Comparison of non-homogeneous regression models for probabilistic wind speed forecasting
In weather forecasting, non-homogeneous regression (NR) is used to statistically post-process forecast ensembles in order to obtain calibrated predictive distributions. For wind speed forecasts, the regression model is given by a truncated normal (TN) distribution, where location and spread derive f...
Main Authors: | , |
---|---|
Format: | Article |
Language: | English |
Published: |
Taylor & Francis Group
2013-11-01
|
Series: | Tellus: Series A, Dynamic Meteorology and Oceanography |
Subjects: | |
Online Access: | http://www.tellusa.net/index.php/tellusa/article/download/21206/pdf_1 |
id |
doaj-99f5eed9fcd543a0b5105ed088951c32 |
---|---|
record_format |
Article |
spelling |
doaj-99f5eed9fcd543a0b5105ed088951c322020-11-25T01:46:35ZengTaylor & Francis GroupTellus: Series A, Dynamic Meteorology and Oceanography0280-64951600-08702013-11-0165011310.3402/tellusa.v65i0.21206Comparison of non-homogeneous regression models for probabilistic wind speed forecastingSebastian LerchThordis L. ThorarinsdottirIn weather forecasting, non-homogeneous regression (NR) is used to statistically post-process forecast ensembles in order to obtain calibrated predictive distributions. For wind speed forecasts, the regression model is given by a truncated normal (TN) distribution, where location and spread derive from the ensemble. This article proposes two alternative approaches which utilise the generalised extreme value (GEV) distribution. A direct alternative to the TN regression is to apply a predictive distribution from the GEV family, while a regime-switching approach based on the median of the forecast ensemble incorporates both distributions. In a case study on daily maximum wind speed over Germany with the forecast ensemble from the European Centre for Medium-Range Weather Forecasts (ECMWF), all three approaches significantly improve the calibration as well as the overall skill of the raw ensemble with the regime-switching approach showing the highest skill in the upper tail.http://www.tellusa.net/index.php/tellusa/article/download/21206/pdf_1ensemble post-processingnon-homogeneous regressionpredictive distributionprobabilistic forecastingweather forecastingwind speed |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Sebastian Lerch Thordis L. Thorarinsdottir |
spellingShingle |
Sebastian Lerch Thordis L. Thorarinsdottir Comparison of non-homogeneous regression models for probabilistic wind speed forecasting Tellus: Series A, Dynamic Meteorology and Oceanography ensemble post-processing non-homogeneous regression predictive distribution probabilistic forecasting weather forecasting wind speed |
author_facet |
Sebastian Lerch Thordis L. Thorarinsdottir |
author_sort |
Sebastian Lerch |
title |
Comparison of non-homogeneous regression models for probabilistic wind speed forecasting |
title_short |
Comparison of non-homogeneous regression models for probabilistic wind speed forecasting |
title_full |
Comparison of non-homogeneous regression models for probabilistic wind speed forecasting |
title_fullStr |
Comparison of non-homogeneous regression models for probabilistic wind speed forecasting |
title_full_unstemmed |
Comparison of non-homogeneous regression models for probabilistic wind speed forecasting |
title_sort |
comparison of non-homogeneous regression models for probabilistic wind speed forecasting |
publisher |
Taylor & Francis Group |
series |
Tellus: Series A, Dynamic Meteorology and Oceanography |
issn |
0280-6495 1600-0870 |
publishDate |
2013-11-01 |
description |
In weather forecasting, non-homogeneous regression (NR) is used to statistically post-process forecast ensembles in order to obtain calibrated predictive distributions. For wind speed forecasts, the regression model is given by a truncated normal (TN) distribution, where location and spread derive from the ensemble. This article proposes two alternative approaches which utilise the generalised extreme value (GEV) distribution. A direct alternative to the TN regression is to apply a predictive distribution from the GEV family, while a regime-switching approach based on the median of the forecast ensemble incorporates both distributions. In a case study on daily maximum wind speed over Germany with the forecast ensemble from the European Centre for Medium-Range Weather Forecasts (ECMWF), all three approaches significantly improve the calibration as well as the overall skill of the raw ensemble with the regime-switching approach showing the highest skill in the upper tail. |
topic |
ensemble post-processing non-homogeneous regression predictive distribution probabilistic forecasting weather forecasting wind speed |
url |
http://www.tellusa.net/index.php/tellusa/article/download/21206/pdf_1 |
work_keys_str_mv |
AT sebastianlerch comparisonofnonhomogeneousregressionmodelsforprobabilisticwindspeedforecasting AT thordislthorarinsdottir comparisonofnonhomogeneousregressionmodelsforprobabilisticwindspeedforecasting |
_version_ |
1725018554027212800 |