Monitoring infiltration rates in semiarid soils using airborne hyperspectral technology

Abstract

Loss of rain and irrigation water from cultivated fields is a matter of great concern, especially in arid and semi-arid regions. The physical crust that forms on the soil surface during rain events is one of the major causes of increased run-off and reduced water infiltration into the soil profile. Based on previous studies that showed significant correlation between crusted soil and soil reflectance properties, we performed a systematic study over Loess soil from Israel, in order to map the infiltration rate from a remote distance, using Hyperspectral (or Imaging Spectroscopy, IS) technology. First, we simulated rain events under laboratory conditions, using the selected soil and varying rain energy treatments. After measuring the reflectance properties of the crusted soils, we developed a spectral parameter for assessment of crust status. The parameter, Normalized Spectral Area (NSA), uses the area under a ratio spectrum across the VIS-NIR spectral region (calculated from the ratio of the crusted (treated) soil spectrum to the non-crusted soil spectrum). The correlation between the NSA and infiltration rate values provided a significant calibration equation. Based on these results, we conducted an airborne campaign, employing the AISA imaging scanner adjusted to 30-channel data in the VIS-NIR, and established control plots (crusted and non-crusted soil) on the ground, to examine the NSA parameter for mapping the infiltration properties of Loess soils. Reasonable agreement was obtained between the two datasets (laboratory and air), suggesting that infiltration rates can be estimated remotely. Further research is necessary to expand the analysis to other areas and conditions (e.g. diverse CaCO₃ and moisture content of soil). The paper shows that spectral reflectance information in the VIS-NIR region can be used to assess soil infiltration affected by the soil crust, in both laboratory and air domains. It is strongly suggested that future study in this regard use the full optical range (VIS-NIR-SWIR-TIR), as well as a spectral library of crusted soils collected in or within the rain simulator environment.

Acknowledgment

This study was conducted under a CIRES fellowship foundation program at the Center for the Study of Earth from Space (CSES), University of Colorado at Boulder.