Remote sensing of vegetation dynamics in West Africa: improved satellite time series for phenological analyses

Abstract

Vegetation dynamics and the lives of millions of people in West Africa are closely interlinked with each other. The high annual variability of the phenological cycle considerably affects the agricultural population with late rainfalls and droughts, often resulting in serious food crises. On the other hand, the rapidly growing population has a high need for space due to expanding cities and a low agricultural efficiency. This situation, together with a changing climate had a strong impact on vegetation dynamics in West Africa and will play a major role in the future. The mapping and monitoring of seasonal and long-term changes in West Africa’s vegetation is essential to understand the implications for nature and population. In order to cover the spatial extent of West Africa and, at the same time, to track the temporal development of vegetation, time series analysis of remote sensing data are a valuable tool. A considerable amount of research has been conducted on this topic in West Africa during the past 30 years which was reviewed and summarized in a first step of this PhD thesis. The result of this effort was handed in as an article and already accepted for publication in the International Journal of Remote Sensing (Knauer et al., 2014). At the moment, there are several remotely sensed time series of vegetation parameters available, but for an application in sub-humid to arid regions like West Africa, all of these products have their advantages and disadvantages mainly related to the sensor characteristics. In areas with persistent cloud coverage like the coastal regions of West Africa, time series of sensors with sun-synchronous orbits such as the MODIS sensor are highly affected by data gaps. In contrast, geostationary satellite sensors such as the MSG SEVIRI sensor, which has a fixed position over Africa, can overcome this problem but have the issue of a rather coarse spatial resolution. The complementary strengths of available satellite sensors could result in an optimized vegetation parameter time series that is suitable for regional application in sub-humid to arid regions like West Africa. In this PhD thesis, a data fusion of two different satellite sensors is intended in order to overcome the issues mentioned above and to allow for a consistent analysis of vegetation dynamics in West Africa. For this purpose, an established fusion algorithm (ESTARFM) will be applied and modified in order to fill the gaps in the MODIS dataset with data from the SEVIRI sensor. Based on this improved dataset, the land surface phenology of West African vegetation will be investigated. Different parameters for the monitoring of seasonal and long-term changes of vegetation will be derived from the time series and tested for their suitability. Currently, the preprocessing of the two sensors’ data as well as the modification of the fusion algorithm is conducted.