Land degradation estimation from global and national satellite based datasets within UN program

Abstract

In this paper, we investigate global and national level datasets, used to estimate trends in land cover and in land productivity in Ukraine within Land Degradation Neutrality (LDN) Target Setting Programme of United Nations Convention to Combat Desertification (UNCCD). To assess land cover changes, the ESA CCI-LC 2000 and 2010 epochs are used, focusing on changes between the 6 main land cover categories. The JRC's Land Productivity Dynamics (LPD) dataset is used as default source for land productivity data. The LPD dataset is derived from a 15-year time series (1999 to 2013) of global NDVI observations composited in 10-day intervals at a spatial resolution of 1 km. To improve the resolution of satellite based products at national level we use regional land cover datasets for Ukraine developed by Space Research Institute (SRI) and available for 1990, 2000 and 2010 respectively. They are developed according to international standards (LUCAS classification) and are comparable with ESA CCI-LC in terms of class nomenclature and notations. These land cover maps at 30 m spatial resolution have been developed for the whole territory of Ukraine based on the Landsat-4/5/7 images for three decades, namely 1990s, 2000s and 2010s within FP-7 SIGMA Project. They are based on our Machine Learning approach for classification of time series of satellite data.