Land Surface Temperature (LST) Retrieval from GOES Satellite Observations - Satellite-based Applications on Climate Change - page 283

The algorithms discussed above are physically based, starting from a theoretical
description of the important radiative transfer processes at the surface and in the
atmosphere and ending in a regression relation with coefficients that are adjustable,
or at least, changing with surface and atmospheric parameters. All of these split-
window approaches utilize the 11- and 12-

μ

m window channels and assume

constant surface emissivities.
Bates and Smith (1985) included the 3.9-

μ

m channel of GOES-5 in SST retrieval
from the split-window channels, and were able to reduce the SST retrieval error.
May ( 1993 ) introduced three-channel nighttime SST algorithm to retrieve SST
from the NOAA/AVHRR imager during nighttime and demonstrated improved
accuracy relative to the split-window algorithm. The transmittance of radiant
energy from the surface through the atmosphere is greater for AVHRR channel 3
(3.75

m) than for channels 4 and 5. This fact results in less atmospheric attenuation
effects in channel 3 data, providing a more accurate SST retrieval when all the three
thermal channels are used. However, this channel has not been used to retrieve LST
and as yet no three-channel LST algorithm has been developed. Channel 3 can be
used only at night because this wavelength contains reflected solar energy during
the daytime, unless a correction to the solar contamination is adopted. Brown et al.
( 1996 ) investigated the possibility of correcting the solar contamination existing in
the MID-IR channels during daytime for the bands of MODIS (Moderate-resolution
Imaging Spectroradiometer). These studies are in progress and results are being
evaluated.
Most studies on LST have focused on the use of polar-orbiting satellite systems,
such as EOS/MODIS and NOAA/AVHRR. The temporal measurement frequency
of the polar-orbiting satellite instruments is approximately two times per day. This
sparse temporal sampling is inadequate to capture the LST diurnal cycle. Moreover,
the NOAA satellites are not strictly sun-synchronous, implying that a drift in time
of measurement may exist at a given location (Gutman 1992; Gleason et al. 2002).
Less work has been done with geostationary satellites. Prata (1999) investigated
LST retrieval from the Japanese Geostationary Meteorological Satellite 5 (GMS-5)
using a split-window LST algorithm. However, the GMS have only two window
channels available for LST retrieval. Faysash and Smith ( 1999 , 2000 )proposeda
simultaneous retrieval of LST and surface emissivity using MODTRAN; however,
radiative transfer models (RTM) are very time consuming, and therefore, not well
suited for large data sets and operational use. Several studies of LST retrieval from
METEOSAT of the European Meteorological Satellite Programme (EUMETSAT)
have been performed (Morcrette 1991; Olesen et al. 1995; Hay and Lennon 1999;
Cresswell et al. 1999; Schadlich et al. 2001; Gottsche and Olesen 2001; Dash et al.
2002). They deal with modeling the brightness temperature of METEOSAT, studying
ground height effects on LST, estimating air temperature from LST, and applications
of LST in disease studies. Recently, a four-channel LST algorithm has been devel-
oped by Sun and Pinker (2007) for the METEOSAT second-generation imager
SEVIRI.

μ