COMPARISON OF THERMAL-BASED SOIL-MOISTURE ESTIMATION TECHNIQUES ON A HISTOSOL

Thermal-infrared remote sensing images have the potential to provide r egional-scale, high-resolution soil moisture datasets for a variety of agricultural and environmental monitoring and modeling efforts. In th is preliminary ground-based study, five different thermal soil-moistur e estimation techniques were evaluated in south Florida on Pahokee muc k soil (euic, hyperthermic Lithic Medisaprist) under conditions of ope rational soil-moisture range and cover types of bare muck, sugarcane, and grass. The morning surface-temperature (MST), afternoon surface te mperature (AST), morning surface-minus-air temperature (MSMA), afterno on surface-minus-air temperature (ASMA), and diurnal surface temperatu re variation (DSTV) techniques were evaluated with and without air-tem perature normalization. Measurements consisted of thermistor and radio meter thermal data and soil samples at 3 cm depth increments to a dept h of 18 cm. For bare muck, a good correlation with gravimetric soil-mo isture content (GSMC) was obtained with the normalized DSTV method (r2 = 0.80, 0.81, 0.69 respectively for depths to 3, 6, 9 cm). For grass cover, the DSTV method produced a fair correlation that was uniform th roughout the rooting depth of the plant (r2 = 0.60, 0.60, 0.63, 0.64 r espectively for depths to 3, 6, 9, 12 cm). Air-temperature normalizati on was beneficial under the condition of bare muck, but not under the condition of vegetated cover. The results of this study indicate that a DSTV-based technique has the potential to discriminate relatively fi ne soil-moisture differences (to +/- 3.86% GSMC) within the operationa l range (53.5 to 220.0% GSMC) for organic soil agriculture.