RETRIEVAL OF TEMPERATURE AND EMISSIVITY OF DUNE FIELDS USING LANDSAT8 AND QUARTZ SPECTRAL LIBRARY IN AN ITERATIVE APPROACH

Retrieval of temperature emissivity (TE) for targets using
thermal infrared (TIR) data is of great importance in
climatology, urban heat island analysis, geological studies,
and many more. In TIR, water vapor content is the largest
source of errors in radiance data (common in humid climate)
requiring an efficient atmospheric correction. Another source
of error is that the radiance is a function (depend) of the TE
of the target, making this an indeterminate problem. The
objective of this work is to propose a method to recover TE
for quartz from its radiance spectrum measured in a
laboratory for mitigating the indeterminate problem. For this
purpose, we used linear regression models in radiance
laboratory spectrum quartz data, applied to LANDSAT 8
image. The results showed that it is possible to recover the
TE with errors of 0.179K in temperature and 0.0031 in
emissivity when compared to other classical methods referred
as Reference Channel Method (RCM), and at the same time
mitigating the indeterminate problem.