Jher plants
S assumed
le classes.
e for each
SS Variance
taset was
orl, 1992,
value of
Yerefore,
] its
'ere taken
my of
'd deviation
| enhanced
data, using
lerived from
jet equation
(9)
ation
transformed
j long wave
Short wave
centage of
assumed à
it heat flux
, for each
1e following
(10)
pi percentage of land use class i in the pixel
d .. regression constant, containing all factors not
considered by the other parameters, like
changing Bowen Ratio within land use classes.
The right side of the equation consists of parameters that
were available in high resolution for the study area. The
land use classification, derived from the 30x30m
LANDSAT channels, a terrain model of 30x30m and a
model of solar irradiation based on the terrain model
were available.
7.2.2 Extended regression model
The parameterization of Scherer assumes, that the
Bowen Ratio, i.e., the ratio between sensible and latent
heat flux, can be assumed as constant within one
vegetation class. This simplifies the real conditions in so
fat as the Bowen Ratio evidently clusters around a
characteristic value for each class, but shows strong
dependence on water supply and on interchange
resistance. Furthermore, the Bowen Ratio is controlled by
parameters like air temperature and water vapor
saturation deficit.
As soil moisture controls evapotranspiration by
influencing stomata apertures, it has no negligible
influence on surface temperature, hence long wave
radiation and the Bowen Ratio. Therefore, soil moisture
must be considered in modelling energy budget
components.
Analyzing the interdepandence between the energy
budget parameters, it can be shown, that with time- site-
and vegetation-specific energy input Es J(1-0)+E; J. the
surface temperature and hence long wave radiation and
Bowen ratio in first approximation are controlled only by
the interchange resistance dependant on stand geometry,
wind velocity and soil moisture (Storl, 1992). Because of
this interdependance between soil moisture and surface
temperature, a spectral soil moisture index partly
explains the variance of long wave emission within the
vegetation classes.
Storl (Storl, 1992, 1993, 1994) derived a spectral soil
moisture index from TM data for the study area that
could be integrated into the regression model and
improved significantly variance explanation. The
regression equation was therefore written as:
n
Eft *aE,| * bh *c SM *X dip; + e (11)
SM .. soil moisture index
The regression coefficients calculated with the image
Processing module GLOREG (Scherer, 1987,
Parlow/Scherer, 1991, Storl 1994) for global multiple
linear regression can be interpreted as a measure for the
contribution of each data set to the reduction or elevation
of long wave radiation and hence surface temperature
(Parlow/Scherer, 1991). Therefore, radiance
temperatures were transformed to longwave radiation
values, using the Stephan-Boltzmann equation:
E so 1^ (12)
663
Where T is the radiation temperature.
In a further step, long wave radiation with 30x30m
resolution was calculated, using the derived regression
coefficients and applying equation (11). The radiation
temperature was then calculated using the Stephan-
Boltzmann-equation (12).
Fig. 1 shows the resolution enhanced radiation
temperature of the surface. These thermal data contain
information about soil moisture, as far as it influences
surface temperature and establish therefore an ideal
dataset to model sensible heat flux and
evapotranspiration.
7.3 Temperature of free atmosphere
The meteorological situation at the day of satellite
overpass was dominated by high pressure conditions
with Abisko Station at the center of the high pressure
area. Wind velocities ranged between 1,5 and 2,5m/s at
Abisko Station at 9h20, the time of satellite overpass.
The 10-minute wind measurements clearly show, that the
wind direction follows the sun in order to form a well
developed local slope/sea wind system perpendicular to
the shores of lake Abisko and the southern slopes of
Mount Njulla.
The measurements of a radio sonde from Kallax/Lulea,
at 100km from the study site were used to interpolate the
temperautes of free atmosphere.
The synoptic weather charts show, that Lulea and the
study area are under the influence of the same air
masses under high pressure. The potential virtual
temperatures from the sondage values confirm, that
stable atmospheric conditions prevailed, one of the
conditions for applying Brehm 's slope wind model.
The uncorrected air pressure, temperature and vapor
pressure values of the radio sonde from Lulea,
interpolated by the program SONDE (Storl, 1992),
showed 967,85 hPa, 12,96 °C and 8,84 hPa respectively
for the elevation of Abisko at 385m above sea. The radio
sonde started at 12hOO am, whereas the satellite
passed at 9h20 am. At Abisko, air temperature rose from
9,13 °C to 10.43 °C in this interval of time, which makes
a difference due to time of 1,3 °C. If the temperature
measured by the radio sonde at Lulea is reduced by 1,3 °
C, according to the time difference, a hypothetic air
temperature at Abisko Station of 11,66 °C at 9h20 is
derived. This slightly higher value (2,56 °C) than the
temperature of 9,1 °C measured at Abisko station is
probably due to the warming effect of the lower
landmasses of Lulea and the cooling effect of advected
air from Lake Abisko.
It cannot be theoretically justified to simply reduce the
temperature profile of the radio sonde by 2,56 °C and
adapt it hereby to the measured temperature values at
Abisko of 9,1 °C. As air masses at Lulea are warmed up
from the underlying land masses, the heating effect is
reduced with height, and it can be assumed that the
temperature at the upper limit of the mixture layer at
2000-3000m behaves very constant over time and space.
Therefore, the radio sonde value for this elevation is
taken as a reference in order to model the temperature
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996