In this paper, we will study the estimation of channel state information of a radar system whose clutter returns can be modeled using a channel matrix with entries computed from fundamental physics using a stochastic Green's function based approach. We develop a simple algorithm for channel estimation using well-known frequency domain techniques. We present a more realistic study of the underlying clutter-to-noise-ratio (CNR) by treating it to be dependent on the channel, pulse, and range bin indices. We will present extensive simulations using high fidelity modeling and simulation tool RFView to discuss channel estimation accuracy, performance metrics, and other challenges pertaining to this problem.