|Bryant, R. - UNIVERSITY OF ARIZONA|
|Rahman, M. - UNIVERSITY OF ARIZONA|
|Holifield Collins, Chandra|
Submitted to: International Geoscience and Remote Sensing Symposium Proceedings
Publication Type: Proceedings
Publication Acceptance Date: July 28, 2004
Publication Date: September 17, 2004
Citation: Thoma, D., Moran, M.S., Bryant, R., Rahman, M., Holifield Collins, C.D., Skirvin, S.M. 2004. Comparison of two methods for extracting surface soil moisture from c-band radar imagery.Proc. Internat'l. Geosci. and Rem. Sens. Sym., Sept. 20-24, Anchorage, AK, 4 p. Interpretive Summary: Surface soil moisture is important for plant growth, ranching, farming, and for determining cross country mobility in vehicles. Radar satellite imagery can be used to measure near-surface soil moisture, but different methods of determing moisture from imagery give different results. The purpose of this experiment was to evaluate two techniques for determining soil moisture from satellite imagery. The first method used a complex mathematical model to calculate soil moisture from image data. It did not work well because it was confused by the abundance of rock fragments in the soil. The second method involved comparison of a wet satellite image with a dry satellite image. The difference in image brightness was directly related to soil moisture. Furthermore, the second method was easier to apply and may work in any sparsely vegetated landscape. The contribution of this research was in identifying the better of two methods for predicting soil moisture from radar imagery in rocky landscapes.
Technical Abstract: The Integral Equation Method (IEM) model and a newly defined delta index were used to estimate near surface soil moisture from C-band radar satellite imagery in a semi-arid rangeland in southern Arizona, USA. Model results were validated against soil moisture measurements made in the field at the time of satellite overpass. The IEM model performed poorly in this environment due to abundant near-surface rock fragments which were not considered in the model. The index performed better than the IEM model and was shown to work with both ERS and Radarsat imagery. Additionally the index was simple to implement and implicitly accounted for both rock fragments and surface roughness.