A Study of Dome-C Ice Sheet Parameter Estimation Using 0.5-2 GHz Ultra-Wideband Radiometry

This paper is a preliminary study of estimation of Antarctica Dome-C ice sheet properties using 0.5-2 GHz Ultra Wide Band Software Defined Radiometer (UWBRAD). The estimated ice sheet parameters include the vertical temperature profile and stochastic parameters that describe the density fluctuations within the ice column. The borehole measured temperature profile is used as a prior. The inversion uses regularization to incorporate both the prior borehole measurement and the UWBRAD data in the cost function. A hybrid model that use both cloud and partially-coherent models is chosen for this study as the forward model [1]. Since the partially-coherent model has a long computation time, a neural network is trained to replace the hybrid forward model and this neural network is used in the inversion. A genetic algorithm inversion scheme is used.