Using image reconstruction methods to enhance gridded resolutionfor a newly calibrated passive microwave climate data record

Abstract:

Spanning over 35 years of Earth observations, satellite passive microwave sensors have generated a near-daily, multi-channel brightness temperature record of observations. Critical to describing and understanding Earth system hydrologic and cryospheric parameters, data products derived from the passive microwave record include precipitation, soil moisture, surface water, vegetation, snow water equivalent, sea ice concentration and sea ice motion. While swath data are valuable to oceanographers due to the temporal scales of ocean phenomena, gridded data are more valuable to researchers interested in derived parameters at fixed locations through time and are widely used in climate studies. We are applying recent developments in image reconstruction methods to produce a systematically reprocessed historical time series NASA MEaSUREs Earth System Data Record, at higher spatial resolutions than have previously been available, for the entire SMMR, SSM/I-SSMIS and AMSR-E record. We take advantage of recently released, recalibrated SSM/I-SSMIS swath format Fundamental Climate Data Records. Our presentation will compare and contrast the two candidate image reconstruction techniques we are evaluating: Backus-Gilbert (BG) interpolation and a radiometer version of Scatterometer Image Reconstruction (SIR). Both BG and SIR use regularization to trade off noise and resolution. We discuss our rationale for the respective algorithm parameters we have selected, compare results and computational costs, and include prototype SSM/I images at enhanced resolutions of up to 3 km. We include a sensitivity analysis for estimating sensor measurement response functions critical to both methods.