Monitoring surface moisture state in the Canadian High Arctic using synthetic aperture radar (SAR)

Research was initiated in July 2004 to determine if change-detection methods employing synthetic aperture radar (SAR) could be applied to estimate change in soil moisture over a growing season for a site in the Canadian High Arctic. Two change-detection methods, image ratioing and principal component analysis (PCA), were assessed for their potential to provide accurate soil moisture change information. Volumetric moisture estimates from bulk soil samples were collected at 36 test sites concomitant with two descending RADARSAT-1 C-HH standard beam images collected on 8 July and 1 August 2004. Regression analyses between soil moisture values and image backscatter were performed for the 8 July and 1 August backscatter images (r2 = 0.099 and 0.413, respectively). The correlation between soil moisture change and image ratio values was poor (r2 = 0.038). A similar regression was performed between the soil moisture change values and the first and second principal component image values (r2 = 0.003 and 0.131, respectively). Weak correlations were largely attributed to saturated ground conditions during the 8 July image acquisition which are believed to have caused a specular reflection mechanism and reduced overall image backscatter. The higher correlation between late season soil moisture and SAR backscatter (1 August, r2 = 0.413) reinforces the suggestion that extremely wet and saturated ground conditions during early season lead to adverse results when assuming a linear relationship between moisture values and image backscatter. Lastly, despite the low correlation coefficients, the ratio and second principal component images displayed spatial similarities with respect to drainage networks and snowbank melt areas, suggesting the utility of these techniques for delineating moisture-saturated areas between dates.