In this study, a highly differenciated land use
classification had been calculated with radiation
corrected data (Parlow, 1988, 1991, 1992) and was
found to be coherent with the field mapping realized by
the author (Storl, 1992).
For a modelling of sensible heat flux on micro scale, the
intraclass variance of roughness length has to be taken
into account and must be adecuately parametrized.
The author could dispose of a PAR-albedo dataset
(Photosynthetically Active Radiation albedo). PAR-
albedo is the albedo for radiation in the
photosynthetically active region of the spectra, and
therefore an excellent biomass parameter that reflects
vegetation density much better than NDVI, as has been
shown by Parlow (Parlow, 1988).
In can be assumed, that PAR-albedo is correlated to
stand height within a certain vegetation class, as
absorption of photosynthetically active radiation is related
to biomass, and more biomass means higher plants
within one vegetation class in general.
As a first approximation, a linear relation was assumed
between plant height and PAR-albedo within the classes.
In order to obtain a roughness length value for each
30x30m pixel, the information about intraclass variance
variation inherent in the PAR-albedo dataset was
integrated with the land use classification.
An image processing procedure (CSCALE, Storl, 1992,
1993, 1994) was written in order to scale each value of
PAR-albedo according to its land use class. Therefore,
representative values for roughness length and its
standard deviation for each vegetation class were taken
from the data accumulated at the Royal Academy of
Science in Abisko.
Tab.1 shows the medium values and standard deviation
of roughness length used in meters.
very dry heath |dry heath moist heath meadow | snow swamp | salix
0,010 0,010 0,020 0,008 0,001 0,010 0,040
0,005 0,005 0,005 0,002 0,000 0,000 0,020
old birch forest | heath birch forest | moist heath birch forest | constr. water
0,500 0,400 0,600 0,900 0,001
0,200 0,200 0,200 0,000 0,000
Tab. 1. Mean roughness length and standard deviation for the land use classes
Conditioned scaling was applied to the PAR-albedo
dataset, with the land use classes as conditioning
parameters. For each land use class, CSCALE
calculated the medium value and the standard deviation
of PAR-albedo. In a second pass through the images,
CSCALE transformed the PAR-albdo values in a linear
way accordingly, assigning pixels with mean PAR-albedo
value for the respective land use class the mean
roughness length of the class, and transforming PAR-
albedo values that varied about standard deviation from
the mean values into roughness length values within the
roughness lenth statistics of the respective class.
7.2 Surface temperature
The thermal channel 6 of Landsat TM with a spatial
resolution of 120m had been calibrated by Parlow
(Parlow, 1988) to radiance surface temperature over the
sea surface of lake Abisko using a calibration formula
suggested by Schott and Volchok (Schott/Volchok,
1985). The atmospheric influence was eliminated using
an improved version of the WINDOW model of J. Price,
NASA (WINDHA, Price, 1987 / ATMKOR, Scherer, 1987)
7.2.1 Resolution enhancement
The channels 1-5 and 7 of Landsat-TM have a spatial
resolution of 30x30m, whereas the thermal channel 6 is
restricted to a resolution of 120x120m.
Scherer (Scherer, 1987) developed a method to improve
the spatial resolution of thermal data, integrating data
sets of higher resolution. He obtained an enhanced
spatial resolution of remotely sensed thermal data, using
a multiple linear regression model, which he derived from
the energy budget equation. The energy budget equation
under equilibrium conditions can be written as:
ESL (oo) * EN - Fit * oO "H*LE-O (9)
a .. integral albedo
Es ... short wave solar irradiation
Ej .. long wave atmospheric counter irradiation
Ej .. long wave terrestrial radiation
G .. soil heat flux
H .. sensible heat flux
LE ... latent heat flux (evapotranspiration)
Scherer showed, that equation (9) can be transformed
with permissible approximations, representing long wave
radiation as a linear combination of short wave
irradiation, terrain elevation and the percentage of
different land use classes in a pixel. He assumed a
constant ratio between sensible and latent heat flux
(Bowen Ratio) within each class. Hence, for each
120x120m thermal pixel of Landsat-TM, the following
equation can be written (Scherer, 1987):
n 0
Ep =a Es) +bh+E cip; + d (10)
i=1
Es .. short wave solar irradiation
h .. terrain elevation
662
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996
ua - ea gemmam "
l2, 0 Dll DU a D aD hn gg w= rv i
cn
<<‘ OEY (9 — "TITS ee