GOES Image Enhancement Study–A Feasibility Test

Abstract: This paper describes a small project to look at the feasibility of computer processing of GOES meteorological images to improve spatial resolution. Using software of the Maximum Entropy Method and Maximum Likelihood Method types, typical images in VIS and IR were selected of several severe weather cases in the U.S. Due in part to the fact that the GOES imager does oversample objects in the East-West scans, it was found possible to increase spatial resolution by from 15 to 31% after computer processing. I.e. A typical GOES image in the VIS that might be at 1.3 km was found to have 0.9 km resolution after processing. Normal computer processing with line sharpening and shading and other simple deconvolution methods does not increase spatial resolution, but with MEM and MLM software we were able to do so, since the imager was recording more information than it typically displayed in the image pixel size.

The before and after image results of severe weather are quite dramatic and the improvement can be seen easily. A panel of NWS experts and others were assembled and offered detailed results from our study. The processing appears to be efficient, and we are still looking at which products would benefit from the processing for the future. With some additional simulation and modeling, it may be possible to get more resolution improvement. It appears, too, that smller temperature features are visible in the IR images, which may improve early definition of storm cells and structure.

We believe the study is carefully documented and shows what a small technical project can achieve with a small team looking for innovation. We would like feedback from the audience and encouragement from other agencies who do digital image processing of this type.