Statistical Forecasts for the Occurrence of Precipitation Outperform Global Models over Northern Tropical Africa. Issue 3 (2nd February 2021)

• Record Type: Journal Article
• Authors: Vogel, Peter; Knippertz, Peter; Gneiting, Tilmann; Fink, Andreas H.; Klar, Manuel; Schlueter, Andreas
• Is Part Of: Geophysical research letters. Volume 48:Issue 3(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-02-02
• Subjects: forecasting -- logistic regression -- postprocessing -- precipitation -- tropical convection -- West Africa; Geophysics -- Periodicals; Planets -- Periodicals; Lunar geology -- Periodicals; 550
• DOI: 10.1029/2020GL091022
• Languages: English
• ISSNs: 0094-8276

Abstract

Abstract: Short‐term global ensemble predictions of rainfall currently have no skill over northern tropical Africa when compared to simple climatology‐based forecasts, even after sophisticated statistical postprocessing. Here, we demonstrate that 1‐day statistical forecasts for the probability of precipitation occurrence based on a simple logistic regression model have considerable potential for improvement. The new approach we present here relies on gridded rainfall estimates from the Tropical Rainfall Measuring Mission for July‐September 1998–2017 and uses rainfall amounts from the pixels that show the highest positive and negative correlations on the previous two days as input. Forecasts using this model are reliable and have a higher resolution and better skill than climatology‐based forecasts. The good performance is related to westward propagating African easterly waves and embedded mesoscale convective systems. The statistical model is outmatched by the postprocessed dynamical forecast in the dry outer tropics only, where extratropical influences are important. Plain Language Summary: Forecasts of precipitation for the next few days based on state‐of‐the‐art weather models are currently inaccurate over northern tropical Africa, even after systematic forecast errors are corrected statistically. In this paper, we show that we can use rainfall observations from the previous 2 days to improve 1‐day predictions of precipitation occurrence. Such an approach works well over this region, as rainfall systems tend to travel from the east to the west organized by flow patterns several kilometers above the ground, called African easterly waves. This statistical forecast model requires training over a longer time period (here 19 years) to establish robust relationships on which future predictions can be based. The input data employed are gridded rainfall estimates based on satellite data for the African summer monsoon in July to September. The new method outperforms all other methods currently available on a day‐to‐day basis over the region, except for the dry outer tropics, where influences from midlatitudes, which are better captured by weather models, become more important. Key Points: Raw and statistically postprocessed global ensemble forecasts fail to predict West African rainfall occurrence A logistic regression model using observations from preceding days outperforms all other types of forecasts The skill of the statistical model is mainly related to propagating African easterly waves and mesoscale convective systems