GLOVEG - Centre of excellence in remote sensing of terrestrial ecosystems dynamics

Context and objectives

The objective of the GLOVEG-project is a better understanding of the dynamics of terrestrial ecosystems by:

The development of remote sensing-based estimates of surface attributes. The analysis of time trajectories of multi-temporal NDVI datasets at seasonal, inter-annual and decadal time scales. The development of a fully water limited ecosystem model. Development of evapotranspiration and soil moisture content products, using thermal remote sensing imagery is a main objective a well.

Techniques of successive land cover mapping have promoted a vision of land-cover change consisting mostly of linear trends of conversion from one land-cover type to another. More advanced remote sensing approaches, that allow monitoring of continuous biophysical indicators, highlight continuous variations in land surface attributes in space and time, at the seasonal and inter-annual scales. These changes are non-linear and may include abrupt, episodic events in addition to cyclic changes or trends.

The development of specific applications in fire risk assessment and vegetation post-disturbance re-growth is an additional objective in the framework of land cover change monitoring. Indicators for fire risk were evaluated with remote sensing time series and meteorological data. The re-growth index is an indicator to assess post-disturbance vegetation carbon uptake. The specific objective of this work is to evaluate the ability of a set of selected indices (NDVI, NDWI, and KBDI), related to vegetation water status, to assess fire risk in savannah ecosystems. Accurate methods to describing post-fire development of vegetation are thereby essential to understand and monitor the changes of post-fire regeneration of terrestrial ecosystems.

Project outcome

Expected scientific results

The extraction method for bio-geophysical variables, e.g., LAI, fAPAR, fCover, chlorophyll a+b, water content, broadband albedo as well as hemispherical and normalized spectral reflectances for different soil-canopy matrices from atmospherically corrected VGT-P reflectance data has been developed. The bio-geophysical products produced with this method are new for the VGT instrument. For example instantaneous evapotranspiration of the European continent was estimated. It was validated with EUROFLUX data for the growing season of 1997. In collaboration with NEO BV, the ERS Scatterometer Soil Water Index was assimilated in C-Fix to estimate short-term water limitation effects. C-Fix was run with and without water limitation and including canopy and soil as heat buffers to convert air temperatures to canopy and soil temperatures. The effect of water limitation on NEP compared to non water limited NEP for May 1997 over Europe is given in Fig. 1.

Using improved metrics, the magnitude, extent, and nature of changes in photosynthetic activity and its timing across Sub-Saharan Africa was investigated. Changes in overall vegetation activity and shifts in its timing have considerable implications for a variety of processes including surface-atmosphere energy exchanges, terrestrial sources and sinks of carbon, the contribution of local evapotranspiration to the water cycle, disturbance regimes such as fires and pests, and the food security of human settlements using these ecosystems for food supply. The overall changes in vegetation were quantified each year between 2000 and 2004 and the proportion linked to differences in phenology and overall photosynthetic activity was examined. In addition, the nature of these changes in terms of frequency and duration, the proportion per ecosystem, identification of areas of intensive change, and the potential consequences of these changes was examined. It was concluded that most inter-annual changes are not from shifts in timing or phenology, but are largely due to differences in the amount of annual photosynthetic activity. The changes were likely climate driven with particular vegetation types most susceptible to inter-annual change and with a high spatial and temporal variability found across the African continent.

Chlorophyll-related indices related to Fuel Moisture Content (FMC) were evaluated using fire activity data. Satellite imagery was used to obtain more accurate maps of fire danger once the fire risk performance of a chlorophyll-related index was characterized. A novel image segmentation approach was applied to delineate vegetation-soil-climate complexes with similar phenological properties. The segmentation technique is based on a difference criterion using NDVI time series as an indicator for phenological differences. Segmentation results were validated based on two thematic vegetation layers. Results reveal scale effects of different ecological processes.
To provide quantitative estimates of burning efficiency based on remote sensing indicators two different techniques were tested. A technique based on time series analysis of sub-pixel fractions and a technique based on temporal variations of vegetation indices. Both techniques were compared and validated with field data containing burning intensity and severity indices based on Landsat imagery with higher spatial resolution. Statistical regression techniques were used to assess the performance of both techniques and the resulting quantitative indicators of burning efficiency.