| Soil moisture is an important water variable that has various effects on the environment, such as the water cycle, crop production and for forecasting changes in regional water balances. In a world where water consumption is increasing, it is important to conserve this valuable resource. Traditionally, soil moisture content has been obtained predominantly through in situ methods (soil moisture probes). The main problem with obtaining data through these means, especially over large areas, is that it is costly and time consuming. On the other hand, remote sensing can be considered as an important tool to estimate soil moisture content if properly calibrated. The shortwave Infrared bands for the majority of multispectral and hyperspectral sensors have the potential for estimating the soil moisture content of the top few centimeters of soil.;For this study, an ASD FieldSpec spectroradiometer (350-2500nm) was used to measure the spectral reflectance of soil samples. Most spectroradiometers have the ability to give more robust spectral measurements due to the number of bands these sensors have. Spectral measurements have shown promise in the wavelengths surrounding the water absorption bands (1450, 1900, and 2800 nanometers). These areas of the electromagnetic spectrum show the most variations between saturation levels in soils. The goal of this study is to observe and analyze the effects that moisture have on soil reflectance, as well as estimate soil moisture content from these reflectance values.;In this study, the band ranges of 1425 through 1475 nanometers, and 1875 through 1925 nanometers were selected to cover the water absorption bands. The results showed that combining the two areas of the spectrum produced the most significant results, with R2 values greater than 0.95. Several soil moisture models were compared for this study. Regression analysis was used to compare the moisture models. The Lobell and Asner (2002) model produced the best results for the individual band ranges and the combined band range when applied to a regression model, with p-values less than 0.05. |
