Improving the combined use of reflectance and thermal channels for ocean and coastal cloud detection for the Sea and Land Surface Temperature Radiometer (SLSTR)

Reflectance imagery is used to aid daytime cloud detection in thermal remote sensing. This paper presents new approaches to utilization of reflectance for sea surface temperature (SST) remote sensing for the Sea and Land Surface Temperature Radiometer (SLSTR). SLSTR is an along-track scanning sensor with a complex instrumental viewing geometry, and no co-registration of the fields-of-view of reflectance and thermal pixels. Reflectance channels have twice the spatial resolution of thermal channels, and observations are placed on compatible image grids for the convenience of users of Level-1 data. We highlight limitations of simple methods, based on these image grids, of using reflectance imagery to inform cloud detection at the thermal resolution. We present improvements from averaging the N-nearest reflectance observations directly to the infrared instrument geometry, where N = 10 is chosen in this study when using the A and B stripes together. We show that the standard deviation of the N-nearest reflectance observations is another calculable quantity of use to improve the discrimination of clouds in the infrared image, beneficially reducing the weight placed on coarser-scale thermal spatial variability. The developments are illustrated by case studies in coastal zones over optically bright waters and around strong ocean fronts, and the benefit for SST products is quantified by the impacts on coverage and validation statistics. In a case study over optically bright waters, the clear-sky fraction increases from 46.2% to 93.1%. Coastal zone validation shows a 23.8-31.6% reduction in the false alarm rate and a corresponding 27.133.3% increase in the cloud detection true skill score. Globally the robust standard deviation for clear-sky matches between SLSTR-A and drifting buoys reduces from 0.3 to 0.29 with a 6% reduction in data due to improved screening of scattered cloud.

Automated summary

This paper presents new approaches to utilizing reflectance imagery for sea surface temperature (SST) remote sensing and improving cloud detection in infrared images. By averaging and calculating the standard deviation of the nearest reflectance observations, the authors were able to enhance the discrimination of clouds in the infrared image. The results show significant improvements in coastal areas and around ocean fronts, leading to increased accuracy and coverage of SST measurements.