Soil Moisture Content from Spectral Reflectance Using Visible, Near-Infrared, and Short-Wave Infrared Light

Abstract

Quantification of soil moisture content is important for understanding physical processes that occur on and within the earth. The methods that are currently used for the determination of soil moisture content are point-based (gravimetry, time-domain reflectometry, neutron scattering, or gamma ray scanning). Remote sensing methods exist but have several disadvantages (poor resolution and dependence on local meteorological conditions). The objective of the study described herein was to determine soil moisture content by means of remote sensing from hyperspectral imagery. Reflectance values obtained from optical remote sensing (wavelengths from 350 nm to 2,500 nm) were acquired from materials with different moisture content levels. Reflectance values, obtained from a spectroradiometer, were obtained for three soils and two commercial sands and then correlated with the initial soil moisture content obtained from gravimetric methods. The correlations were based on the collected reflectance values at prescribed wavelengths and a continuum analysis across all wavelengths. For all investigated materials, the reflectance values decreased with increasing moisture content. The best results were obtained when 1,900 nm was considered (coefficients of determination greater than 0.94, indices of agreement greater than 0.97, coefficients of efficiency greater than 0.41, mean absolute error less than 1.2 percent, and root mean square error less than 1.6 percent). With the continuum analysis, 1,450 nm was the best wavelength. Based on these results, a correlation was observed to exist between soil moisture content and the spectral reflectance values. Using this technique, the determination of soil moisture content can be rapidly obtained (within seconds).