Suitability of remote sensing for dust storm detection and quantification in Sahelian Africa

Abstract

Wind erosion is one of the main soil degradation processes that play a role in desertification in Sahelian Africa. Development in remote sensing techniques during the last decade has increased its potential for desertification assessment. The purpose of this study was to develop an empirical method to quantify wind erosion with satellite imagery for dryland areas in Sahelian Africa. Firstly, dust storms in the Sahel region were identified on the available InfraRed Imager of the Meteosat Second Generation instrument (SEVIRI-MSG) images in 2013. Additionally, the storms had to be recorded by at least one Aerosol Robotic Network (AERONET) observation station in the Sahel. Secondly, the relation between Aerosol Optical Thickness (AOT) and dust concentration at ground level was determined. Since dust concentration measurements were only available for 2006, AOT was used a proxy for dust concentration. Thirdly, the statistical relation between brightness temperature (BT) and AOT measured in-situ was determined and eventually the relation between BT and dust concentration was obtained. Finally, the empirical method was applied on two SEVIRI-MSG images. Two linear regression equations were developed to describe dust concentration in months February-March of the dry season at two measurement stations in the Sahel. Poor relations were found for the early rainy season months. Application of the obtained equations on two SEVIRI-MSG images revealed unrealistic dust concentrations, which could not be validated with currently available ground data. Another important conclusion was that the dust storms causing most wind erosion problems, which occur in the early rainy season, are often not detectable on the SEVIRI-MSG images because their signal is blocked by the cloud of the thunderstorms above it. Therefore, quantification of annual wind erosion with the obtained equations without inclusion of these events will lead to significant underestimation of sediment transport.