|Bindlish, Rajat - SSAI|
|Van Der Velde, Rogier - INTL TECH CTR,NL|
Submitted to: International Geoscience and Remote Sensing Symposium Proceedings
Publication Type: Proceedings
Publication Acceptance Date: May 1, 2006
Publication Date: July 31, 2006
Citation: Bindlish, R., Jackson, T.J., van der Velde, R. 2006. High resolution soil moisture mapping using AIRSAR observations during SMEX 02. In: Proceedings of the International Geosciences and Remote Sensing Symposium, July 31-August 4, 2006, Denver, Colorado. p. 2324-2327. Technical Abstract: Soil moisture mapping using Synthetic Aperture Radar (SAR) can be achieved at high spatial resolution but has not been rigorously demonstrated over vegetated areas.. Radar measurements are sensitive to the structure and dielectric constant of the target. Of some concern are the crop type and row direction which are known to influence the backscatter observations. In this work, L- and P-band SAR images are used to map crop type and row orientation. The VV-polarized L-band 'o values are used to identify the crop type and the ratio of P-band VV-polarized over VH-polarized 'o measurements are used to determine the row orientation. Validation results show classification accuracies of 86% for the crop type classification and an accuracy of 94% accuracy for the detection of the row orientation. A vegetation correction approach derived from this analysis was used in combination with a semi-empirical surface scattering algorithm to estimate the soil moisture content using high resolution aircraft SAR data. Three different retrieval methods were employed to estimate spatially distributed soil moisture; 1) the Dubois model without vegetation correction, 2) the Water Cloud model using vegetation water content, 3) the Water Cloud model using vegetation water content and row direction information. These three methodologies were applied to airborne SAR) data collected during the soil moisture experiments in 2002 (SMEX02). For corn and soybean fields with an east-west row direction a standard error of estimate of approximately 0.075 cm3cm-3 was obtained and for north-south oriented crop rows accuracies up to 0.043 cm3cm-3 were achieved.