ABSTRACT
Introduction Remote sensing in the thermal infrared (TIR) bands provides a means of measuring both surface emissivity and surface temperature. Measurements of these two parameters are very useful in many applications, particularly those related to climate studies, for the following reasons: (1) surface temperature is not only a good indicator of both the energy equilibrium of
However, besides the effects of the atmosphere, which occur in all the spectral bands, there are several problems, specific to TIR bands, that need to be solved before a relevant quantitative interpretation of the data can be performed. Firstly, as it is well known, the radiance, R;, measured from space depends on both spectral emissivity and surface temperature. Therefore it is generally not possible to interpret a variation of radiance in the TIR bands in terms either of a temperature variation or of an emissivity variation, as the separation of these two contributions is a complicated task. Secondly, the strong vertical and horizontal heterogeneity of the land surface, at the scale of a pixel, makes it difficult to define effective temperatures and
emissivities, as has been shown by Becker (1981) and Caselles and Sobrino (1989). In these cases the surface temperature Ts can be defined physically as:
where Bj1 is the inverse of the Planck function in channel i, Ra,n is the downwelling atmospheric radiance, and £-, the emissivity of an heterogeneous pixel in channel i. This definition of surface temperature is used in the following discussion.
