Date of Award
Master of Science (MS)
Earth System Science & Policy
Dr. Xiaodong Zhang
Wet weather cycle since 1993 has brought ground water level closer to the surface of the soil in many areas in the Red River Valley (RRV) of the North basin. Many farmers have to delay spring plantation or autumn harvest due to excessive moisture in', their farmland. In such case, it becomes increasingly important to have timely and accurate information on the soil moisture conditions. Conventional soil moisture measurement techniques are point-based that results in poor spatial representation of soil moisture. Remote sensing techniques offer many potential advantages over traditional means such as repetitive coverage and areal representation. The objective of this study was to develop a remote sensing algorithm to be used by Landsat 5 TM and Aerocam to map surface soil moisture during the early stage of growing season in the RRV of the North basin.
Soil samples were collected and hyperspectral reflectances of the soil at various moisture levels were recorded under laboratory conditions. The first two experiments were carried out with Halogen lamps as the source of light whereas the third experiment was performed outdoor. Landsat 5 TM and Aerocam spectral response function were applied to the measured hyperspectral values to simulate the multispectral reflectance for each sensor. The results from these three experiments were consistent to each other and therefore were binned together. By using simple mathematical computations, band or band combinations that best estimates the surface soil moisture was found out. For validation, soil moisture was measured in different agricultural fields in the RR V during the time of Landsat overpass, and soil moisture was continuously monitored in a farm field in Fairmount, Richland County, North Dakota.
The difference of bands 5 and 1 was shown to correlate best with soil moisture concentration while the NIR band itself is the best for Aerocam. The estimated soil moisture using Landsat 5 TM agreed with the measurements with an R2= 0.90. The model performed well in dry or moderately wet conditions, but slightly underestimated by 3-4% for excessively wet conditions of more than 40% soil moisture in the field. The evaluation at the Fairmount experimental field also showed that the model performed well. Field verification for Aerocam imagery remained incomplete due to lack of irradiance value.
Rijal, Santosh, "Developing A Remote Sensing Algorithm For Deriving Soil Moisture From Spectral Reflectance" (2011). Theses and Dissertations. 2516.