Abstract
Woody biomass production is a critical indicator in evaluation of land use management and the dynamics of the global carbon cycle (sequestration/emission) in terrestrial ecosystems. The objective of the present study was to develop, through a case study in Sudan, an operational multiscale remote-sensing-based methodology for large-scale estimation of woody biomass in tropical savannahs. Woody biomass estimation models obtained by different authors from destructive field measurements in different tropical savannah ecosystems were expressed as functions of tree canopy cover (CC). The field-measured CC data were used for developing regression equations with atmospherically corrected and reflectance-based vegetation indices derived from Landsat ETM+ (Enhanced Thematic Mapper Plus) imagery. Among a set of vegetation indices, the normalized difference vegetation index (NDVI) provided the best correlation with CC (R2 = 0.91) and was hence selected for woodland woody biomass estimation. After validation of the CC-NDVI model and its applicability to Moderate Resolution Imaging Spectroradiometer (MODIS) data, time-series MODIS NDVI data (MOD13Q1) were used to partition the woody component from the herbaceous component for sparse woodlands, woodlands and forests defined by the Food and Agriculture Organization (FAO) of the United Nations Land Cover Map. Following the weighting of the estimation models based on the dominant woody species in each vegetation community, NDVI-based woody biomass models were applied according to their weighted ratios to the decomposed summer and autumn woody NDVI images in all vegetation communities in the whole of Sudan taking the year 2007, for example. The results were found to be in good agreement with those from other authors obtained by either field measurements or other remote sensing methods using MODIS and lidar data. It is concluded that the proposed approach is operational and can be applied for a reliable large-scale assessment of woody biomass at a ground resolution of 250 m in tropical savannah woodlands of any month or season.
Acknowledgements
Landsat ETM+ and MODIS data employed in this research were freely obtained from the USGS and NASA data servers (http://glovis.usgs.gov and https://lpdaac.usgs.gov/lpdaac/products/). We thank Google for making the very-high-resolution satellite images available on Google Earth, and the FAO Africover Project for providing the Land Use/Cover Map of Sudan (2003). Finally, sincere thanks are due to Dr John Ryan, Dr Jack Durrell of ICARDA, and two anonymous reviewers for their helpful suggestions and comments that helped improve this paper.
Notes
1. 1. In this paper, we are talking about the Sudan prior to 9 July 2011, when South Sudan became independent.
2. 2. Method developed based on a personal communication with Dr Rolf Sommer (ICARDA), October 2008.