Modeling biophysical controls on land surface temperature and reflectance in grasslands

Abstract

The influence of soil moisture and vegetation density on land surface reflectance and temperature is investigated at a grassland site in northeastern Kansas. To do this, a forward model is developed and validated against measurements from a helicopter mounted multi-spectral sensor. Soil background reflectances are calculated as a function of soil type and moisture content and are incorporated into calculations of surface reflectance using a canopy radiative transfer model. Surface temperature is predicted using a two-layer framework that includes the effects of soil moisture on soil thermal properties. Variation in vegetation density is included in both models via the canopy leaf area index. Results from model simulations show good correlation between normalized values of modeled and observed surface spectral reflectance. Comparison of modeled vs. observed temperature data show that a relatively parsimonious set of variables (net radiation, air temperature, soil thermal properties, and leaf area index) may be used to model land surface temperature with good accuracy. These results support previous empirical and modeling studies suggesting that for land areas characterized by fractional vegetation cover, the slope of the relationship between spatially distributed NDVI and surface temperature measurements is diagnostic of mean soil moisture, while variation around the slope is indicative of local variation in soil moisture.