A5086897062- Meteorological Phenomena and Simulations
- Climate variability and models
- Forecasting Techniques and Applications
- Hydrology and Drought Analysis
- Precipitation Measurement and Analysis
- Cryospheric studies and observations
- Energy Load and Power Forecasting
- Data Analysis with R
- Financial Risk and Volatility Modeling
- Computational Physics and Python Applications
- Hydrological Forecasting Using AI
- Statistical Methods and Inference
- Distributed and Parallel Computing Systems
- Geophysics and Gravity Measurements
- Hydrology and Watershed Management Studies
- Landslides and related hazards
- Flood Risk Assessment and Management
- Atmospheric and Environmental Gas Dynamics
- Risk and Portfolio Optimization
- Remote Sensing in Agriculture
- Engineering Technology and Methodologies
- Advanced Statistical Methods and Models
- Calibration and Measurement Techniques
- Wind and Air Flow Studies
- Advanced Data Processing Techniques
Météo-France
2016-2025
Centre National de la Recherche Scientifique
2016-2025
Centre National de Recherches Météorologiques
2016-2025
Université de Toulouse
2020-2025
École Nationale de la Météorologie
2022-2023
Laboratoire des Sciences du Climat et de l'Environnement
2016
Statistical postprocessing techniques are nowadays key components of the forecasting suites in many National Meteorological Services (NMS), with for most them, objective correcting impact different types errors on forecasts. The final aim is to provide optimal, automated, seamless forecasts end users. Many now flourishing statistical, meteorological, climatological, hydrological, and engineering communities. methods range complexity from simple bias corrections very sophisticated...
Abstract Ensembles used for probabilistic weather forecasting tend to be biased and underdispersive. This paper proposes a statistical method postprocessing ensembles based on quantile regression forests (QRF), generalization of random regression. does not fit parametric probability density function (PDF) like in ensemble model output statistics (EMOS) but provides an estimation desired quantiles. is nonparametric approach that eliminates any assumption the variable subject calibration. can...
Rainfall ensemble forecasts have to be skillful for both low precipitation and extreme events. We present statistical post-processing methods based on Quantile Regression Forests (QRF) Gradient (GF) with a parametric extension heavy-tailed distributions. Our goal is improve quality all types of events, included, subject good overall performance. hybrid proposed are applied daily 51-h 6-h accumulated from 2012 2015 over France using the M{\'e}t{\'e}o-France prediction system called PEARP....
Abstract. Statistical post-processing of ensemble forecasts, from simple linear regressions to more sophisticated techniques, is now a well-known procedure for correcting biased and poorly dispersed weather predictions. However, practical applications in national services are still their infancy compared deterministic post-processing. This paper presents two different using machine learning at an industrial scale. The first station-based surface temperature subsequent interpolation grid...
Abstract. Statistical postprocessing of medium-range weather forecasts is an important component modern forecasting systems. Since the beginning data science, numerous new methods have been proposed, complementing already very diverse field. However, one questions that frequently arises when considering different in framework implementing operational relative performance for a given specific task. It particularly challenging to find or construct common comprehensive dataset can be used...
Proper scoring rules are an essential tool to assess the predictive performance of probabilistic forecasts. However, propriety alone does not ensure informative characterization and it is recommended compare forecasts using multiple rules. With that in mind, interpretable providing complementary information necessary. We formalize a framework based on aggregation transformation build multivariate proper Aggregation-and-transformation-based able target specific features forecasts; which...
A point-forecast is defined as a single-value forecast expressed in the unit of variable interest. A deterministic for 2m temperature at Vienna tomorrow is point-forecast. Point-forecasts are required by some users and various applications. When an ensemble prediction system hand, point-forecast can take form of a distribution functional such mean or quantile. In this context, we introduce new type based on concept crossing-point (Ben...
Proper scoring rules are an essential tool to assess the predictive performance of probabilistic forecasts. However, propriety alone does not ensure informative characterization and it is recommended compare forecasts using multiple rules. With that in mind, interpretable providing complementary information necessary. We formalize a framework based on aggregation transformation build multivariate proper Aggregation-and-transformation-based can target application-specific features forecasts,...
How much rain can we expect in Toulouse on Wednesday next week? It is impossible to provide a precise and definitive answer this question due the limited predictability of atmosphere. So ideally, forecast would be probabilistic, that expressed form probability of, say, having at least some rain. However, for users applications, an mm per 24h needed. A so-called point-forecast output single deterministic model. But with ensemble forecasts hand, how summarize optimally information into...
How much rain can we expect in Toulouse on Wednesday next week? It is impossible to provide a precise and definitive answer this question due the limited predictability of atmosphere. So ideally, forecast would be probabilistic, that expressed form probability of, say, having at least some rain. However, for users applications, an mm per 24h needed. A so-called point-forecast output single deterministic model. But with ensemble forecasts hand, how summarize optimally information into...
Verifying probabilistic forecasts for extreme events is a highly active research area because popular media and public opinions are naturally focused on events, biased conclusions readily made. In this context, classical verification methods tailored such as thresholded weighted scoring rules, have undesirable properties that cannot be mitigated, the well-known continuous ranked probability score (CRPS) no exception. paper, we define formal framework assessing behavior of forecast evaluation...
Abstract. Height of new snow (HN) forecasts help to prevent critical failures infrastructures in mountain areas, e.g. transport networks and ski resorts. The French national meteorological service, Météo-France, operates a probabilistic forecasting system based on ensemble detailed snowpack model provide ensembles HN forecasts. These are, however, biased underdispersed. As for many weather variables, post-processing methods can be used alleviate these drawbacks obtain meaningful 1 4 d In...
The theoretical advances on the properties of scoring rules over past decades have broaden use in probabilistic forecasting. In meteorological forecasting, statistical postprocessing techniques are essential to improve forecasts made by deterministic physical models. Numerous state-of-the-art based distributional regression evaluated with Continuous Ranked Probability Score (CRPS). However, such minimization CRPS mostly considered unconditional framework (i.e. without covariables) and...
Abstract. Statistical Postprocessing of medium-range weather forecasts is an important component modern forecasting systems. Since the beginning data science, numerous new postprocessing methods have been proposed, complementing already very diverse field. However, one questions that frequently arises when considering different in framework implementing operational relative performance for a given specific task. It particularly challenging to find or construct common comprehensive dataset...
Abstract. Statistical post-processing of ensemble forecasts, from simple linear regressions to more sophisticated techniques, is now a well-known procedure in order correct biased and misdispersed weather predictions. However, practical applications National Weather Services still its infancy compared deterministic post-processing. This paper presents two different using machine learning at an industrial scale. The first station-based surface temperature medium resolution system. second...
The implementation of statistical postprocessing ensemble forecasts is increasingly developed among national weather services. so-called Ensemble Model Output Statistics (EMOS) method, which consists generating a given distribution whose parameters depend on the raw ensemble, leads to significant improvements in forecast performance for low computational cost, and so particularly appealing reduced computing architectures. However, choice parametric has be sufficiently consistent as not lose...
Statistical postprocessing of forecasts from numerical weather prediction systems is an important component modern forecasting systems. A growing variety methods has been proposed, but a comprehensive, community-driven comparison their relative performance yet to be established. Important reasons for this lack include the absence fair intercomparison protocol, and, difficulty constructing common comprehensive dataset that can used perform such intercomparison. Here we introduce first version...
Verifying probabilistic forecasts for extreme events is a highly active research area because popular media and public opinions are naturally focused on events, biased conclusions readily made. In this context, classical verification methods tailored such as thresholded weighted scoring rules, have undesirable properties that cannot be mitigated, the well-known continuous ranked probability score (CRPS) no exception. Here, we define formal framework assessing behavior of forecast evaluation...
The theoretical advances in the properties of scoring rules over past decades have broadened use probabilistic forecasting. In meteorological forecasting, statistical postprocessing techniques are essential to improve forecasts made by deterministic physical models. Numerous state-of-the-art based on distributional regression evaluated with continuous ranked probability score (CRPS). However, such evaluations CRPS solely considered unconditional framework (i.e. without covariates) and...