Modeling the hyperspectral reflectance signal of moist sandy soils

摘要

Removing soil moisture effects in spectral images is critical for agricultural remote sensing. Yet, current available soil moisture reflectance models fail to properly address subtly, yet significant, moisture induced reflectance changes within soils of the same texture class. The operational implementation of these models is as such limited, particularly in agricultural fields where within-field variations in soil characteristics such as organic matter and clay content prevail. In this study, the effect of soil moisture content (SMC, water content by weight) on the reflectance in the 400 to 2500 nm spectral domain is studied for six sandy soils located in Citrus orchards in the Western Cape Province, South Africa. Novel insights in soil moisture reflectance modeling of sandy cultivated soils are provided. The wavelength and soil specific variations in model parameters are mechanistically modeled and a general model for moist reflectance of cultivated sand soils is presented and successfully tested.