Probabilistic temperature forecasting with a heteroscedastic autoregressive ensemble postprocessing model

Abstract

Weather prediction today is performed with numerical weather prediction (NWP) models. These are deterministic simulation models describing the dynamics of the atmosphere, and evolving the current conditions forward in time to obtain a prediction for future atmospheric states. To account for uncertainty in NWP models it has become common practice to employ ensembles of NWP forecasts. However, NWP ensembles often exhibit forecast biases and dispersion errors, thus require statistical postprocessing to improve reliability of the ensemble forecasts. This work proposes an extension of a recently developed postprocessing model utilizing autoregressive information present in the forecast error of the raw ensemble members. The original approach is modified to let the variance parameter depend on the ensemble spread, yielding a two‐fold heteroscedastic model. Furthermore, an additional high‐resolution forecast is included into the postprocessing model, yielding improved predictive performance. Finally, it is outlined how the autoregressive model can be utilized to postprocess ensemble forecasts with higher forecast horizons, without the necessity of making fundamental changes to the original model. We accompany the new methodology by an implementation within the R package ensAR to make our method available for other researchers working in this area. To illustrate the performance of the heteroscedastic extension of the autoregressive model, and its use for higher forecast horizons we present a case‐study for a dataset containing 12 years of temperature forecasts and observations over Germany. The case‐study indicates that the autoregressive model yields particularly strong improvements for forecast horizons beyond 24 h.

The EUPPBench postprocessing benchmark dataset v1.0

“…The adjusted ensemble forecasts are employed to estimate the mean and variance parameter of the predictive Gaussian distribution. Estimation of the predictive variance was further refined in Möller and Groß (2020). The method is implemented in the R package ensAR (Groß and Möller, 2019).…”

CRPS-based online learning for nonlinear probabilistic forecast combination

“…SLP adjusts the spread of the component forecasts and can consequently mitigate-to a certain extent-overdispersion caused by linearly combining calibrated forecasts. Möller and Groß (2020) apply SLP to post-processed temperature forecasts issued by the European Center for Medium-range Weather Forecasts (ECMWF) ensemble prediction system and show that it effectively lowers CRPS compared to the component forecasts. However, a limitation of SLP is that the method fails to be flexibly dispersive, which is to say that it is unable to sufficiently adjust the spread to produce neutrally dispersed forecasts, especially when the component forecasts are neutrally dispersed or underdispersed (Gneiting and Ranjan, 2013).…”

Which precipitation forecasts to use? Deterministic versus coarser‐resolution ensemble NWP models

“…It is generally recognised that raw forecasts produced by NWP models are not suitable for direct use (Scherrer et al ., 2004; Wu et al ., 2019; Möller and Groß, 2020). Raw deterministic and ensemble forecasts are often biased and can even be less skilful than naïve climatology forecasts (Li et al ., 2017), especially at long lead times.…”