A Neural Network Approach to Inversion of Snow Water Equivalent from Passive Microwave Measurements

Abstract

The inversion of snow water equivalent, SWE, from passive microwave remote sensing measurements may be accomplished by using a neural network trained with a dense media multiple scattering model. Brightness temperatures from 19 GHz vertical and horizontal polarizations, 22 GHz vertical polarization and 37 GHz vertical and horizontal polarizations which are available from the SSM/I sensors, are used as input to the neural network. The percentage error for estimated SWE varies from 9 % to 57 % for different snow conditions.