Development of a satellite-based dynamic regional vegetation model for the Drâa catchment

Abstract

Analysing and modelling land cover dynamic of the vegetation under a changing hydrological cycle inside the semi-arid area resulting from the global climate change are a difficult task. It is important to be able to understand and predict the characteristics and availability of vegetation as result of the global climate. This study was carried out inside the upper and middle Drâa catchment in south Morocco, focusing on the natural vegetation outside rural and agricultural areas. Development of a dynamic regional land cover model is traditionally driven by site specific plant growing parameters or by spatial information from remote sensing (e.g. NDVI). By scaling both approaches to a regional level plant activity can be analysed with the MODIS sensor and interpreted by local measurements. By using signal processing techniques, a double regression approach was developed and tested under the conditions of temporal trends and performance parameters. Completed by a regional adopted vegetation model, important productivity parameters could be extracted. This semi-automatic approach is realized in the conceptual model MOVEG Drâa, bringing together remote sensing, meteorological and other data and techniques. An extensive phenological database was built up by integrating Terra MODIS NDVI time series (2000 until 2008), a vegetation monitoring network and 10 years of meteorological measurements. In order to validate the method a comprehensive field measurement along a North-South transect was established. The results show that a robust point conclusion on vegetation trends and parameters on a statistical significant level is possible. Based on these findings a spatial explicit output was realized by a spatial extrapolation technique considering the annual and intra-annual vegetation trends. Based on the IPCC Scenarios (A1B and B1) a forecast of vegetation activity and productivity was implemented until 2050.<br /> MOVEG DRAA is an improvement to the hitherto state of unknown atmospheric-vegetation-relationship for the semi-arid area of southern Morocco. The study reveals that the semi automatic modular model approach is capable of handling the highly variable vegetation signal and projecting further scenarios of environmental changes. The model output will help to refine all models using land cover information (e.g. pastoral modelling), hydrological modelling (e.g. SWAT) and meteorological parameterisations (e.g. FOOD3DK). The output of the MOVEG DRAA model can also built a valuable information source for all kind of land users.