Diurnal Reflectance Model Using Grass: Surface-Substrate Interaction and Inverse Solution

The accuracy of using remote sensing data from earth orbiting radiometers can be improved by using a model that helps to separate the green-fraction in a canopy reflectance (rho) from thatch and soil background, accounts for their diurnal changes, and inverts to a solution of a biophysical plant property of interest. Previous studies addressed one or more of these needs separately. Because reflectance components are interdependent, difficulties remain in obtaining a combined inverse solution. We combined a conditional probability method with a novel experimental procedure to predict grass dry weight (dw). Using simple ratio (SR) = near-infrared reflectance (NIR)/red that varied with the normalized time T = (local time - sunset)/daylength, we predicted the mean differential grass dry weight, delta dw = dw1 - dw2, of two grass patches. SR1 and dw1 defined the first patch, which included a background and the second, SR2 and dw2, a predefined background. The inverted solution for delta dw was an ellipse with axes formed by the diurnal reflectance SR1 and SR2 coordinates. It described previously studied soil line and the zone of canopy x canopy x ground interactions. The standard error of predicting delta dw was 17%. We separately tested for plant height SR = f(h) or fresh weight SR = f(fw) using SR, and for dw as a function of normalized difference vegetation index NDVI = f(dw). SR = f(dw) produced superior results. Potential applications include noninvasive prediction of other biophysical plant properties in a single or in hyperspectral bidirectional reflectance in agronomic and ecological remote sensing.