6-meter resolution surface soil moisture using L-band airborne SAR data

The spatial resolution of the current satellite SAR (synthetic aperture radar) systems are about 1km to tens of kms. These are insufficient for soil moisture mapping to support important science applications such as the prediction of wildfire and its damages, precision agriculture, and landslide monitoring. In comparison, airborne SAR offer superior spatial resolution. The assessment of the ratio of the soil moisture signal to the SAR noise suggests that 6-m resolution is feasible. In this paper, we present recent advances in inverting physical models for SAR observation to retrieve surface soil moisture (of ~ top 5 cm) at ~ 6m resolution. In the first example, for the agricultural cases in Canada, the soil moisture estimates are accurate to unbiased rmse of 0.041, 0.059 and 0.060 m3/m3 and correlation of 0.71, 0.83, and 0.69 for grass, wheat, and soybean fields, respectively over the soil moisture dynamic range up to 0.5 m3/m3, and soybean's full growth cycle. Second, over Canadian aspen forests, the retrievals over the full dynamic ranges of wetness are accurate to 0.04 m3/m3 unbiased rmse with correlations of 0.71 to 0.84, which is very encouraging for retrieval under the forest canopy. The high-spatial resolution of the airborne SAR have responded to the gaps among the trees (which is not feasible at coarse resolution), and helped the retrieval performance. The next case present an independent validation where physical models developed elsewhere were inverted for the retrieval at independent sites. In land-slide occurring steep hills (~15 degrees slope) in northern California, 6-m resolution maps of surface soil moisture were generated with unbiased rmse of 0.054 m3/m3. The spatial-scale of the hills is small and can only be resolved at such high resolution. In summary, the approach was validated successfully over croplands, steep terrains, and forests.