The retrieval methdodology for the operational AATSR LST product is described in the original AATSR LST Algorithm Theoretical Basis Document: Several restrictions imposed by the near-real time processor have affected the choice of algorithm. These are: (1) the algorithm must not utilise the visible channels of the AATSR, (2) the algorithm must be computationally fast, must not occupy large amounts of computer memory and must not utilise large amounts of disk space, (3) a method for cloud detection over the land must be implemented, and (4) a protocol for validation of the product must exist. The algorithm uses pixel-by-pixel top-of-the-atmosphere cloud-free, calibrated and navigated day and night brightness temperatures from the 11 and 12 µm AATSR channels to produce a global LST product. Additional seasonally-dependent land cover classiﬁcation, fractional vegetation and precipitable water data are required for the operation of the algorithm. The algorithm is fast and can be used globally by employing regression coefﬁcients in a look-up table that is updatable.
Efforts are currently underway to improve the algorithm by using auxiliary data with higher spatial resolution. This includes both a new dataset for global biome distribution and a new dataset for fractional vegetation cover.