OPTIMISING VEGETATION-INPUT FOR DROUGHT ASSESSMENT WITH SENTINEL-2A DATA

Abstract

As a consequence of climate change, in some regions, more intense rain showers go hand in hand with longer dry periods. The subsequent more and more severe droughts can have devastating effects on many economic and social sectors. Therefore, it is necessary to be able to predict and assess the consequences of these droughts on a local scale, in order to develop policies to cope. Drought assessment needs a lot of detailed and accurate input-data, such as land use, land cover, soil moisture, vegetation, evapotranspiration, etc., often obtained by continuous earth monitoring by satellites. Satellite images are generally converted into indices, of which the Normalized Difference Vegetation Index (NDVI) is one of the most widely used. It was developed for use with Landsat imagery and allows for the classification of satellite images for land use and the assessment of the vegetation�s vitality. In this research, a new composite index is presented and compared to the NDVI to be used with Sentinel-2A imagery, having higher resolution and more spectral bands than Landsat. This new composite index can be used to detect water and vegetation. Test results show that this newly developed composite index achieves a better accuracy through Support Vector Machine (SVM) classification than the widely used NDVI. Although further validation is necessary, the results promise a possible amelioration of vegetation related input data for drought assessment and management.