Coastal-zone sea surface temperature at ~100 m resolution

This presentation was held at the 24^th International SST Users’ Symposium and GHRSST Science Team Meeting (GHRSST24) held in Ahmedabad, India/Online from 16 – 20 October 2023.

Explore the full program at GHRSST24 on the GHRSST Website here.

Abstract

Sea surface temperature observation in highly dynamic and spatially variable coastal zones will be advanced by future high-resolution (<100 m) thermal remote sensing missions. Meanwhile, the Landsat missions offer opportunity to prepare for these later missions that will bring finer resolution, lower noise, more thermal channels and more frequent coverage. We adapt well-established sea surface temperature processing applied to global meteorological-type sensors to Landsat observations. Cloud detection and temperature retrieval are the two principal steps, along with quality-flag derivation and uncertainty estimation. Bayesian cloud detection based on thermal and reflectance channels is expected to be effective, as the approach is by design generic and adaptable given a suitable radiative transfer model. Use of a channel set as would be used for processing (for example) an AVHRR-like sensor confirms this. Points of difference when classifying high resolution imagery for sea surface temperature retrieval are the need for finer land-sea boundary auxiliary information and the greater importance of tides in changing the land-sea boundary. For Landsat specifically, the available (regridded) thermal data “bleed” across high temperature contrasts, making assessment of the cloud edges more ambiguous. Similarly to Bayesian cloud detection, optimal estimation used for the temperature retrieval is a highly transferrable technique, reliably and rapidly giving good results when applied to new instruments. Simulation studies are a useful means of assessing optimal estimation in comparison to other temperature retrieval options. In the application to Landsat, the calibration is a significant source of error in retrieved surface temperature. 

Results for cloud detection and sea surface temperature will be presented, and discussed in terms of lessons towards future missions that will have smaller errors and enable more frequent observation, first of all TRISHNA.