Characterizing surface soil water with field portable diffuse reflectance spectroscopy

Surface soil moisture plays a key role in regulating a variety of processes associated with water and energy, especially evaporation. Most prevailing methods measure soil water content only for soil blocks or layers of a certain depth, rather than directly on the soil surface. Diffuse reflectance spectroscopy (DRS) provides a way to approximate surface soil water content by utilizing surface reflectance to quantify soil water. The direct measurement of surface water content could potentially improve the understanding and modeling accuracy of the processes associated with surface water dynamics. The present study investigated the reflectance variations in one artificially constructed sample set, one set of natural soil cores, and one set of natural surface soils, with various moisture levels. Results showed that the reflectance decreased non-linearly with an increase in soil moisture at single wavelengths for all three sample sets. The models derived from partial least square (PLS) regression with segmented cross-validation using natural logarithm transformed wave bands or full spectra provided accurate and stable prediction for all moisture ranges studied. Although it is still problematic to construct a single, universal model for surface soil moisture, segmented models based on soil property classes (i.e. texture, color) are able to accurately predict surface soil water content from measured reflectance. Specific calibrations for soils with similar soil texture or color are expected to provide promising prediction of surface soil moisture.