Pasture Status from Visible and Near Infrared Radiometric Measurements

The monitoring of pasture status often has relied on the satellite-derived quantity known as the normalised difference vegetation index (NDVI). However, theoretical modelling of canopy reflectances by Sellers (1985) suggests that NDVI does not provide a good measure of green matter quantity, particularly when the leaf-area index (LAI) is large. For ryegrass/clover pastures in the Waikato region of New Zealand, LAI rarely drops below 2, and can range as high as 12. In-situ radiometric measurements of Waikato pastures have confirmed that NDVI is not very sensitive to vegetation bulk amount (total dry matter per hectare). We have found that a better indicator of bulk amount can be obtained by taking a linear combination of reflectances in three bands: green, red, and NIR (near infrared). We investigate these observations using a theoretical model of visible and NIR reflectance from a layered vegetation canopy. This work is an adaptation of the two-stream approach of Dickinson (1983) and Sellers (1985). Assuming isotropic scattering by the leaves, we derive albedos for the green, red, and NIR bands. Model predictions are found to be in broad agreement with the reflectance measurements.