831 
As mentioned in the Introduction, the split-window approach is currently utilized to produce 
surface temperature retrievals in terms of a linear combination of brightness temperatures measured in AVHRR 
Channels 4 and 5, according to, T s = T 4 + a (T 4 -T 5 ) + p, where a and P are the split-window coefficienst, 
which depend only on spectral emissivities and not on atmospheric conditions (Becker and Li, 1990). However 
this is only true for a relatively dry atmosphere since the linear approximations with respect to water vapor 
content introduced in the derivation of local split-window method (Becker 1987) are true in this case but not in 
case of a wet atmosphere. In order to correct for the nonlinear effects and make the split-window method 
applicable to most atmospheric situations encountered in the world, the split-window coefficients could be varied 
with the classes of atmospheric humidity. This can be made, dividing the 60 atmospheres extracted from TIGR 
(see above section) into 4 classes with respect to channel 5 transmission ( 15 ), and using a least-squares fit 
method for each class. In this way the coefficients for each class of atmosphere can be obtained, leading to an 
improved split-window with water vapour corrections (Li et al., 1994). 
To use this improved split window method, we have to classify the atmosphere, with the help 
of the covariance and the variance of brightness temperatures measured in channels 4 and 5 of AVHRR. It should 
be noted that, if emissivity is known, using any of the more recent split-window algorithms (Sobrino et al., 
1993; Coll et al., 1994, Li et al., 1994) one can obtain the land surface temperature to an accuracy of 1 K. 
5-CONCLUSIONS 
For both prediction and detection, the general circulation models need ST data that cannot be given by the 
traditional techniques. The main problems are the lack of measurements and the low accuracy of the 
observations. Substantial differences exist between coincident observations from VOP (Voluntary Observing 
Program) ships of the World Meteorological Organization (WMO) and from NOAA moored buoys even when 
they are in close proximity to one another (Wilkerson and Earle, 1990). To solve for these shortcomings the best 
solution is to obtain this data (ST) from thermal infrared satellite imagery. However this constitutes a difficult 
task that has been severely limited because of the difficulty in evaluating the effects of the atmosphere. The 
present paper should help solving this drawback through the development of a technique that permits modeling 
the transmittance ratio, the channel atmospheric transmittance and the total water vapor from brightness 
temperatures measured at Channels 4 and 5 of AVHRR on board NOAA-11 satellite alone. The technique relies 
on the simple condition that the atmosphere is unchanged over the neighboring points where the surface 
temperature and emissivity changes. This permits derive a split-window algorithm which results in an 
improvement over the standard algorithms for relating ST and the brightness temperatures measured by the 
satellite borne sensor (T 4 and T 5 ). 
Although it is clear that even more statistics will be necessary to test the technique, the 
applications to real and simulated data given in this paper show that the technique is promising and could be 
implemented in a production environment necessary to document global climate change. 
ACKNOWLEDGMENTS 
The authors wish to express their gratitude to the Laboratoire de Météorologie Dynamique (Paris, France) for 
supplying the TIGR data sets, and to the Air Force Geophysics Laboratory (Massachusetts, USA) for supplying 
the LOWTRAN-7 computer code. We also thank M. Arbelo for providing the TOVS data and C. Badenas for 
assistance in the processing of AVHRR data. The work was supported in part by the Commission of the 
European Communities (Projects No. EV5V-CT91-0033/0035). 
references 
Barton I. J., Zavody A. M., O’Brien D. M., Cutten D. R„ Saunders R. W. and Llewellyn-Jones T., 1989. 
Theoretical algorithms for satellite-derived sea surface temperatures. J. Geophys. Res. 94: 3365-3375. 
Becker F., 1987. The impact of spectral emissivity on the measurement of land surface temperature from a 
satellite. Int. J. Remote Sens. 8 : 1509-1522. 
Becker F., Li Z.-L., 1990. Towards a local split window method over land surfaces. Int. J. Remote Sens. 
11: 369-394. 
Chesters D„ Uccellini L. W., and Robinson W. D., 1983. Low level water vapor fields from the VISSR 
atmospheric sounder (VAS) split-window channels. J. Climate Appl. Meteorol., 22: 725-743. 
Coll C„ Caselles V., Sobrino J. A., and Valor E„ 1994. On the atmospheric dependence of the split- 
window equation for land surface temperature. Int. J. Remote Sens., (in the press). 
Dalu G., 1987. Satellite remote sensing of atmospheric water vapor. Int. J. Remote Sens. 7: 1089-1097. 
Deschamps P.Y., and Phulpin T., 1980. Atmospheric correction of infrared measurements of sea surface 
temperature using channels at 3.7, 11 and 12 pm. Bound. Layer Met. 18: 131-143.