Mapping Crop Leaf Area Index from Multi-Spectral Imagery Onboard an Unmanned Aerial Vehicle

Leaf area index (LAI) is a significant biophysical parameter used in many agronomic and ecological models. In comparison with satellite remote sensing, unmanned aerial vehicles(UAV) technology can obtain imagery with high spatial resolution for better accuracy of LAI estimation. This study conducted global sensitivity of input variables in PROSAIL model by the extended Fourier amplitude sensitivity test (EFAST) method and determined the most sensitive bands and vegetation indices (VIs) to LAI. Estimation accuracy of five input variable combinations in cost-functions was compared. Results of global sensitivity analysis show that green and red band are sensitive to LAI, and the correlation coefficient between measured LAI and estimated LAI from combination of these two bands as input variables in cost-functions is 0.85. For VIs, the most sensitive input variables are LAI, average leaf angle(ALA) and chlorophyll content(Chl). VIs of NDVI, RVI and MSR are sensitive to LAI with corresponding total sensitivity of 0.80, 0.69 and 0.72 respectively. The correlation coefficient between measured LAI and estimated LAI from VIs is over 0.75, indicating that it may be an alternative way for LAI inversion in PROSAIL model through LUT method.