Biomass estimates of semi-arid vegetation from airborne remote sensing in Sorbas, South-east Spain

This study utilises CASI and ATM airborne data from three time periods; 1996, 2001 and 2006, to assess the changes in semi-arid vegetation in the area immediately south of Sorbas, South East Spain. The data was orthocorrected using a combined LiDAR/Photogrammetric DEM within NERC’s AZGCORR software, producing a closely co-registered data series. The image mosaics produced were then manipulated to produce a variety of Vegetation Indices (VI).. Using Pearson’s correlation, each VI has been correlated with ground truth data to evaluate the relationship between the VI and the Above Ground Biomass (AGB) of the area. The ground truth data comprises AGB estimates from three meter square clear-cut quadrats adjusted for moisture content to provide dry-weight biomass estimates. Quadrats were selected from areas that covered the range of expected biomass values, from <8 tonnes/Ha (wet weight) to >80 tonnes/Ha (wet weight) and were reasonably representative of the surrounding contiguous area to reduce pixel mis-location errors. Initial correlations from the 2006 ATM data for NDVI, OSAVI and SAVI are all around 0.94. It is expected that the CASI data will also have very high correlations as band midpoints are more suited to a number of the VI used. An assessment of the VI and how they correlate with AGB will provide information on which VI is most appropriate for the semi-arid vegetation types seen in this area. Validation ground data have also been collected, effectively providing an independent dataset for confirmation of the VI predicted biomass values throughout the image. Initial validation quadrat results have been correlated using Pearson’s Correlation with the 2006 ATM data. Correlations for NDVI are around 0.85, demonstrating the ability of the NDVI to predict AGB accurately enough for a regional assessment of biomass values with this VI. Total biomass in the imaged area will be estimated using a sub-pixel classification technique to determine vegetation type and, in combination with the AGB estimates, ratios for Below Ground Biomass from the research literature. Carbon per Biomass measures undertaken on samples collected from the field can then be applied to the total biomass and a carbon value can be estimated for the area for each year data was collected. This will allow a temporal comparison to be carried out, providing information on the total carbon budget for the natural vegetation, whether increasing or decreasing. The study will enhance the body of scientific knowledge of Carbon sequestration in semi-arid Europe, and may have implications for the European Union Common Agricultural Policy, which currently provides subsidies for extensive agriculture in the region.