Earth Observation-Based Land Degradation Mapping and Prediction: The Moroccan Test Region Case Study

Abstract

This paper presents an innovative technique for remote sensing-based mapping and predicting land degradation (LD) in the European frontier regions. The mapping is carried out by geospatial data fusion of relevant biophysical indicators of land degradation. The indicators are represented by existing higher level satellite-derived data products of Earth observation (EO). The foundation of our approach to land degradation assessment is not the current state analysis, but the extraction of degradation conditions from the satellite products data cube. Data analysis and fusion are performed using an innovative probabilistic model based on the statistics of these indicators. The prediction is provided by annual time series analysis using the grey model (GM) (1,1). Conducted field observations within the study area confirm the validity of the proposed technique. The Kendall correlation coefficient value between remotely acquired and ground-based classes of land degradation is 0.816. The developed technique is being implemented in the cloud service of the land degradation early warning system's (EWS) prototype.