Soil Moisture

Introduction

The U.S. Geological Survey (USGS) is engaged with the U.S. Department of Agriculture’s (USDA) Agricultural Research Service (ARS) and the University of Georgia’s National Environmentally Sound Production Agriculture Laboratory (NESPAL) both in Tifton, Georgia, USA, to develop transformations for medium and high resolution remotely sensed images to generate moisture indicators for soil. The Institute for Technology Development (ITD) is located at the Stennis Space Center in southern Mississippi. ITD has developed hyperspectral sensor systems that, when mounted in aircraft, collect electromagnetic reflectance data of the terrain. The sensor suite consists of sensors for three different sections of the electromagnetic spectrum; the Ultra-Violet (UV), Visible/Near InfraRed (VNIR) and Short Wave InfraRed (SWIR). The USDA/ ARS’ Southeast Watershed Research Laboratory has probes that measure and record soil moisture. Data taken from the ITD SWIR sensor and the USDA/ARS soil moisture meters were analyzed to study relationships between SWIR data and measured soil moisture.

Project Description

The ITD sensors collect two dimensional spatial (image) data sets at a variety of different wavelengths. When stacked together, these image data sets create hyperspectral data cubes with two dimensions containing spatial information and the third dimension containing spectral information. These HyperSpectral Imaging (HSI) sensors record hyperspectral data cubes for targets of interest. On November 30, 2005, ITD used the SWIR HSI sensor to collect data over 18 soil moisture meters managed by USDA/ARS.

The soil moisture meters managed by the USDA/ARS are near Tifton, Georgia. Data is recorded from these sensors every half hour at three different depths; two inches below the surface, eight inches below the surface and twelve inches below the surface. The data is recorded continuously throughout the year and is available online (http://www.tifton.uga.edu/sewrl/radio/lrdata.htm)

The geographic locations of 18 soil moisture meters in the vicinity of Tifton, Georgia were provided by USDA/ARS. Flightlines were drawn over the 18 soil moisture meters. Data was collected over these flightlines using the ITD SWIR HSI sensor. At the same time, radiometer data was collected at selected locations within the flightlines on the ground. The Analytical Spectral Devices (ASD) SWIR radiometer that was used was provided by the USGS in collaboration with the University of Missouri – Rolla’s Department of Civil, Architectural & Environmental Engineering.

The SWIR data was georeferenced and calibrated using the field spectroradiometer data (Figure 1). Regions of Interest (ROI) were drawn using the ENVI image processing application over the image data set corresponding with the location of the soil moisture meters. These ROIs were used to extract the spectral information out of the SWIR data set for each of the soil moisture meter locations.

Figure 1: SWIR data over NESPAL and USDA office

Figure 2 shows the image surrounding soil moisture meter labeled 32, with the location of the soil moisture meter communications tower indicated by a white plus mark in the center. The display color components were provided by the same image wavelengths as used in Figure 1.

Figure 2 : SWIR Image From the Area Around Soil Moisture Meter 32

The 3x3 window of pixels within each ROI were averaged to create one spectra for each of the soil moisture meters. Figure 3 shows the spectra extracted from around the soil moisture meter labeled 32.

Figure 3 : Spectra from Soil Moisture Meter 32

The soil moisture data for each soil moisture meter was provided by USDA/ARS. The data for November 30, 2005 was extracted from the larger data set. The data for the hour and a half between 10am and 11:30am were averaged. The average values for the 2 inch, 8 inch and 12 inch probes were recorded and used in analysis.

Analysis

Correlations between these spectra per band and the 2 inch soil moisture data were generated. The most significant correlations were with bands 15, 55, and 81 which correspond with wavelengths 1062, 1422 and 1664 nanometer. A graph of the correlations is shown in Figure 4.

Figure 4: Correlations with Spectrum and 2 Inch Soil Moisture

A new GLM was generated to model the 2 inch, 8 inch and 12 inch soil moisture data with bands 15, 55 and 81. On the 2 inch model, band 15 did not exhibit a significance value below 0.05. Thus, the model was run again using only bands 55 and 81. The results, shown in Table 1, show the 2 inch and 12 inch models to be significant at the .05 level. Although close, the 8 inch model is not significant at the 0.05 level. The 2 inch model has an R2 of 0.79. The R2 value of the 8 inch and 12 inch model are significantly lower.

Table 1. Model coefficients and statistical significance

The band coefficients for the 2 inch model and their significance levels are shown in Table 2. The same information for the 8 inch and 12 inch model are shown in Tables 3 and 4.

Table 2. Band significance for 2 inch model

Table 3. Band significance for 8 inch model

Table 4. Band significance for 12 inch model

Conclusions

The conclusion of this preliminary study is that there is relationship between SWIR data and soil moisture. The results for the 2 inch model, shown in Table 1, were highly significant and had an R2 value of 0.79. This indicates that the model is significant and with a correlation factor R= 0.89 = Square Root (.79). This data indicates that the wavelengths that best model the soil moisture at the 2 inch depth are at 1422 and 1664 nanometers.

The data analysis also showed that the ability for remotely sensed SWIR data to estimate soil moisture deteriorates at the 8 inch or 12 inch depth. Although the model for soil moisture at the 8 inch depth did have a somewhat reasonable R2 of 0.49, the model was barely significant at the 0.05 level. The model for soil moisture at the 12 inch depth was significant with an R2 of 0.56.

For more information on this or any other current research, contact us.