Non-uniform Sampling Method On Observed Brightness Temperature Scene For Aperture Synthesis Microwave Radiometer Simulation

The aperture synthesis radiometer(ASR)is an interferometric system in which many small antennas are used to form the receiving antenna array,but the actual hardware of the system is highly complex,it is much expensive and hard to analyze accurately.So simulations for the aperture synthesis radiometer is significant in the research.In the “power level” simulation system of aperture synthesis radiometer,the target scene module and signals receiving module form the forward model from brightness temperature to visibility function(BT2V forward model),the BT2 V forward model for aperture synthesis radiometer(ASR)is of great importance to predict the instrumental degradation or evaluate part of the visibility outside the object area.By far,simulations of the BT2 V forward model are based on the equidistant sampling method,but the uniform sampling method on the observed brightness temperature scene suffers from high computational complexity.In this letter,a non-uniform sampling method is proposed for aperture synthesis radiometer.The sampling intervals in this method vary with the distribution regularity of brightness temperature.Large sampling interval is used when the brightness temperature is stable,but it will be partitioned into sufficient many small intervals in the area brightness temperature changes rapidly to preserve more high-frequency contents.The size of sampling intervals,the number of sampling points,the length of baselines,the type of the target scene and their influences on numerical accuracy and on computational complexity in BT2 V forward model simulation are discussed for the non-uniform sampling method and the uniform sampling method.Compared with the equidistant uniform sampling method,the high efficiency of the non-uniform sampling method in ASR BT2 V forward model simulation is verified by numerical simulations and the considerable improvements are obtained.The approximation of visibility function in non-uniform sampling method is much closer to analytic Fourier transform,ensuring the better reconstructed brightness temperature image.According to the simulation results,for given desired calculation accuracy,the number of sampling points in non-uniform sampling method will be reduced to at most one-third,it improves the rate of convergence of the visibility function numerical integration and saves the computation resources.And for the given computation complexity,the calculation error of BT2 V forward model will degrade two or three orders,it guarantees much lower reconstructed errors,especially in the coastal areas.