| Summary: | Short abstract Landsat recently released its Collection 2 Level 2 Surface Temperature product, marking the first comprehensive calculations of surface temperatures from the four-decade-long Landsat mission sequence. Producing surface temperatures requires complex integration of Landsat imagery with external atmospheric datasets and model outputs, to compensate for the lack of dual thermal bands (Landsat 4/5/7) and other bands required for atmospheric correction (Landsat 4/5/7/8/9). Although this work provides a reliable surface temperature product, gaps still exist for sea surface temperature (SST) retrievals: the algorithm is optimised for acquiring land surface temperatures (i.e., not SST), and surface temperatures are not produced at night or around Antarctica. Here, we develop Landsat single-channel and split-window SST algorithms that will allow for integration with GHRSST products and will also be developed as an on-demand, cloud-based, user-customizable data product. The single channel algorithm uses coincident atmospheric profiles from reanalysis data for temperature, relative humidity, and geopotential height as inputs into a radiative transfer model to account for the atmospheric effects on thermal retrievals. The Non-Linear SST algorithm for split-window Landsat data (Landsat 8/9) will use the Canadian Meteorological Center Global Foundation SST product as a SST reference, and radiative transfer model-based simulations of at-sensor brightness temperatures to derive the algorithm coefficients. Our cloud-based workflow—modelled after the ICESat-2 SlideRule project—will provide a framework for on-demand data product generation and serving, allowing users to specify algorithms and atmospheric data inputs and models to retrieve Landsat SSTs optimised for their scientific needs.
|
|---|