ratur. 30 
Tempe 
i 7,86 
7 3 
        
  
Schnee/bebaut 
Bd Hasser 
  
10 11 12 19 14 15 18 17 18 19 20 21 
x30m 
in "tr 
  
E 
zz 
ET 
  
  
Fig. 1. Surface radiation temperature, resolution enhanced to 30x30m. 
gradient between the station values at Abisko and the 
upper limit of the boundary layer. It is assumed that 
temperature decreases with elevation in a linear way. 
This assumption, however, is supported by the linear 
form of the temperature and vapour pressure curve over 
Lulea. 
Mannstein (Mannstein, 1991) did not consider the 
modification of air density due to the higher temperature 
near the surface and the vapor content of the air. The 
author calculated air density as used in formula (5) from 
arithmetic means between atmosphere- and surface 
temperature and the complete formula, considering vapor 
pressure. 
Vapor pressure was modeled in the same way as air 
temperature, as the profiles showed similar 
characteristics. 
8. SENSIBLE HEAT FLUX MODELLING 
The image processing program WAERME (Storl, 1992) 
developed by the author uses as input four images with 
the parameters elevation, inclination, land use class, 
roughness length and surface temperature. Furthermore, 
a tabular file containing the radio sonde values for air 
temperature, air- and vapour pressure in an interval of 
1m is required. 
The program reads for each pixel elevation, inclination, 
surface temperature, land use class and roughness 
length and accesses the air temperature, air- and vapour 
pressure values for the corresponding elevation. 
The nucleus of the program calculates potential 
temperature of air and surface first, so that the difference 
A, can be derived. The difference A between potential 
surface temperature at the height n+zo and free 
atmosphere, as required by Brehm's model, is derived by 
an iterative algorithm. First, the unknown temperature 
difference A is set to 0,5 x As (where As is the 
temperature difference between radiance temperature 
and temperature of free atmosphere). R is calculated 
according to equation (3), and c, and n are derived from 
Rg and a, according to the functional dependance 
between these parameters given by Brehm's results. Ag 
is then calculated by equation (7). If Aq*A is bigger than 
As, À is reduced by a fraction of itself, and Ro, cg, n and 
Ad are repeatedly calculated, until equation (8) holds and 
the difference A is known. 
The coefficients c, and n have to be derived for each 
pixel depending on soil rosby number (as output from the 
iterative procedure) and slope angle. 
664 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996