GOES-R Series 3.9-μm Channel Artificially Cold Pixels: Causes and Mitigation Strategies

The Level-1b product delivered by the Advanced Baseline Imager (ABI) on board the GOES-R series of satellites consists of detector radiances that have been resampled into pixels aligned to the Fixed Grid (FG). The FG is a projection based on the perspective of a satellite in an ideal geosynchronous orbit. It provides a coordinate system that is used as a basis for the Navigation and Resampling steps in the ABI L0-L1b algorithm. Navigation consists of solving for the FG coordinates for each detector sample after it has been calibrated into radiance units. This knowledge of the location of each sample is used as input to the Resampling process, which populates the output pixels by weighted sums of multiple input samples. Sample weights are determined based on their location along a two-dimensional kernel that is centered on each pixel in the output image.

The exact shape of each channel's kernel in both the north-south and east-west directions has been carefully designed by the ABI instrument vendor to both mitigate aliasing and preserve radiometric integrity. The generic shape of the kernel is similar to that of a SINC function, with negative lobes on either side of center. In nominal conditions, these negative weights are needed to maintain edge fidelity. However, when observing wildland fires these negative lobes occasionally give rise to artificially "cold" pixel radiances in the 3.9 μm channel. This presentation will give an overview of the Resampling process, how it gives rise to low radiances near fires, and recent efforts to mitigate this issue by way of alternate kernels.