Calibrating airborne vegetation data for hydrological applications under dry conditions

Accurate spatial vegetation data are essential for hydrological modelling since vegetation processes directly influence biomass production and affect the distribution of surface water. Spatially distributed vegetation data are difficult and expensive to collect on the ground. Ground-collected data rarely provide complete spatial coverage at a single time. Remotely sensed data provide spatially extended maps of the surface cover in catchments, but require calibration. In this study, values of the airborne Normalized Difference Vegetation Index (NDVI), obtained with the Compact Airborne Spectrographic Imager (CASI), were calibrated with ground biomass samples in a 27km(2) catchment consisting of 65% partially grazed pastures and grasses and 35% open and medium density woodland. Linear, quadratic and exponential regressions were applied to six waveband combinations of CASI NDVI and the best result was an exponential correlation of r(2) = 0.62. This suggests that CASI NDVI has an exponential relationship with biomass. Calibration was affected by vegetation type and height, grazing, possible saturation of the near-infrared (NIR) bands and the narrow swathe width of aircraft data. Ground validation between Leaf Area Index (LAI) and biomass gave an r2 = 0.80. No statistically significant correlation was found between LAI and airborne NDVI. Significant fractions of non-green biomass at some sites, due to dry conditions, were seen as a contributing factor.