Nowadays, weather prediction is based on numerical weather prediction (NWP) models to produce an ensemble of forecasts. Despite of large improvements over the last few decades, they still tend to exhibit systematic bias and dispersion errors. Consequently, these forecasts may be improved by statistical postprocessing. This work proposes an extension of the ensemble model output statistics (EMOS) method in a time series framework. Besides of taking account of seasonality and trend in the location and scale parameter of the predictive distribution, the autoregressive process in the mean forecast errors or the standardized forecast errors is considered. The models can be further extended by allowing generalized autoregressive conditional heteroscedasticity (GARCH). Last but not least, it is outlined how to use these models for arbitrary forecast horizons. To illustrate the performance of the suggested EMOS models in time series fashion, we present a case study for the postprocessing of 2 m surface temperature forecasts using five different lead times and a set of observation stations in Germany. The results indicate that the time series EMOS extensions are able to significantly outperform the benchmark EMOS and autoregressive adjusted EMOS (AR-EMOS) in most of the lead time-station cases. To complement this article, our method is accompanied by an R-package called tsEMOS.

翻译：现今，天气预报基于数值天气预报（NWP）模型生成集合预报。尽管在过去几十年间取得了显著进步，但其仍普遍存在系统性偏差和离散度误差。因此，可通过统计后处理方法来改进这些预报。本研究提出了一种在时间序列框架下对集合模型输出统计（EMOS）方法的扩展。除考虑预测分布的位置参数和尺度参数中的季节性与趋势外，还引入了平均预报误差或标准化预报误差的自回归过程。通过允许广义自回归条件异方差（GARCH）效应，可进一步扩展模型。此外，本文概述了如何将这些模型应用于任意预报时效。为展示所提出的时间序列型EMOS模型的性能，我们以德国多个观测站点的2米地表温度预报后处理为例，采用了五种不同预报提前期进行案例研究。结果表明，在大多数预报提前期-站点组合中，时间序列型EMOS扩展方法能显著优于基准EMOS和自回归调整的EMOS（AR-EMOS）。为补充本文内容，我们的方法还配有名为tsEMOS的R语言包。