A COMPARISON OF SOIL-MOISTURE RETRIEVAL MODELS USING SIR-C MEASUREMENTS OVER THE LITTLE-WASHITA RIVER WATERSHED

SIR-C L-band measurements over the Little Washita River watershed in C hickasha, Oklahoma during 11-17 April 1994 have been analyzed for stud ying the change of soil moisture in the region. Two algorithms develop ed recently for estimation of moisture content in bare soil were appli ed to these measurements and the results were compared with those samp led on the ground. There is a good agreement between the values of soi l moisture estimated by either one of the algorithms and those measure d from ground sampling for bare or sparsely vegetated fields. The stan dard error from this comparison is on the order of 0.05-0.06 cm(3)/cm( 3), which is comparable to that expected from a regression between bac kscattering coefficients and measured soil moisture. Both algorithms p rovide a poor estimation of soil moisture or fail to give solutions to areas covered with moderate or fail to give solutions to areas covere d with moderate or dense vegetation. Even for bare soils the number of pixels that bear no numerical solution from the application of either one of the two algorithms to the data is not negligible. Results from using one of these algorithms indicate that the fraction of these pix els becomes larger as the bare soils become drier. The other algorithm generally gives a larger fraction of these pixels when the fields are vegetation-covered. The implication and impact of these features are discussed in this article. (C)Elsevier Science Inc., 1997.