A STUDY OF AN INVERSION MODEL FOR SEA ICE THICKNESS RETRIEVAL USING SIMULATED ANNEALING

Previously, an inverse microwave scattering model based on radiative transfer was developed for the retrieval of sea ice thickness using radar backscatter data. The model, called the Radiative Transfer Inverse Scattering Model (RTISM), is a combination of the Radiative Transfer-Dense Medium Phase and Amplitude Correction Theory (RT-DMPACT) forward model and the Levenberg-Marquardt Algorithm (LMA). In this paper, the LMA in the RTISM is replaced with Simulated Annealing (SA) as the optimizer to allow a wider range of inversion capability. SA is a global optimizer, and its settings make it convenient to switch between different target parameters to be optimized for inversion. In this study, the model will first be tested using different data sets to verify its applicability. Next, the model is used to estimate the sea ice thickness around Ross Island, Antarctica using data from ground truth measurements together with satellite data from Radarsat-1 from the year 2006. In order to further validate the model, the data collected from measurements performed during an experiment to grow an ice sheet within a refrigerated facility at the U.S. Army Cold Regions Research and Engineering Laboratory (CRREL) are used to perform the retrieval of saline ice thickness. Preliminary results show that the new model performs as expected and shows potential. However, there are still limitations to the inverse model, and further improvements in the future need to be considered.