Methods to retrieve LST from satellite infrared measurements
From the previous chapter we can conclude that the major difficulty when measuring LST is the uncertainty about land surface emissivity. The different algorithms that have been proposed to solve the RTE can be divided in two distinct categories:
- Land Surface Emissivities (LSEs) are assumed to be known beforehand
- LSEs are not known in advance.
Fig. 10: Methods to estimate LST assuming surface emissivity is known in advance are classified in single channel, multi-channel and multi-angle methods.
If the emissivity is assumed to be known in advance, the RTE equation is solved in order to retrieve the LST. There are three different ways of solving these equations: single channel, multi-channel and multi-angle methods (Figure 10).
If no emissivity value is known in advance, the purpose of the algorithms is to solve the RTE for the two unknowns (LST and LSE). They can be classified as: stepwise retrieval methods, simultaneous retrieval of LST and LSE with known atmospheric information, and simultaneous retrieval of LST and LSE with unknown atmospheric information (Figure 11).
Fig. 11: Methods to estimate LST assuming surface emissivity is not known in advance: stepwise retrieval methods, simultaneous retrieval of LST and LSE with known atmospheric information, and simultaneous retrieval of LST and LSE with unknown atmospheric information.
The list of the advantages and drawbacks of these methods is quite extensive, and allow different accuracy values on LST retrievals. However, the choice of approach cannot be based solely on accuracy. The choice is governed by many factors, such as the availability and quality of the input data, the geographical area of application and allowed computation time. With this in mind, it should be mentioned that the methods that try to solve the LSE and LST simultaneously are computationally very heavy, and therefore not suitable for operational LST retrievals from a high temporal frequency satellite sensor, such as MSG's SEVIRI.
The next section provides a brief description of the different methods with examples of some reference algorithms. A detailed description of the methods and their assumptions, strengths and disadvantages, is found in Zhao-Liang Li et. al (2013).