Simulation of solar zenith angle effect on global vegetation index (GVI) data

Publisher Website
Google Scholar
Abstract
The aim of this paper is to investigate the effect of solar zenith angle on the normalized difference vegetation index (NDVI) derived from the Advanced Very High Resolution Radiometer (AVHRR) channel 1 (red) and channel 2 (near-infrared) data. The NDVIs as a function of solar zenith angle are simulated for a nadir-looking sensor. Simulation is essentially the inverse of atmospheric correction of remotely-sensed data. Calculations are made for four surface cover types, namely, the high, moderate and low green-leaf vegetation densities and bare soils. The results show that the NDVI for bare soil remains constant for solar zenith angles up to about 60°, then decreases for solar zenith angles above this. This constant value of NDVI is found to be about 1 per cent larger than the correct (surface) value. The NDVIs for high, moderate and low green-leaf vegetation densities remain constant up to about 30° solar zenith angle. However, for larger solar zenith angles these NDVIs decrease smoothly and significantly. With the exception or the bare soil case, the relative deviation of the NDVI from its constant value has been fitted to a curve as a function of solar zenith angle. The equation of this curve is found to make solar zenith angle corrections to the NDVI with a maximum possible error of about 9 per cent for solar zenith angles up to about 70°. This error decreases as the vegetation leaf density decreases. Since the global vegetation index (GVI) data are usually from the near-nadir view (≤ 10°) and since within this view angle there is negligible off-nadir view effect, the relation obtained in this work may be used to correct GVI data for solar zenith angle effect. But it cannot be used for correcting the data which correspond to larger off-nadir view angles used by the AVHRR. Finally, the results of this investigation show that the multitemporal AVHRR data from about 4 months are needed for producing one GVI image.