SST Reliability Categories and Reflectance Table Documentation

The Naval Research Laboratory (NRL) Sea Surface Temperature (SST) software suite categorizes each produced SST retrieval into one of three categories based on the estimated reliability of the data. Reliability categories, from most to least reliable, are respectively labeled clear (1), probably clear (2) and questionable (3). Quantitative reliability estimates in the form of bias and standard deviation are then derived from comparisons to buoys for each category, and assigned to each retrieval. This report documents how the NRL SST software suite categorizes the SST retrievals. Two methods currently coexist to categorize the SST data. The first method is the 20-year-old legacy categorization, which mostly depends on the field test, a comparison between SST retrievals and the corresponding analyzed SST field. The field test-based categorization is complemented by a series of ad hoc test to recover incorrectly categorized retrievals as the field test performs poorly in dynamic regions. This method has become less effective with the improved detection and discrimination of SST contamination that has been implemented within the NRL SST suite. The second method is the new standard categorization method, which relies on the tests that detect contamination of the SST retrievals within the NRL SST suite, and how well the retrievals pass the contamination tests, to assign reliability categories. For daytime, the ref2d reflectance table test is shown to be most valuable to categorize SST retrievals. For nighttime, the 4 m brightness temperature inter-comparison test appears most effective for that categorization. Additional details on the standard categorization method and on other contamination tests which are used to estimate the reliability of the retrievals are included within this report.