An inverse model for sea ice physical parameter retrieval using simulated annealing

In this paper, an inverse model for applications in sea ice parameter estimation is investigated. The algorithm utilizes a forward model based on Radiative Transfer theory and Dense Medium and Amplitude Correction Theory (DMPACT), together with a global optimizer known as Simulated Annealing. The purpose of the forward model is to calculate the radar backscatter data from a set of input parameters. Simulated Annealing is then applied to minimize the difference between the forward model calculation and the measurement data by changing one or more of the unknown parameters. By deducing the value of the unknown parameter which gives the best minimum, the model is able to predict the corresponding sea ice parameter. In this paper, the data from ground truth measurements at Ross Island, Antarctica and the radar backscatter data from satellite images of the same area have been used for the simulation of the inverse model with promising results.