and outlook for the ECMWF Integrated Forecasting System , EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-10861, https://doi.org/10.5194/egusphere-egu25-10861, 2025.
National Oceanic and Atmospheric Administration’s (NOAA’s) Environmental Modeling Center (EMC) is a lead developer of operational Numerical Weather Prediction (NWP) systems at the National Weather Service (NWS), which are used for the protection of life and property and the enhancement of the economy. EMC transitions to operations and maintains more than 20 numerical prediction systems that are used by NWS, NOAA, other United States (U.S.) federal agencies, and various other stakeholders. These systems are developed through a close collaboration with academic, federal and commercial sector partners. EMC maintains, enhances and transitions-to-operations numerical forecast systems for weather, ocean, climate, land surface and hydrology, hurricanes, and air quality for the U.S. and global domains.
NOAA’s operational predictions are transitioning to the Unified Forecast System (UFS) framework in order to simplify the operational prediction suite of modeling systems. The UFS is being designed as a community-based, comprehensive atmosphere-ocean-sea-ice-wave- aerosol-land coupled Earth modeling system with coupled data assimilation and ensemble capabilities, organized around applications spanning from local to global domains and predictive time scales ranging from sub-hourly analyses to seasonal predictions. Disparate legacy operational applications that have been developed and maintained by EMC in support of various stakeholder requirements are being transitioned to the UFS framework. The transition started several years ago and is planned to continue over the next few years. Fewer resulting applications will consolidate NCEP’s Production Suite that shares a set of common scientific components and technical infrastructure. This streamlined suite is expected to accelerate the transition of research into operations and simplify maintenance of operational systems.
There is also a major development effort in the area of AI/ML for NWP, and EMC has stepped up its efforts in adopting and testing the new technologies that show significant promise in revolutionizing operational NWP for NOAA.
This talk describes major development and operational implementation projects at EMC over the last couple of years, and progress in advancing UFS applications for operations. We will also present EMC plans for AI/ML for NWP, within the overall NOAA strategy, and how planned efforts link with other modeling efforts within NOAA, in the broader U.S. and international community.
How to cite: Tallapragada, V., Kleist, D., Yang, F., Barton, N., Carley, J., and Levit, J.: NOAA’s Environmental Modeling Center Update: Transitioning to Unified Forecast System Applications for Operations: Accomplishments till date and Future Plans, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-11527, https://doi.org/10.5194/egusphere-egu25-11527, 2025.
In this study, a latent heat nudging lightning data assimilation (LDA) method independent of the flash rate was developed and tested with data from the Lightning Mapping Imager (LMI) onboard the Feng-Yun-4A (FY-4A) satellite based on the Weather Research and Forecasting (WRF) model. In this LDA method, the positive temperature perturbations at the lightning location are first calculated by the difference between the moist adiabatic temperature of a lifted air parcel and the model temperature. The positive temperature perturbations in the mixed-phase region are then assimilated by a nudging method to adjust the latent heat within the convective system. Meanwhile, the water vapor mixing ratio is adapted to the temperature perturbations accordingly to constrain the relative humidity to remain unchanged. This method considers the physical nature of the convective system, in contrast with other LDA methods that establish an empirical or statistical relationship between the lightning flash rates and model variables.
The impact of this LDA method on short-term (≤6 h) forecasts was evaluated using two severe convective events in eastern China: a multi-region heavy rainfall event and a thunderstorm high-wind event. The results showed that LDA could add thermodynamic information associated with the convective system to the WRF model during the nudging period, leading to a more reasonable storm environment. In the forecast fields, the simulations with LDA produced more realistic convective structures, resulting in an improvement in forecasts of precipitation and high winds.
How to cite: Gao, Y. and Wang, X.: Impact of Assimilating FY-4A Lightning Data with a Latent Heat Nudging Method on Short-Term Forecasts of Severe Convective Events in Eastern China, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-3879, https://doi.org/10.5194/egusphere-egu25-3879, 2025.
FengYun-3E (FY-3E), the fifth satellite in China's second-generation polar-orbiting satellite FY-3 series, was launched on 5 July 2021. FY-3E carries a third-generation microwave temperature sounder (MWTS-3) and a second-generation microwave humidity Sounder (MWHS-2). In this study, the influence of assimilating FY-3E MWTS-3 and MWHS-2 clear-sky radiance data on tropical cyclone forecasts in a regional model is investigated through a series of data assimilation experiments. More than five typhoons from the northwest Pacific Ocean during 2024 typhoon season are selected for the numerical experiments of assimilation and forecasts, and the assimilation effects of FY-3E MWTS-3 and MWTS-2 are carefully evaluated. The results show that assimilation of MWTS-3 and MWTS-2 has positive impact on typhoon track forecasts, especially for forecasts beyond 12 hours, in terms of intensity forecasts, the impact of the data is neutral or slightly positive.
How to cite: Zhang, L., Niu, Z., Weng, F., and Huang, W.: Direct Assimilation of FY-3E Microwave Sounding Channel Data in a Regional Model to Improve Typhoon Forecasts, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7592, https://doi.org/10.5194/egusphere-egu25-7592, 2025.
An ensemble down-selection method is proposed to improve the analysis and forecast with a small ensemble, reducing computational needs. A usual problem with ensemble down-selection is that, despite of the reduction of forecast error, ensemble spread sharply decrease. To limit ensemble spread collapse, this study introduces two variations of a novel down-selection method seeking to minimize the sub-ensemble’s Continuous Ranked Probability Score (CRPS), thereby preserving ensemble spread while minimizing forecast error. The approaches are then tested with a regional-scale model whose precipitation forecast we seek to improve. The precipitation forecast performance of sub-ensembles obtained by these CRPS-based methods is evaluated against the full ensemble, and 100 randomly down-selected sets using various verification metrics measuring precipitation forecast skill. Results demonstrate that the CRPS-based sub-ensembles improve probabilistic forecast accuracy by achieving lower CRPS with the lowest Root Mean Square Error (RMSE) value, especially for short forecasts, without increasing false alarms. Additionally, the Brier Score shows improved forecasts, while Fraction Skill Score (FSS) confirms the improved spatial accuracy in light precipitation. These findings suggest that CRPS-based methods are viable sub-ensembling approaches for balancing accuracy, reliability, and computational efficiency in operational forecasting. By preserving ensemble spread, they improve the sub-ensemble's capacity to represent uncertainty, offering a practical and robust solution for ensemble down-selection.
How to cite: Lee, M.-T., Yau, M.-K., Jacques, D., and Fabry, F.: Ensemble precipitation down-selection methods using Continuous Ranked Probability Score (CRPS): Balancing accuracy and spread under computational constraints, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-6903, https://doi.org/10.5194/egusphere-egu25-6903, 2025.
A parameterisation suite is the combination of all parameterisation schemes that is used by a numerical model of the atmosphere. These parameterisation (or “physics”) suites are widely seen as the most uncertain components of atmospheric models.
In MUMIP we compare deterministic parameterisation suites from across different modelling centres under common prescribed large-scale dynamics. In the first MUMIP experiment, these dynamical tendencies have been derived by coarse-graining the convection-permitting ICON DYAMOND simulation to 0.2 degree resolution. We use these realistic spatiotemporal dynamical patterns to drive millions of single column model simulations over the tropical Indian Ocean with prescribed SSTs. We use this data to estimate the uncertainty from their physics across four models, each using their default convection-parametrised physics suites. The models are: IFS, GFS, RAP and ARPEGE.
The distributions of precipitation rate, convective available potential energy (CAPE), convective inhibition (CIN) and level of neutral buoyancy are analysed, as well as individual model tendencies and rate of change of CAPE and CIN as a function of lead time and, for instance, the diurnal cycle . We find notable differences across the physics suites and even more strongly between convection-parameterised physics suites and the convection-permitting ICON DYAMOND benchmark. Furthermore, we relate these diagnostics to biases in temperature and specific humidity. We also develop a framework for the detection of statistical relations among diagnostics and/or their change. The framework may for instance be used to quantify the impact of spin-up compared to persistence ("memory") and randomness within a dataset and to identify similarity in the physics across modelling centres.
In this contribution some of the early results of the international MUMIP project will be presented and we hope to encourage other researchers to use and/or complement the data of MUMIP. Please refer to https://mumip.web.ox.ac.uk for details of how to get involved.
How to cite: Groot, E., Christensen, H., Sun, X., Newman, K., Lfarh, W., Roehrig, R., Singh, K., Lambert, H., Beck, J., Williams, K., Bernadet, L., and Berner, J.: Precipitation rate, convective diagnostics and spin-up compared across physics suites in the model uncertainty model intercomparison project (MUMIP) , EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7206, https://doi.org/10.5194/egusphere-egu25-7206, 2025.
This study has been inspired by the activities developed within the IRIDE project in the service chain on “Hydro-meteorological mapping and monitoring atmospheric structure”. The work presents a preliminary evaluation of three numerical weather prediction models, WRF (Weather Research and Forecasting), ICON (ICOsahedral Non-hydrostatic), and COSMO (COnsortium for Small-scale MOdelling), by comparing synthetic and observed brightness temperatures (BTs) from the Meteosat Second Generation geostationary satellite. Synthetic satellite images were generated using the Radiative Transfer for TOVS (RTTOV) model, version 13.2. The analysis spans a verification period of over one month, with all models operating at a horizontal resolution of approximately 2 km and a temporal resolution of 1 hour.
A special focus of the study is the evaluation of the models' ability to reconstruct the intense weather events that struck some Italian regions during the recent years. This severe event caused widespread damage and highlighted the critical need for accurate and timely forecasting capabilities. By analyzing the models' performance during this extreme weather event, we aim to identify strengths and limitations in their ability to simulate localized and high-impact phenomena.
To assess the performance of the models, key verification metrics were calculated to provide a quantitative basis for understanding the accuracy and reliability of the models in predicting atmospheric conditions as represented by BTs.
The results of the verification are thoroughly discussed, with particular emphasis placed on their broader implications for both the development and refinement of numerical weather prediction models. This discussion delves into how these findings can inform improvements in various aspects of model design, from enhancing their ability to simulate complex physical processes to addressing persistent biases and inaccuracies. Differences in model performance are meticulously analyzed to identify potential sources of error, which may arise from a range of factors such as deficiencies in physical parameterizations, limitations in boundary condition specifications, or inaccuracies stemming from radiative transfer assumptions.
These analyses aim to provide a deeper understanding of the underlying causes of discrepancies, paving the way for more targeted adjustments. This work represents a significant contribution to the ongoing evolution of high-resolution numerical weather prediction models, offering a wealth of valuable insights for both researchers striving to push the boundaries of modeling capabilities and operational forecasters seeking to improve real-time prediction accuracy. By shedding light on the intricate interplay between model dynamics and observational data, it underscores the importance of continuous innovation and refinement in the pursuit of more reliable and precise forecasting tools.
How to cite: Giugliano, G., Fedele, G., Bonfiglio, A., Campanale, A., Raffa, M., Antonelli, P., and Mercogliano, P.: Performance Evaluation of High-Resolution Numerical Weather Prediction Models Using MSG Brightness Temperatures, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9393, https://doi.org/10.5194/egusphere-egu25-9393, 2025.
2023 was record-breaking for wildfires in Canada with unprecedented impacts on local ecosystems as well as large scale smoke hazards. These exceptional fire impacts rose the public demand for accurate forecasts of smoke plumes as well as analysis of air quality impacts. However, fire smoke plumes are extreme air quality events with exceptionally high concentrations and related uncertainties fall outside statistical ranges. These particular conditions induce specific challenges for data assimilation algorithms, because error estimates need to capture the high uncertainties and spatial gradients. At the same time, operational forecast systems require high computational efficiency to deliver fast, yet accurate forecasts to the public.
This study explores the potential of a novel assimilation approach, called parametric Kalman filter (PKF), for operational air quality forecasting during extreme air quality events. By explicitly evolving the main error parameters, the PKF has been proven to provide accurate uncertainty estimates at very low computational costs. In this work, a dynamical propagation of error standard deviations is implemented in the Canadian atmospheric-chemical forecast model GEM-MACH. This extended forecast model is applied to a case study of Quebec wildfires in early July 2023. First results indicate that the forecast error distributions during this events can be sufficiently approximated by a passive error-tracer. It is demonstrated that vertical diffusion is a critical component for dynamical error forecasting of extreme air quality events. The error standard-deviation forecasts are used in the current objective analysis (OA) for surface air quality at ECCC (Environment and Climate Change Canada) and compared to operational OA results.
How to cite: Vogel, A., Ménard, R., Abu, J., and Chen, J.: Potential of error-evolving tracer forecasts for operational assimilation of PM2.5 during wildfire smoke episodes, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9428, https://doi.org/10.5194/egusphere-egu25-9428, 2025.
NASA is developing an Earth Science Modeling Strategy. This is motivated by recommendation 4.2 from the United States National Academies of Sciences, Engineering, and Medicine, Thriving on Our Changing Planet: A Midterm Assessment of Progress Toward Implementation of the Decadal Survey (2024, https://doi.org/10.17226/27743): “To ensure continued advances in modeling in conjunction with Earth observation: NASA should develop a long-term strategic plan for a strong sustained commitment to Earth system modeling in concert with observations. Success in observation-driven modeling holds the key for maintaining the end-to-end capability that has served NASA well in its effectiveness and service to society.” Moreover, one of the main objectives of NASA’s Earth Science to Action Strategy (https://science.nasa.gov/earth-science/earth-science-to-action/) is to Deliver Trusted Information to Drive Earth Resilience Activities. This Objective will rely on comprehensive Earth system modeling as a key result that will enable NASA to advance and integrate Earth science knowledge to empower humanity to create a more resilient world.
In this presentation we will overview the approach being taken to develop the Earth Science Modeling Strategy, within NASA and with the broader community. Some of the key aspects being considered include comprehensive state-of-science modeling representation of the coupled Earth system, from global to local scales, analyses and predictions at different lead times, from short term predictions to climate projections, and using ensembles. Another key facet is data assimilation into Earth system models and improved utilization and demonstration of the value of Earth observations. It is envisioned that bold innovation, new disruptive technologies, including artificial intelligence and machine learning, and utilization of large Earth science datasets will be key enablers for cutting edge research, increased understanding of the Earth system, and improved ability to provide actionable Earth science information for societal applications to meet broad user needs. Modeling underpins NASA’s Earth Science to Action strategy as a key capability for advancing foundational knowledge, Earth system science and applied research, as well as increasing societal value of NASA’s data and information leading to improved public understanding.
How to cite: Stajner, I.: Development of NASA’s Earth Science Modeling Strategy, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-3327, https://doi.org/10.5194/egusphere-egu25-3327, 2025.
The KIM(Korean Integrated Model)-based local ensemble model, which has a 3km horizontal resolution and 13 members, was developed to improve the prediction of heavy rainfall. The members of the local ensemble model were generated by downloading the KIM global ensemble model. The initial and boundary fields for the members were provided by the KIM global ensemble model. The local ensemble model covers the Korean Peninsula and surrounding areas and produces a 5-day forecast twice a day.
With the introduction of the KIM local ensemble model, the CSI score for precipitation were improved alleviating the underestimation of precipitation in the KIM global model. In summer, the 75% and 90% percentiles of the local ensemble model show the best performance in heavy rainfall forecasting, while in winter, the median provides the best results.
The analysis verification(RMSE) results also showed that the KIM local ensemble model generally provided improved outcomes compared to the KIM regional model and exhibited similar performance to the UM (Met Office Unified Model)-based local ensemble.
The summer season on the Korean Peninsula is characterized by frequent extreme rainfall events, and this extreme rainfall presents a major challenge for forecasters in producing accurate forecasts. Therefore, various strategies using local ensembles have been developed to predict these extreme rainfall events. Probability matching and percentiles are representative methods, and by employing these techniques, many of the issues associated with the underestimation of extreme rainfall in numerical weather prediction have been largely addressed.
Keywords: local ensemble model, regional model, member, RMSE, CSI, underestimation, extreme rainfall, probability matching, percentiles
How to cite: Shin, H., Kim, E. J., Yun, S., Park, J.-I., Ha, J.-C., and Kim, D.-J.: Development and Application of KIM-based Local Ensemble Model, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-10029, https://doi.org/10.5194/egusphere-egu25-10029, 2025.
Improving the representation of the initial state of the atmosphere in the Numerical Weather Prediction (NWP) model is critical for advancing the quality of weather forecasts which are vital for our daily life. Wind, cloud and precipitation are driving factors for Earth’s water and energy cycles and sometimes they can represent weather-related threats. Uncertain measurements of these variables present challenges for NWP models.
The WIVERN (Wind Velocity Radar Nephoscope) mission (Illingworth et al., 2018) is one of two candidate missions in Phase A studies for potential selection as the Earth Explorer 11 mission under the European Space Agency’s FutureEO programme. WIVERN would be the first-ever satellite to measure global in-cloud winds. The data from WIVERN is expected to provide significant benefits across multiple sectors, including advancing our understanding of weather phenomena, validating climate statistics, and improving the NWP models performance.
We focus on the NWP performance after assimilating WIVERN Doppler data, specifically Line of Sight (LoS) winds, for the high-impact case study of the Medicane Ianos, occurred in mid-September 2020 in the central Mediterranean. The experimental results of WIVERN Doppler assimilation are compared with those obtained from the output of similar experiments assimilating other data types: the Advanced SCATterometer (ASCAT) radar data, radiosoundings, and Atmospheric Motion Vectors (AMV).
WIVERN pseudo-observations were generated by running an ensemble of WRF at a 4 km horizontal resolution, using the European Centre for Medium range Weather Forecast – Ensemble Prediction System (ECMWF-EPS) analysis/forecast cycle issued at 12 UTC on 16 September 2020 as initial and boundary conditions. The approach consisted of the following steps:
- The WRF model was run using the initial and boundary conditions from all 51 ECMWF-EPS members.
- The forecast trajectories of Medicane Ianos from the 51 WRF ensemble members were compared to the observed trajectory.
- The best WRF member, i.e., the one with the closest agreement between the simulated and observed trajectories, was selected.
- Pseudo-observations were generated from the output of the selected WRF best member.
- These pseudo-observations were assimilated into all other members of the WRF ensemble.
For consistency, all observations in this study were pseudo-observations. Assimilation and forecast were performed at 12 UTC on 17 September, followed by a 24-hour forecast.
The trajectories followed by the Medicane are evaluated considering the assimilation of different data sources. Results show marginal improvement of the Ianos’ trajectory when radio-soundings or Atmospheric Motion Vector (AMV) are assimilated, while the trajectory forecast is substantially improved by ASCAT data assimilation (20% improvement). The assimilation of WIVERN data is very important, as the trajectory forecast was improved by over 40%.
A similar positive impact is shown when WIVERN data are assimilated together with other data sources. Specifically, two additional experiments were conducted: in the first, all data sources except WIVERN were assimilated, while in the second, WIVERN data were included. The results show an important improvement of over 10% in the trajectory forecast of Medicane Ianos when WIVERN data are used in combination with ASCAT, AMV and radio-soundings observations.
References
Battaglia, A., et al., 2022, https://doi.org/10.5194/amt-15-3011-2022.
Illingworth, A. J., et al., 2018, DOI: 10.1175/BAMS-D-16-0047.1, 1669-1687.
How to cite: Federico, S., Torcasio, R. C., Transerici, C., Montopoli, M., Pourshamsi, M., and Battaglia, A.: Assimilation of WIVERN Doppler Data in Weather Research and Forecasting (WRF) Model for the Medicane Ianos: A Comparison with Alternative Data Sources, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-11144, https://doi.org/10.5194/egusphere-egu25-11144, 2025.
Hailstorms are highly localized severe weather events that can cause extensive damage to agriculture, infrastructure, and property, necessitating accurate forecasting for effective risk mitigation. The Weather Research and Forecasting (WRF) model, a numerical model which is able to simulate features from a wide range of scales, offers a range of physics parameterizations to simulate sub-grid scale processes which are essential for hail storm forecast. However, the vast number of possible configurations complicates the identification of an optimal setup for hail simulation. This study leverages a genetic algorithm (GA) to systematically optimize WRF physics parameterizations for hail prediction over Central Europe, focusing on the severe hail events of June 2022.
The GA framework encodes WRF physical parameterizations configurationsas individuals within a population, evolving through selection, crossover, and mutation across multiple generations. Fitness is evaluated using the F2 score, prioritizing recall to address the imbalance between observed hail and non-hail events. By exploring over 2.4 million potential configurations, the GA provides the best combinations of physical parametrizations to capture the spatial and temporal characteristics of hailstorms. The results show that this methodology enables the exploration of a wide range of possible configurations, demonstrating its potential to optimize parameterizations for high-impact weather events effectively. This novel methodology represents a substantial step toward advancing hail forecasting capabilities using high-resolution NWP models.
How to cite: Guerrero-Calzas, I., Rossetto, L., Cortés Fité, A., Hanzich, M., and Miró, J. R.: Applying a Genetic Algorithm to Optimize Hail Prediction Using the Weather Research and Forecasting Model, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-12631, https://doi.org/10.5194/egusphere-egu25-12631, 2025.
Aircraft data are considered one of the best platforms for obtaining atmospheric spatial information in the observation gap over the ocean. The National Institute of Meteorological Sciences (NIMS) has operated an atmospheric research aircraft to mitigate this observation gap. In particular, the dropsonde and AIMMS-20 systems installed on the aircraft generate vertical distributions of meteorological variables over the ocean, and these specialized observation data enhance the accuracy of the initial model fields. These aircraft observation data provide continuous distributions of meteorological variables and significantly contribute to improving the performance of numerical predictions. In this study, we evaluated the effectiveness of data assimilation (DA) on the prediction of severe meteorological phenomena affecting the Korean Peninsula using high-resolution numerical modeling using atmospheric research aircraft observation data. To analyze the sensitivity of the difference in the background error covariance in the data assimilation method, three sets of simulation experiments were performed. First, an experiment was conducted using the background error covariance option CV3 based on the NMC method, which is suitable for simple settings or when the computational resources are limited. Second, an experiment using option CV5 is suitable for studying more complex situations or high-accuracy forecasts. This option generates a covariance structure that adapts to atmospheric conditions by using an ensemble-based method. The last is an experiment using the CV7 option, which is a hybrid background error covariance option that combines static methods (such as CV3) and ensemble-based methods (such as CV5), and has the advantage of combining climate statistics and flow-dependent features to improve model prediction performance.
How to cite: Han, S.-B., Goo, T.-Y., Jung, S.-P., Kim, M.-S., Kang, D.-D., and Lee, C.: Sensitivity analysis of severe weather events to different background error covariances in meteorological aircraft data assimilation, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-14710, https://doi.org/10.5194/egusphere-egu25-14710, 2025.
Numerical Weather Prediction (NWP) has recently lost its hegemony in weather forecasting, as more machine-learning-based models achieve results comparable to NWP. It turns out that data-driven models are capable of identifying patterns and distilling physical laws that, until now, have only been formulated by atmospheric physics specialists. Although ML-based models are already being used by meteorological institutes alongside NWP-based models, this does not mean that NWP will fade into irrelevance. In this talk, we will show how NWP and ML can interoperate to achieve the shared goal of providing more accurate weather forecasts. From NWP providing high-quality training data for ML models to ML models replacing specific parameterizations, the spectrum of collaboration is vast. ML models cannot succeed without high-quality training data provided by NWP, and NWP can benefit from this new technology by incorporating ML models in places where conventional physics parameterizations are found lacking. None of the currently successful ML-based models would exist without the high-quality reanalysis data generated through numerical models. Decades of expertise and extensive research in numerical modelling now serve as a solid foundation for the remarkable achievements of data-driven applications. The future of weather forecasting is built on this synergy — numerical modelling and machine learning working together to achieve what neither could accomplish on its own.
How to cite: Rognvaldsson, O. and Stanislawska, K.: Numerical Weather Prediction meets Machine Learning - a synergy for better forecasts, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-18692, https://doi.org/10.5194/egusphere-egu25-18692, 2025.
Medicanes are very dangerous meteorological phenomena with large uncertainty on genesis and intensification usually case dependent. The peculiarity of medicane Daniel analyzed in this study is the long life and strong tropical-like characteristics even on land with baroclinic atmosphere. It is essential to deepen the knowledge of these events to improve operational forecasts and waring systems. In this perspective, the models used in this analysis have reported results sufficiently close to observations. In particular, the WRF model performed better in terms of temporal synchronization of the phenomenon, internal structure of the cyclone and spatial distribution of precipitations; while ICON better modeled lower layers and highlighted different feature on track and tropical transition.
For both the models the tracks obtained from the simple algorithm used are discrete, with major errors in the initial phase. The landfall was simulated with acceptable errors. Minimum Mean Sea Level Pressure values are modeled as lower than the observed one, with WRF simulating a most intense cyclone. Wind speed data correctly passed the threshold for classification as a Category 1 hurricane, although WRF overestimated the mid-tropospheric wind. The Hart's Cyclone Phase Space diagram consistently highlighted the tropical phase of the medicane with a symmetric deep warm core at its most intense period, but the values of the parameters differ from simulation to simulation. Finally, the point values of precipitation are not satisfactory in any model even if the field of cumulated precipitation is consistent with the observations.
How to cite: Serafini, P., Ricchi, A., Marsigli, C., and Ferretti, R.: Multi-model high resolution analysis of Mediterranean Hurricane Daniel with WRF and ICON, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-19128, https://doi.org/10.5194/egusphere-egu25-19128, 2025.
The evaluation of two reanalysis precipitation datasets, CRA40 and ERA5, was conducted over the Ganjiang River Basin, utilizing precipitation records from 37 ground rainfall gauges and surface-observed stream flow data spanning from January 1998 to December 2008. Both CRA40 and ERA5 were found to effectively capture the spatial and temporal precipitation characteristics at the basin scale. However, significant differences in precipitation quality were observed between the two. ERA5 demonstrated superior accuracy in depicting short-term precipitation changes, particularly on a daily basis. In contrast, CRA40 exhibited better performance on a monthly scale, offering more stable and long-term precipitation trends. Simulations of stream flow using the VIC hydrological model driven by these two precipitation products revealed that (1) CRA40 outperformed ERA5 in both daily and monthly stream flow simulations, with a higher Nash-Sutcliffe Efficiency (NSE, 0.65 for CRA40 vs. 0.6 for ERA5) and a greater correlation coefficient (CC, 0.96 for CRA40 vs. 0.91 for ERA5). Although ERA5 had a relatively good CC (0.86 and 0.93 respectively), its NSE was notably poor (0.29 and 0.30 respectively); (2) both CRA40 and ERA5 tended to overestimate stream flows in the basin, especially during the flood season (April-September), with ERA5's overestimation being more evident. This study is anticipated to offer a foundation for selecting reliable reanalysis products for precipitation and hydrological simulations in the Ganjiang River Basin.
How to cite: li, Z., Zhou, Z., and Chen, S.: Hydrological Evaluation of CRA40 and ERA5 Reanalysis Precipitation Products over Ganjiang River Basin in Humid Southeastern China, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-20414, https://doi.org/10.5194/egusphere-egu25-20414, 2025.
Recent advances in artificial intelligence (AI), specifically with applications of deep learning, have brought paradigm-shifting changes to Numerical Weather Prediction. Recent AI-based NWP systems have rivaled traditional physics-based global NWP systems according to some verification metrics. However, the performance of these systems for extreme events, and their implications for atmospheric predictability, has not yet been fully explored. In this study, the practical predictability for winter storms in eastern North America will be compared using forecasts generated by several traditional NWP and AI-NWP systems. In addition to domain-wide verification statistics, the realism of cyclone structure and evolution will be evaluated at different forecast lead times. We plan to discuss the ensemble predictability of events, evaluating the sensitivity of the AI-NWP systems to initial condition perturbations, with implications for data assimilation. Finally, at the mesoscale, we will demonstrate a convection initiation nowcasting system that utilizes deep learning to generate probabilities of new convection forming at lead times under one hour, which we interpret using explainable AI and uncertainty quantification.
How to cite: Greybush, S. and Spallone, C.: Implications of AI for Atmospheric Predictability of Convection and Winter Storms, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-20960, https://doi.org/10.5194/egusphere-egu25-20960, 2025.
Reconstructing historical weather and climate at the daily scale during the early 19th century (1806–1821) is crucial for understanding variability and predictability in this data-scarce period. This study assimilates additional historical pressure and temperature observations into the 20th Century Reanalysis Version 3 (20CRv3) to improve the reliability of daily reconstructions over Europe.
We use state-of-the-art data assimilation methods, with the Ensemble Kalman Filter as the primary framework for integrating historical series. Alternative techniques, including the Ensemble Square Root Filter and the Iterative Ensemble Smoother, are also investigated to study their performance in capturing daily weather.
We assess the impact of these methods using evaluation metrics, probabilistic measures, and comparisons to independent observations. Our results reduce the ensemble spread and uncertainty of 20CRv3, providing insights into internal variability at the daily scale and short-term climate dynamics.
How to cite: Corbella, C.: Reconstructing Daily Weather in the Early 19th Century: New Insights from Data Assimilation , EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-3762, https://doi.org/10.5194/egusphere-egu25-3762, 2025.
Ocean surface albedo (OSA) plays a important role in the energy balance of the climate systems. In climate models, it is typically treated as a constant or represented by a simplistic function of Solar Zenith Angle (SZA). However, research by Jin (2011) indicates that OSA can be significantly influenced by whitecaps under moderate to high surface wind conditions (denoted as Jin11 scheme). Whitecap coverage is a key factor in this parameterization, often expressed as a power function of surface wind speed. Given that water depth and wave height are associated with wave breaking—of which whitecaps are the primary manifestation—the ratio of theoretical wave height to water depth has been incorporated into the Jin11 scheme for adjustment. This modification reflects the characteristic that certain areas are more prone to whitecaps under identical wind conditions.
In this study, we incorporated this improved OSA parameterization scheme into the Community Earth System Model Version 2 (CESM2) and conducted coupled simulation experiments. The numerical results show an alleviation of the excessive reduction in sea surface temperature in the equatorial ocean, the North Pacific subtropical gyre circulation, and the southern westerly wind belt as simulated by the Jin11 scheme. Additionally, longwave radiative heating in the tropical regions is significantly altered after accounting for the wave breaking factor. Precipitation simulations over the northwest Pacific, the tropical Indian Ocean, and the Indo-Pacific Convergence Zone show improvements, while the induced substantial changes in latent heating have further affected vertical motion and convective activity.
How to cite: Jing, X. and Wang, L.: Development of a Ocean Surface Albedo Scheme with Wave Breaking Factor, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-3809, https://doi.org/10.5194/egusphere-egu25-3809, 2025.
The level of uncertainty of reanalysis datasets varies greatly based on the quality and amount of available observations and the uncertainty of physical parameterizations used in the background forecast model. Ensemble data assimilation (EDA) schemes are used to quantifiy this combined uncertainty. However, isolating the effects of observational and model uncertainties based on a given ensemble reanalysis is not straightforward. Here, we use the 9 member EDA ensemble produced for the ECMWF 5th Generation Reanalysis product (ERA5) to investigate synoptic scale model uncertainty and its connection to the occurrence of specific weather regimes.
To control for ensemble spread caused by observation uncertainty - especially on long time scales - we devise grid-point-wise statistical models for the logarithmic ensemble variance with temporal predictors. We use a binary segmentation algorithm to objectively identify change points in ensemble spread time-series caused by abrupt changes in the observation system.
The set of statistical models allows for statements about the relative impact of changes in the observation system on the total background forecast uncertainty between different grid-points. After filtering out the impact of changes in the observation uncertainty, we obtain a long time series of model uncertainty estimates, which we analyze climatologically with respect to flow characteristics, regime structure and impact of physical parameterizations. This provides regions of high model uncertainty for the respective regimes as well as differences in the role of model uncertainty among the regimes.
How to cite: Schoeller, H. and Pfahl, S.: Isolating the effects of model uncertainty on ensemble reanalysis data and their relation to North Atlantic flow regimes, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-4243, https://doi.org/10.5194/egusphere-egu25-4243, 2025.
The Numerical Weather Prediction (NWP) gray zone (GZ) represents a critical challenge in modeling, occurring at spatial resolutions typically ranging from approximately 500 m to 5 km, depending on factors such as the modeling framework, the prevailing atmospheric conditions, and the geographical context where neither full parameterization nor explicit simulation of physical processes is feasible. Within this range, convection parameterizations often become unreliable, particularly for cumulus clouds and turbulence, leading to uncertainties in weather forecasts. High-resolution models (below 4 km) assume explicitly resolved convection, yet this approach does not consistently improve prediction accuracy. Recent advancements in scale-aware parameterizations offer a promising solution, enabling a gradual transition from parameterized to resolved convection, enhancing model performance and reducing biases within the GZ. To explore these challenges, the Weather Research and Forecasting (WRF) model was employed to simulate eight precipitation events across Schleswig-Holstein and Baden-Württemberg in Germany, all exceeding the severe weather threshold of 40 mm/h (warning level 3) set by the German Weather Service. A comprehensive suite of 1,440 simulations was conducted, combining 10 microphysics schemes, 6 cumulus schemes, 8 event cases, and 3 spatial setups. The model setups included a single domain with a 9 km grid size and two two-way nesting configurations with spatial resolutions of 9 km and 3 km. To investigate the role of convection schemes in the convective GZ and the benefits of higher spatial resolution, simulations at 3 km resolution were run both with and without active convection schemes. Initial and boundary conditions were provided by the ERA5 dataset at a spatial resolution of 0.25°. A detailed performance analysis was carried out using pairwise comparisons and the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS), which ranked the parameterization combinations based on multiple criteria. Results revealed that non-convection-permitting setups performed better during summer precipitation events, where convection is more localized and intense. On the other hand, winter events, influenced by larger-scale processes, showed similar accuracy between convection-permitting and non-convection-permitting configurations. Interestingly, increasing resolution from 9 km to 3 km did not consistently improve model performance. Furthermore, the best-performing parameterizations at 9 km resolution outperformed those at 3 km across all configurations, challenging the common assumption that higher resolution inherently improves model accuracy. These findings emphasize the need to carefully balance resolution and parameterization choices in severe weather forecasting, particularly for convective systems. The study underscores the critical influence of model physics and nesting configurations on simulation outcomes, offering valuable insights for future research and operational modeling efforts.
How to cite: Sotiropoulou, R.-E. P., Stergiou, I., Traka, N., Kaskaoutis, D. G., and Tagaris, E.: Enhancing Precipitation Predictions in the WRF Model: The Role of Convection Schemes and Increased Spatial Resolution in the Convective Gray Zone, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-5157, https://doi.org/10.5194/egusphere-egu25-5157, 2025.
The Geostationary Interferometric Infrared Sounder (GIIRS) on board the Fengyun 4B (FY-4B) satellite is the first hyperspectral interferometer flying in geostationary orbit. It can provide atmospheric information with high spatial and temporal resolution, which has significant potential for application in regional numerical weather prediction (NWP) models. Due to the high correlation between the infrared hyperspectral channels, it is critical to accurately characterize the inter-channel observational error correlation (IOEC) for assimilating the GIIRS radiance data effectively. This study firstly constructed an observation error covariance matrix for considering the inter-channel correlation of FY-4B GIIRS radiance data based on the NWP system developed by the China Meteorological Administration Beijing Urban Meteorological Institute (CMA-BJ). There was a strong error correlation between adjacent channels and channels with similar detection for the GIIRS radiances. Single-point observation assimilation experiments indicated that the IOEC had a significant impact on the magnitude and structure of temperature and humidity analysis increments. Two groups of assimilation experiments over a 10-day period were carried out and compared. The results showed that an average improvement of 1.5% could be obtained in the RMSE of the temperature and humidity forecasts within the first 12 hours incorporated the IOEC. With the IOEC, a positive impact was also achieved on the precipitation forecast skill, although it was not significant.
How to cite: Xie, Y.: Assimilation and Impact of FY-4B GIIRS Radiances in CMA-BJ Numerical Weather Prediction System, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-5543, https://doi.org/10.5194/egusphere-egu25-5543, 2025.
Precipitation datasets derived from reanalysis products play a critical role in weather forecasting and hydrological applications. This study aims to assess the performance of two distinct reanalysis precipitation products, i.e., the first-generation Chinese global land-surface reanalysis precipitation product (CRA40) and the fifth-generation European reanalysis precipitation product (ERA5), over mainland China. The assessment is conducted with daily-scale gridded-point rain gauge data obtained from Chinese surface meteorological stations as reference, and the continuous and categorical statistical indicators as assessment metrics. The findings of this study are as follows: 1) CRA40 outperforms ERA5 in terms of the 13-year daily mean precipitation and seasonal daily precipitation. CRA40 shows better correlation coefficients (0.97), relative biases (5.25%), root mean square errors (0.34 mm), and fractional standard errors (0.05). 2) Both CRA40 and ERA5 generally exhibit an overestimation of precipitation over mainland China. The degree of overestimation is particularly pronounced in dry climatic regions (e.g., Xizang-Qinghai plateau, Xinjiang province), while wet regions (e.g., the middle and lower reaches of Changjiang River, and South China) demonstrate relatively less overestimation. 3) ERA5 shows better performance in the detection of daily precipitation than CRA40. Neither CRA40 nor ERA5 can well capture heavy precipitation events. These findings are expected to advance our understanding of the strengths and limitations of the reanalysis precipitation products, CRA40 and ERA5, over China.
How to cite: Zhou, Z., Chen, S., Li, Z., Li, Y., and Wei, C.: Performance Assessment of CRA40 and ERA5 Precipitation Products over China, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9643, https://doi.org/10.5194/egusphere-egu25-9643, 2025.
State augmentation in a data assimilation cycle can be used as an objective method to estimate uncertain empirical constants in parameterization schemes. In this approach, empirical parameters are appended to the model state vector. They cannot be observed, but, like any other unobserved state variable, they can be updated based on their correlations with the model equivalents of observable quantities. State-parameter correlations are likely flow-dependent, therefore they are best estimated with an ensemble of simulations.
Despite its potential usefulness in parameterization design, ensemble-based parameter estimation has been used so far as a way of accounting for model errors in the assimilation process, and as a method to increase ensemble spread. In this study, we discuss if and how it can aid parameterization optimization. As a case study, we consider a simple first-order parameterization of turbulence in the atmospheric boundary layer. We run several idealized assimilation experiments, partly in a perfect-model scenario (the forecast ensemble and the nature run providing the assimilated observations are instances of the same model), partly in a more realistic imperfect-model scenario (the models providing the forecast ensemble and the nature run the have different formulations).
We demonstrate that, in our case, sensible parameter estimation results are obtained only under restrictive conditions. First, initial conditions must be very accurate, so that the spread of the forecast ensemble is determined primarily by the uncertain parameter. Second, the error variance of the assimilated observations must be low enough for the state perturbations induced by the estimated parameter to be accurately sampled. We show that, when these conditions are met, optimized parameters can compensate for sources of model error, and argue that this property can be used to extend the flexibility of parameterization schemes. For instance, this could be achieved by using parameter estimation experiments to populate lookup tables for adaptive parameters.
How to cite: Serafin, S. and Weissmann, M.: Optimizing parameterization schemes with ensemble-based parameter estimation, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-10218, https://doi.org/10.5194/egusphere-egu25-10218, 2025.
Climate change is already here, but our understanding of its local impacts in Israel is still lacking. Although large networks of in-situ observations cover Israel, and there is an increasing amount of information coming from satellites, there are still spatial and temporal gaps that are not expected to be solved in the coming decades. The problem is more pronounced in Israel than in other locations due to its complex terrain and high climate variability. These characteristics necessitate more observations (relative to other regions) to reliably sample the regional variability and allow for regular temporal and spatial data interpolation. Reanalysis datasets have become more popular in the last few decades due to their regularity in space and time, which is achieved by combining observations with model outputs using a predefined data assimilation method. However, current reanalysis products are still too coarse to represent the high climate variability in Israel, and therefore, their use is limited. In this presentation, we will describe our effort to generate a prototype high-resolution convection-permitting ensemble-based data-assimilation system and a reanalysis product for Israel.
How to cite: Strobach, E., Rozenstein, O., and Rostkier-Edelstein, D.: A Prototype High-Resolution Data-Assimilation System for Israel, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-12777, https://doi.org/10.5194/egusphere-egu25-12777, 2025.
On 16 April 2024, the United Arab Emirates experienced torrential rain, with values exceeding 200 mm in less than a day. In this study the WRF-based forecasting system Weather On Demand (WOD), developed by the Belgingur consortium in Iceland, is employed to explore the medium-range forecasts of the event, by simulations with initial values at different times. A simulation of the event with 120h lead time was very bad, while a 72h simulation was quite accurate in terms of reproducing an extreme precipitation event. A comparison of the simulations corresponding to the two forecasts reveals that the error in the 120h simulation is related to incorrect advection of dry air from the desert into the path of the convective storm across the Persian Gulf. The incorrect advection is associated with a wrong perturbation in the atmospheric flow, extending throughout the troposphere. This perturbation error is associated with an erroneous simulation of a mesoscale convective complex occurring in the vicinity of Bahrain a day before the 16 April event.
How to cite: Ólafsson, H. and Rögnvaldsson, Ó.: Analysis of an incorrect forecast of a torrential rain event in the UAE, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-10827, https://doi.org/10.5194/egusphere-egu25-10827, 2025.
Extreme rainfall events often lead to significantly heavier rainfall over urban areas compared to their surrounding regions. Predicting these positive urban precipitation anomalies during heavy rainfall remains a critical challenge in numerical weather modeling. This study explores the sensitivity of the Weather Research and Forecasting (WRF) model to approximately 70 combinations of parameterization schemes, including microphysics, cumulus convection, planetary boundary layer options, and urban canopy model schemes, focusing on urban precipitation anomalies.
The analysis is based on two significant heavy rainfall events over Chennai, India: October 22, 2006, and November 8, 2015. Model simulations are validated against Integrated Multi-satellite Retrievals for GPM (IMERG) precipitation data to evaluate their ability to capture urban anomalies. High-resolution simulations demonstrate that specific combinations of parameterization schemes, particularly those incorporating multi-level urban canopy models, enhance the model’s capacity to predict significant positive anomalies during intense rainfall events.
The findings underscore the critical role of urban canopy models in shaping precipitation intensity and spatial distribution and the interplay between cumulus convection and boundary layer processes in driving urban precipitation dynamics. These insights provide practical guidance for optimizing WRF parameterization settings, advancing the accuracy of urban-scale weather prediction, and deepening the understanding of urban hydrometeorological processes.
How to cite: Kalamalla, L. and Satyanarayana, A.: Sensitivity Analysis of Parameterization Schemes in the WRF Model for Predicting Urban Precipitation Anomalies Over Chennai, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-15854, https://doi.org/10.5194/egusphere-egu25-15854, 2025.
Land surface temperature (LST) on Iceland has been retrieved by means of remote sensing for the period 2001-2023. The trend in the data shows substantial geographical variability on different scales and is partly very different from the general upward trend in the 2 m temperatures. There are areas with strong negative trend and other areas with strong positive trend. The variability may be attributed to changes in snow cover and vegetation. Impact of volcanic eruptions and retreat of glaciers are also detected. The results suggest that using data decades back in time to train forecasting models may lead to systematic errors in surface temperature forecasts.
How to cite: Rousta, I. and Ólafsson, H.: Land surface temperature trends in Iceland, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-17036, https://doi.org/10.5194/egusphere-egu25-17036, 2025.
Persistence is the first approximation to seasonal and sub-seasonal temperature forecasting. In the present study, summer and autumn temperature persistence in mean monthly temperatures in the circumpolar Arctic is explored in time-series of monthly mean data.The temperature correlations extend from being negative to very high.
The spatial variability of temperature persistence may be linked to the Bowen ratio, static atmospheric stability, snow cover and sea-ice extent. The variability in these factors may contribute directly to seasonal variability in the radiation budget as well as in surface fluxes. In some regions there are also detectable impacts that can be associated with regional circulation patterns.
How to cite: ekrami, N. and Olafsson, H.: Arctic Temperature Persistence in Summer and Autumn and Seasonal Forecasting, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-17466, https://doi.org/10.5194/egusphere-egu25-17466, 2025.
Atmospheric profiles are indispensable for operational weather forecasting across a wide range of scales and latitudes. Despite their importance, the assimilation of tropospheric wind and temperature profiles remains a complex task with considerable potential to markedly improve weather predictions. This research investigates the impact of Ground-based Microwave Radiometer profile measurements on Numerical Weather Prediction (NWP) using a real rainfall case study. Employing the Local Error-Subspace Transform Kalman filter (LESTKF), we assimilate temperature and wind profiles derived from the Ground-based Microwave Radiometer observation network. The coupled WRF-PDAF (Parallel Data Assimilation Framework) system is utilized to conduct twin experiments. These experiments, which vary observation variables and localization distances, offer valuable insights into the assimilation process. The study evaluates potential configurations for future profile measurements and discusses recommended localization distances. The results demonstrate that incorporating multiple observation variables leads to substantial forecast improvements compared to using individual variables alone. The research culminates in a recommendation for an optimal localization distance, which has the potential to enhance the accuracy and reliability of weather forecasting.
How to cite: Shao, C., Guo, Y., and Dong, Y.: The Impact of Ground-based Microwave Radiometer Data Assimilation: A Case Study, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-18433, https://doi.org/10.5194/egusphere-egu25-18433, 2025.
We present an investigation in which two planetary boundary layer (PBL) schemes are compared at the parameterized physical process level in a fog simulation case study. The two PBL schemes in question are two options in the Unified Forecast System (UFS) for global and regional applications. We investigated the difference between the two schemes using both 3-D regional and single-column configurations of the UFS. We found that there are no significant differences in terms of parameterized physical processes. The two schemes differ mainly in the closure assumptions and the magnitudes of parameters used in the parameterization formulations, resulting in more or less success in simulating the development of the observed fog layer. Both schemes have their own error characteristics in representing essential processes for fog formation and dissipation, pointing to the uncertainty in PBL process parameterizations when observations and realistic large-eddy simulations are insufficient for process evaluation.
How to cite: Grell, E., Michelson, S., and Bao, J.-W.: A process-level comparison of two PBL schemes in the Unified Forecast System in a fog case study, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-20650, https://doi.org/10.5194/egusphere-egu25-20650, 2025.
We introduce a paradigm shift for developing parameterizations of subgrid motion in numerical weather prediction (NWP) models, which is based on recent developments in the theory of computational fluid dynamics. The governing equations of an NWP model are based on the same Navier-Stokes (N-S) equations used in computational fluid dynamics. They must be averaged over a grid-cell volume before transforming into discrete forms in order to be solved numerically. Consequently, extra terms of turbulent subgrid motion appear in these governing equations that must be approximated or parameterized. The recent development of the formal theory on the N-S equations filtered via numerical discretization concludes that such parametrizations cannot be exact for the equations to be solvable numerically. Approximation in these parameterizations is necessary for the N-S equations to be solvable as a well-posed problem. Practically, this has two implications for parameterizing subgrid motion in NWP models. First, the development of the parameterizations of subgrid motion is required to be driven by improving the accuracy of the parameterizations to address specific prominent performance issues of an NWP model, and the improvement should be based on observations of forecast variables and subgrid processes for it to be physically relevant. Second, the parameterizations should be as simple as possible for feasible performance tuning based on observations and for computational efficiency. In this presentation, we will use two examples to discuss what the problem-driven and observation-based approach means in developing subgrid convection and turbulent mixing parameterizations in NWP models. We will use the two examples to advocate that the problem-driven and observation-based approach should be used to develop all subgrid physics parameterizations in weather and climate models for developing parameterizations of subgrid motion.
How to cite: Bao, J.-W. and Michelson, S.: A paradigm shift for developing parameterizations of subgrid motion in NWP models, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-20640, https://doi.org/10.5194/egusphere-egu25-20640, 2025.
In general, data assimilation systems analyze values on regularly distributed grids according to the irregularly distributed observations. As long as a data assimilation system was established on a certain grid, it cannot adapt to another grid. In this study, the gridless method was introduced into the three-dimensional variation (3DVar) system. Compared with grid-based method, the gridless method uses discrete points for calculation and does not require grid division, thus being immune to grid distribution. Therefore, the data assimilation system based on gridless method can adapt to most model grid structures without the need to write new code. In the data assimilation system based on gridless method, the Cressman analysis technique is adopted as observation operator and the physical transformation matrix is handled using the Taylor expansion method. The idealized experiments based on the Rankine vortex demonstrate that the 3DVar system based on gridless method can handle structured grid, unstructured grid, and mixed (structured and unstructured) grid. Furthermore, the study showed that data assimilation can be performed simultaneously for different grid resolutions, resulting in higher consistency between the grids than when data assimilation is performed separately.
How to cite: Yuan, X.: Application of Gridless Method in Data Assimilation, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7684, https://doi.org/10.5194/egusphere-egu25-7684, 2025.
A 20-member WRF-based regional Ensemble Prediction System (WEPS) is operationally run at the Central Weather Administration (CWA) to provide up to 5-day ensemble forecasts over the East Asian region with a 15-km grid spacing and a 3-km nest centered over Taiwan. Since becoming operational in 2011, WEPS has been under continuous development that aims to improve the construction of its perturbations in initial conditions (IC), boundary conditions (BC), as well as model uncertainties due to numerical approximations and physical parameterizations. Among these uncertainties, construction of IC perturbations for WEPS is the focus of this study.
An initialization method named ensemble partial cycling (EnPC) is proposed for the WEPS. The EnPC method combines partial cycling data assimilation (DA) and the ensemble of DA approach with an additional blending procedure that merges large-scale global features with small-scale regional information, leveraging the DA efforts from the deterministic system of CWA. EnPC is compared with three other initialization methods that are popularly used for regional ensemble forecasting, including dynamic downscaling from a global EPS, Ensemble Adjustment Kalman Filter (EAKF) based regional ensemble DA, and a blended version of the two, the last of which is equivalent to the current operational configuration of WEPS. Among all 4 methods, EnPC is the only method that allows separate initializations for the parent and the nested domains while the initialization for the nested domain in the other three methods is simply a downscale-interpolation from the corresponding parent grid.
Several sets of WEPS experiments are conducted over a 5-week period, including five typhoons. EnPC-initialized WEPS forecasts are found to be comparable to the dynamically downscaled forecasts in many evaluation metrics and have more accurate near-surface forecasts over the first 12 h and better precipitation forecast discrimination ability for typhoon events. Compared to the EAKF and the blended methods, forecasts initialized from EnPC have overall smaller errors in most of the evaluation metrics by both deterministic and probabilistic measures and better spread-to-error ratios. As an alternative initialization method, EnPC not only adds some regional benefits on top of downscaling, but also shows some advantages over the operational method. With the planned retirement of EAKF and the anticipation of a more unified production suite at CWA, EnPC will replace the current operational method.
How to cite: Su, Y.-J., Wu, T.-C., Li, C.-H., Lien, G.-Y., and Hsieh, C.-H.: Initializing the Taiwan WRF-based Regional Ensemble Prediction System with an Ensemble Partial Cycling Strategy, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-2643, https://doi.org/10.5194/egusphere-egu25-2643, 2025.
Numerical Weather Prediction (NWP) models with a horizontal resolution of 2 km or finer need detailed information for estimating the initial state of the atmosphere. Ground-based remote-sensing instruments like Sodars, Doppler lidars and Profilers provide already meteorological information of the Atmospheric Boundary Layer (ABL). Although observational networks have been extended over the years, there are still gaps in data gathering particular on the finer scales. Therefore we have commenced research to investigate data from third parties. Here we focus on wind-information in the ABL from recreational Hot-air Balloon (HaB) flights. In the basic equipment of a HaB pilot there is a professional navigation device, which is compulsory for safety reasons. Similarly to routinely launched weather balloons, the Global Navigation Satellite System (GNSS)-data from consecutive positions and the elapsed time are the basis of the calculation of the horizontal wind vector. Yearly about 6000 flights take place in the Netherlands, mainly during the morning- and evening transition. As soon as the surface is covered with snow and when convection is strongly reduced, flights may also occur during the day. The HaB data are validated with observations from the meteorological site of Cabauw and we compare the HaB winds with mast data and other available observations like a RASS wind profiler. To explore the possibilities of this new type of wind observations in more complex terrain, we will present the results of an intriguing HaB flight in Austria, revealing a striking mountain-valley circulation. We also compare the HaB data with the results of an NWP model and we will report about a first attempt to assimilate the HaB data in a NWP model.
How to cite: de Bruijn, E. I. F. (.: Sensing the Wind with Hot-air Balloons and their Application in an NWP model, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-3280, https://doi.org/10.5194/egusphere-egu25-3280, 2025.
Super typhoons can pose severe threats to coastal cities. For instance, Typhoon Yagi caused hundreds of fatalities and extensive property damages while sweeping across southern China and Southeast Asia in 2024. Accurately predicting the dynamic motion and intensity of super typhoons in high resolution is critical for effective disaster prevention and mitigation. Over the past few decades, the accuracy of typhoon track predictions has notably improved due to the advancements in global numerical weather prediction models. However, their capability to assess typhoon intensity and dynamic structure is still limited by coarse spatiotemporal resolution. The application of physics-based regional models, such as the Weather Research and Forecasting (WRF) model, presents a promising solution to this challenge.
To simulate high-resolution wind fields during super typhoons through WRF, it is essential to determine optimal and robust physical parameterization schemes. In previous studies, sensitivity analysis is often carried out solely based on the error criteria related to typhoon track and intensity, which are inadequate for the performance evaluation of local wind simulation. Additionally, there is a lack of consistent physical parameterization settings for different super typhoons. Furthermore, due to the inherent biases and model errors, a dynamic bias correction strategy is required for local wind forecasting. To this end, we aim to develop an integrated framework in this study that combines typhoon simulation, multi-metric evaluation, and dynamic bias correction.
The super typhoons that have significantly impacted Hong Kong over the past two decades have been chosen as study cases, i.e. Hato, Mangkhut, and Saola. A series of numerical experiments were designed to assess the impact of various physical models. By comparing simulation results with best track data and field observations from the Hong Kong Observatory and the Shenzhen Meteorological Gradient Mast, the multi-metric evaluation method provides a comprehensive understanding of both global and local wind field simulation performance. The best-performing physical models were thereby identified, achieving consistent typhoon tracks (MAE < 30 km), relatively accurate typhoon intensity predictions (RMSE < 5 m/s), and highly correlated wind fields (r > 0.9) between simulation and observation results. To further reduce the effects of systematic biases, a dynamic linear bias correction strategy was introduced to adjust local wind predictions dynamically based on real-time observations. Given the time-evolving local wind data, the linear bias correction factor can reach convergence and provide reliable forecast corrections with a lead time of approximately 15 hours. The proposed framework shows great potential to enhance disaster warning systems and improve local wind prediction accuracy in typhoon-prone regions.
How to cite: Wang, F. and Wang, L.: A multi-metric evaluation and dynamic correction framework for local wind field prediction during super typhoons, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-3638, https://doi.org/10.5194/egusphere-egu25-3638, 2025.
The Korea Meteorological Administration(KMA) is producing an impact-based forecast data based on Multi-Model Ensemble(MME) system which integrates Unified Model(global, global ensemble, local, and local ensemble models), ECMWF(global and global ensemble models) and KIM(Korean Integrated Model) global model for heat waves (HW) and cold waves (CW). MME-based impact forecast(MEPS) determines the impact(safe, concern, caution, warning, alarm) by using the probability of occurrence of maximum feels-like temperature for HW and lowest temperature for CW in Korea.
The distribution from 93 MME members was converted to a GEV(Generalized Extreme Value) distribution until 2023, but there is a problem that only the daily temperature can be considered. Definition of HW and CW should take into account the 2-day duration/falling temperature compared to the previous day. Therefore, the probability calculation method was modified with the ratio of the number of members satisfying the HW and CW condition among all members and its performance was compared with the previous method.
Verification was conducted by evaluating how well impact-based forecast was matched to the observed impact in 177 regions about 1~9 forecast day. HW was verified for August and September 2023, and CW was verified for December 2023 and January 2024.
As a result, in the case of HW forecasting, impact-based forecast with new method showed a little better performance than previous method with GEV. New method has better Bias at concern(4-9day), warning(3-9day), alarm, and Equitable Thread Score at safe(2-9day), concern(2-9day), alarm. In addition, there are cases in which the definition of guidance is more accurately satisfied compared to the previous MEPS guidance, which was overestimated. Also, new method required much less calculation time than previous method. On the other hand, new method are not applied to CW MEPS due to its overall low performance.
It is presumed that the reason for the degration of performance of new method for CW is that the probability table for determining the impact in the probability distribution has not been tuned. If this table is optimized, it is expected that the performance can be improved in CW case, too.
How to cite: Yun, S., Shin, H.-C., Ha, J.-C., and Kim, D.: Improvement of Impact-based Forecast Using Multi Model Ensemble in 2024, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-13971, https://doi.org/10.5194/egusphere-egu25-13971, 2025.
The GNSS Radio Occultation (RO) Modeling Experiment (ROMEX) seeks to quantify the benefit of the increasing quantity of RO observations available for use in operational numerical weather prediction (NWP) systems and products. ROMEX includes participation from multiple operational NWP centers and NWP models, among them are NASA’s Global Earth Observing System (GEOS) model produced and run at the Global Modeling and Assimilation Office (GMAO). The design of the numerical experiments core to ROMEX include: 1) a control experiment that includes all the RO observations currently used operationally with the sole exception of those from commercial sources and 2) a ROMEX experiment that adds to the control over 25 thousand additional RO profiles per day from commercial RO providers, with both experiments run over the three-month period of September through November 2022. The ROMEX experiment greatly augments the relatively small subset of the currently available commercial RO profiles which have been purchased for routine use in operational NWP by the various NWP centers. While this smaller subset of commercial RO profiles currently used in operations has been shown to have a positive impact on NWP forecasts, the additional impact from the ROMEX RO dataset has yet to be determined and is the focus of ROMEX. Results from GEOS are presented, including the impact on both analyses and forecasts over the study period and statistics using the forecast sensitivity-based observation impact (FSOI) method.
How to cite: Murphy, Jr, M. J., Chattopadhyay, M., El Akkraoui, A., Gelaro, R., Anthes, R. A., and Jin, J.: Assimilation of High-volume commercial GNSS Radio Occultation (RO) Observations during ROMEX in NASA’s Global Earth Observing System, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-12320, https://doi.org/10.5194/egusphere-egu25-12320, 2025.
Posters virtual: Tue, 29 Apr, 14:00–15:45 | vPoster spot 5
EGU25-7852 | ECS | Posters virtual | VPS2
Extreme rainfall hotspots in India based on spatio-temporal variability of rainfall using unsupervised clustering techniquesTue, 29 Apr, 14:00–15:45 (CEST) | vP5.2
In past few decades there has been a noticeable increase in the frequency and intensity of extreme rainfall events (EREs) globally, including India. The Clausius-Clapeyron relationship explains how the warmer air can significantly hold more moisture. Hence, in present climate change scenario increasing temperature along with other factors can lead to further increase in EREs. Effective management strategeis in various sectors like disaster preparedness, smart-city planning, water quality, public health, agriculture planning, etc. can get improved, through proper understanding on the distribution and frequency of EREs. Keeping in mind the socio-economic impacts of EREs; this study aimed to identify the hotspot regions for EREs in India.
India is a country with vast spatio-temporal variability in rainfall pattern. Hence, this study implemented objective criteria to identify the spatio-temporal rainfall variability of EREs over four rainfall homogeneous regions for pre-monsoon, monsoon and post-monsoon seasons. Based on frequency distribution of daily accumulated rainfall, suitable rainfall threshold values for defining EREs are identified for each homogeneous region and each season. These threshold values vary region-wise as well as season-wise. Distribution of EREs show interannual as well as seasonal variability.
Clustering algorithms, popular unsupervised Machine Learning (ML) techniques, are handy tools to identify hotspots of extreme rainfall regions with similar spatial variability. To understand the ERE distribution and to identify rainfall hotspots based on long term daily gridded rainfall data, this study implemented K-means clustering and DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithms. Comparative area distribution study between K-means and DBSCAN clustering help to identify the EREs hotspots in India. Overall, the K-means method shows more scattered hotspots compared to DBSCAN method, which are further validated using Davies-Boulding Index (DBI), Silhouette score, Calinski-Harabasz (CH) score and Dunn's Index. These score analysis methods serve as potential tools to support the clustering validation method. In addition to the area distribution, this study has addressed the temporal variability of the EREs hotspots. ST-OPTICS ( Spatio-Temporal Ordering Points to Identify the Clustering Structure) algorithm results clustering of hotspots based on their spatial and temporal similarity. This study shows that ML algorithms prove to be promising techniques for detecting and analyzing spatial as well as temporal variability of EREs hotspots which is effective for future management practice in various sectors.
Keywords: Extreme Rainfall Events; DBSCAN Clustering; K-Means Clustering; ST-OPTICS.
How to cite: Putatunda, I. and Vasudevan, R.: Extreme rainfall hotspots in India based on spatio-temporal variability of rainfall using unsupervised clustering techniques, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7852, https://doi.org/10.5194/egusphere-egu25-7852, 2025.
EGU25-125 | ECS | Posters virtual | VPS2
State and Stochastic Parameters Estimation with Combined Ensemble Kalman and Particle FiltersTue, 29 Apr, 14:00–15:45 (CEST) | vP5.3
Quantifying uncertainties is a key aspect of data assimilation systems since it has a large impact on the quality of the forecasts and analyses. Sequential data assimilation algorithms, such as the Ensemble Kalman Filter (EnKF), describe the model and observation errors as additive Gaussian noises and use both inflation and localization to avoid filter degeneracy and compensate for misspecifications. This introduces different stochastic parameters which need to be carefully estimated in order to get a reliable estimate of the latent state of the system. A classical approach to estimate unknown parameters in data assimilation consists in using state-augmentation, where the unknown parameters are included in the latent space and are updated at each iteration of the EnKF. However, it is well-known that this approach is not efficient to estimate stochastic parameters because of the complex (non-Gaussian and non-linear) relationship between the observations and the stochastic parameters which can not be handled by the EnKF. A natural alternative for non-Gaussian and non-linear state-space models is to use a particle filter (PF), but this algorithm fails to estimate high-dimensional systems due to the curse of dimensionality. The strengths of these two methods are gathered in the proposed algorithm, where the PF first generates the particles that estimate the stochastic parameters, then using the mean particle the EnKF generates the members that estimate the geophysical variables. This generic method is first detailed for the estimation of parameters related to the model or observation error and then for the joint estimation of inflation and localization parameters. Numerical experiments are performed using the Lorenz-96 model to compare our approach with state-of-the-art methods. The results show the ability of the new method to retrieve the geophysical state and to estimate online time-dependent stochastic parameters. The algorithm can be easily built from an existing EnKF with low additional cost and without further running the dynamical model.
How to cite: Guillot, J.: State and Stochastic Parameters Estimation with Combined Ensemble Kalman and Particle Filters, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-125, https://doi.org/10.5194/egusphere-egu25-125, 2025.
