Derivation of Volumetric Liquid Water Content from the RADARSAT-1 SAR Images over a Permafrost Region in interior Alaska
Tuesday, 16 December 2014
The empirical adopted integral equation model (EA-IEM) is implemented as a promising algorithm for liquid water content from the microwave data over a bare soil and sparsely vegetated conditions. The EA-IEM provides simplified mathematical expressions to calculate the soil dielectric constant. The Newton-Rhapson iteration is used to generate the calibrated rms height and calibrated correlation length by using the absolute difference between the calculated liquid water content (LWC) and the measured liquid water content. The absolute difference is less than the threshold value set to 1e-8. The calibrated rms height shows a constant value of 0.02 m while the calibrated correlation length varies for different sample points. A simple exponential regression model is established between the calibrated correlation length values and the backscattering coefficient observations. In addition, the regression model is incorporated into the EA-IEM as a robust way in determining the roughness parameters for retrieval of LWC over a large area. Liquid water content is then calculated directly from radar backscattering coefficient without iteration. Seven strategies were adopted to calibrate and validate the two NCRS-SCAN sites: Nenana and Ward Farm. A comparison between the predicted LWC and the measurements is performed for each strategy, and the root-mean-square (rms) error is found to be 3.60%, suggesting that the strategy one performs well compared to other strategies. All these strategies indicate that the EA-IEM can be used to retrieve soil moisture under the tested range of model parameters: incidence angles between 10o and 60o, surface rms height from 10 to 25 mm, and correlation length from 30 to 100 mm.