Comparison of downscaling methods for mean and extreme precipitation in Senegal

M.A Sarr,O Seidou,S El Adlouni,Y Tramblay

doi:10.1016/j.ejrh.2015.06.005

Abstract

Abstract Study region The study considers six precipitation stations located in Senegal, West Africa. Senegal is located in the Sahel, an area that is threatened by climate variability and change. Both droughts and extreme rainfall have been an issue in recent years. Study focus Two different statistical downscaling techniques were applied to the outputs of four regional climate models at six selected precipitation stations in Senegal. First, the delta-change method was applied to the mean annual precipitation as well as the 5, 10, 20, 50 and 100-year return period daily precipitation events. Second, a quantile–quantile transformation (QQ) was used to downscale the monthly distributions of precipitation simulated by regional climate models (RCMs). The 5, 10, 20, 50 and 100-year daily precipitation events were afterward calculated. All extreme events were calculated assuming that maximum annual daily precipitations follow the generalized extreme value (GEV) distribution. The two-sided Kolmogorov–Smirnov (KS) test was finally used to assess the performance of the quantile–quantile transformation as well as the GEV distribution fit for the annual maximum daily precipitation. New hydrological insights for the region Results show that the two downscaling techniques generally agree on the direction of the change when applied to the outputs of same RCM, but some cases lead to very different projections of the direction and magnitude of the change. Projected changes indicate a decline in mean precipitation except for one RCM over one region in Senegal. Projected changes in extreme precipitations are not consistent across stations and return periods. The choice of the downscaling technique has more effect on the estimation of extreme daily precipitations of return period equal or greater than ten years than the choice of the climate models.

Highlights

Study region: The study considers six precipitation stations located in Senegal, West Africa
Statistical downscaling results could be strongly impacted by data errors if the predictor–predictand relationship does not consider important climatic features, it generally focuses on precipitation and/or temperature (e.g. Di Vittorio and Miller, 2013)
This study presents the first evaluation of the projected changes in precipitation in Senegal with regional climate models (RCMs) and two downscaling techniques

Summary

Introduction

Study region: The study considers six precipitation stations located in Senegal, West Africa. Study focus: Two different statistical downscaling techniques were applied to the outputs of four regional climate models at six selected precipitation stations in Senegal. The delta-change method was applied to the mean annual precipitation as well as the 5, 10, 20, 50 and 100-year return period daily precipitation events. Statistical downscaling results could be strongly impacted by data errors if the predictor–predictand relationship does not consider important climatic features (such as large scale circulation and local characteristics of the study area), it generally focuses on precipitation (in the present paper) and/or temperature The objective of this paper is to compare the impact of two downscaling techniques (the delta-change and quantile–quantile transformation) on projected changes, in average and extreme precipitation of the 2000–2050 period at six locations across Senegal (Fig. 1).

Objectives

Methods

Results

Discussion

Conclusion