arXiv subjects

Camille Le Coz

Publications and source records attributed to Camille Le Coz.

3 recordsLinked to original sources

arXiv2026

A barycenter-based approach for the multi-model ensembling of subseasonal forecasts

Ensemble forecasts and their combination are examined from the perspective of probability spaces. Manipulating ensemble forecasts as discrete probability distributions, multi-model ensemble (MME) forecasts are reformulated as barycenters of these distributions. We consider two barycenters, each defined with respect to a different distance metric: the L2 barycenter, which correspond to the traditional pooling method, and the Wasserstein barycenter, which better preserves certain geometric properties of the input ensemble distributions. As a proof of concept, we apply the L2 and Wasserstein barycenters to the combination of four models from the Subseasonal to Seasonal (S2S) prediction project database. Their performance is evaluated for the prediction of weekly 2m temperature, 10m wind speed, and 500hPa geopotential height over European winters. By construction, both barycenter-based MMEs have the same ensemble mean, but differ in their representation of the forecast uncertainty. Notably, the L2 barycenter has a larger ensemble spread, making it more prone to under-confidence. While both methods perform similarly on average in terms of the Continuous Ranked Probability Score (CRPS), the Wasserstein barycenter performs better more frequently.

stat.AP↗

arXiv2025

Quantile Regression, Variational Autoencoders, and Diffusion Models for Uncertainty Quantification: A Spatial Analysis of Sub-seasonal Wind Speed Prediction

This study aims to improve the spatial representation of uncertainties when regressing surface wind speeds from large-scale atmospheric predictors for sub-seasonal forecasting. Sub-seasonal forecasting often relies on large-scale atmospheric predictors such as 500 hPa geopotential height (Z500), which exhibit higher predictability than surface variables and can be downscaled to obtain more localised information. Previous work by Tian et al. (2024) demonstrated that stochastic perturbations based on model residuals can improve ensemble dispersion representation in statistical downscaling frameworks, but this method fails to represent spatial correlations and physical consistency adequately. More sophisticated approaches are needed to capture the complex relationships between large-scale predictors and local-scale predictands while maintaining physical consistency. Probabilistic deep learning models offer promising solutions for capturing complex spatial dependencies. This study evaluates three probabilistic methods with distinct uncertainty quantification mechanisms: Quantile Regression Neural Network that directly models distribution quantiles, Variational Autoencoders that leverage latent space sampling, and Diffusion Models that utilise iterative denoising. These models are trained on ERA5 reanalysis data and applied to ECMWF sub-seasonal hindcasts to regress probabilistic wind speed ensembles. Our results show that probabilistic downscaling approaches provide more realistic spatial uncertainty representations compared to simpler stochastic methods, with each probabilistic model offering different strengths in terms of ensemble dispersion, deterministic skill, and physical consistency. These findings establish probabilistic downscaling as an effective enhancement to operational sub-seasonal wind forecasts for renewable energy planning and risk assessment.

cs.LG↗

arXiv2025

Improving sub-seasonal wind-speed forecasts in Europe with a non-linear model

Sub-seasonal wind speed forecasts provide valuable guidance for wind power system planning and operations, yet the forecast skills of surface winds decrease sharply after two weeks. However, large-scale variables exhibit greater predictability on this time scale. This study explores the potential of leveraging non-linear relationships between 500 hPa geopotential height (Z500) and surface wind speed to improve sub-seasonal wind speed forecast skills in Europe. Our proposed framework uses a Multiple Linear Regression (MLR) or a Convolutional Neural Network (CNN) to regress surface wind speed from Z500. Evaluations on ERA5 reanalysis indicate that the CNN performs better due to its non-linearity. Applying these models to sub-seasonal forecasts from the European Centre for Medium-Range Weather Forecasts, various verification metrics demonstrate the advantages of non-linearity. Yet, this is partly explained by the fact that these statistical models are under-dispersive since they explain only a fraction of the target variable variance. Introducing stochastic perturbations to represent the stochasticity of the unexplained part from the signal helps compensate for this issue. Results show that the perturbed CNN performs better than the perturbed MLR only in the first weeks, while the perturbed MLR's performance converges towards that of the perturbed CNN after two weeks. The study finds that introducing stochastic perturbations can address the issue of insufficient spread in these statistical models, with improvements from the non-linearity varying with the lead time of the forecasts.

cs.LG↗