Assessment of radar resolution requirements for soil moisture estimation from simulated satellite imagery

Radar simulations were performed at five-day intervals over a twenty-day period and used to estimate soil moisture from a generalized algorithm requiring only received power and the mean elevation of a test site near Lawrence, Kansas. The results demonstrate that the soil moisture of about 90% of the 20-m by 20-m pixel elements can be predicted with an accuracy of + or - 20% of field capacity within relatively flat agricultural portions of the test site. Radar resolutions of 93 m by 100 m with 23 looks or coarser gave the best results, largely because of the effects of signal fading. For the distribution of land cover categories, soils, and elevation in the test site, very coarse radar resolutions of 1 km by 1 km and 2.6 km by 3.1 km gave the best results for wet moisture conditions while a finer resolution of 93 m by 100 m was found to yield superior results for dry to moist soil conditions.