Estimation of moisture content of sandy beaches from X-band synthetic aperture radar

Abstract

Moisture content is a critical parameter for estimating the strength of partially saturated sand for engineering challenges such as beach trafficability. A framework for estimating the volumetric moisture content of sandy beaches using satellite-based HH-polarized X-band synthetic aperture radar imagery is presented and used to test the applicability of three theoretical scattering models: Oh, Dubois, and the Integral Equation Model (IEM). The developed framework relies on the measured backscatter coefficient, soil surface root-mean square (RMS) height, and the geometric characteristics of the image. Models for estimating the RMS height were developed based on field measurements for two distinct sites composed of predominately quartz sand and approximately uniform beach slopes: Duck, North Carolina and Cannon Beach in Yakutat, Alaska. The models were developed and tested for incidence angles of 23.3°-54.2° using data obtained from the Cosmo-SkyMED and TerraSAR-X satellites. Four sets of RMS height models were tested: Oh with moisture contents greater than 0 %, Oh with moisture contents greater than a very dry threshold, Dubois, and the IEM. Unique RMS height models, specific to a moisture content model, were developed for incidence angles ranging from a single incidence angle to a range of consecutive incidence angles. Applying the RMS height models, the root mean square error (RMSE) of moisture content was 0.7–6.9 %. Images with incidence angles of 30°- 46° and 40°-50° resulted in the best estimates of moisture content when compared to other ranges of incidence angles for the models tested. Deviations generally represented underestimates of the moisture content (0.1 %–1.3 %), with greater underestimates observed for the IEM. Spatial estimates of moisture content resulted in two distinct zones, one with low moisture contents and a second with slightly elevated moisture contents for all models except the IEM. Challenges associated with differences in scattering mechanisms, a lack of data with high moisture content, and sensitivity of models to small changes in RMS height are discussed.