Agricultural and Forest Meteorology 100 2000 127–136
Modelling global radiation in complex terrain: comparing two statistical approaches
Helfried Scheifinger
a,∗
, Helga Kromp-Kolb
b
a
Central Institute for Meteorology and Geodynamics, Hohe Warte 38, A – 1190 Vienna, Austria
b
Institute for Meteorology and Physics, University of Agriculture, Türkenschanzstraße 18, A – 1180 Vienna, Austria Received 1 June 1999; received in revised form 27 September 1999; accepted 7 October 1999
Abstract
Two simple approaches for assessing global radiation in complex terrain are tested and compared. A parameterisation scheme for global radiation based on cloud cover observations was compared with interpolation of measured global radiation
values from the Austrian climate observation network. Interpolation appears to be a useful method for a station density which has been available after 1992 in Austria about 1000 km
2
station. In that case interpolation is superior to parameterisation. The quality of interpolated data quickly drops with height and for elevations above 1500 m neither method delivers useful
results. ©2000 Elsevier Science B.V. All rights reserved.
Keywords: Global radiation; Atmospheric transmittance; Parameterisation; Interpolation; Complex terrain
1. Introduction
The atmospheric environment constitutes a deci- sive factor controlling ecosystem processes. Most pro-
cesses in the atmosphere and biosphere, such as evap- oration, sensible heat flux, soil heat flux, photosynthe-
sis or transpiration, are driven directly or indirectly by solar radiation. Many of them have been investigated
and modelled in recent years by various disciplines not only in flat but also in complex terrain, where so-
lar radiation constitutes a basic input.
In order to investigate and quantify possible rela- tions between temporal and spatial distributions of
ecosystem and atmospheric variables in complex ter- rain, the temporal and spatial resolution of both vari-
∗
Corresponding author. Tel.: +43-1-36026; fax: +43-1-36026-74.
E-mail address: helfried.scheifingerzamg.ac.at H. Scheifinger.
able sets must match. Because of the high spatial vari- ability of ecosystems in complex terrain, atmospheric
information with a high spatial resolution is often seen as a prerequisite for ecological research there. Some-
times it is not easy to acquire the atmospheric data with the necessary spatial resolution in remote areas.
Putting up and running a fully instrumented meteo- rological station can be costly. Points of investigation
can be spread across a huge area, and not at each of them all atmospheric variables can be measured. If
the measurement of only a selected set of atmospheric variables can be afforded, information about others
which are also important, might be still lacking. In such situations one has to look for other sources of
information: 1. Luckily most countries are running meteorological
networks, which provide information about the at- mospheric environment to be utilised. A straight-
forward solution would be to extrapolate and use
0168-192300 – see front matter ©2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 1 9 2 3 9 9 0 0 1 4 1 - 0
128 H. Scheifinger, H. Kromp-Kolb Agricultural and Forest Meteorology 100 2000 127–136
the data from the nearest meteorological station. Because of the horizontal and vertical distance be-
tween the meteorological station and the place of interest, such a procedure can be dangerous and can
deliver unsatisfying results. Unfortunately ecosys- tem research focuses often on such remote places
where data are scarce. National climate observation networks have not been designed to supply data
with a high spatial resolution in remote alpine ar- eas. The more one moves from flat and inhabited
areas to remote and complex terrain the more sta- tion density drops. Station density is also rapidly
decreasing with increasing height.
2. In many cases a strong relationship between the desired atmospheric variable and topogra-
phy can be found. Topographical information can be available more readily and with a much
higher spatial resolution than for any atmospheric variable.
3. A third source of information can be found in the observation that atmospheric variables do not
behave independently. One would have to select proper independent atmospheric variables which
allow a tight relationship with the desired depen- dent variable. In case the independent atmospheric
variable is measured with a higher spatial density than the dependent variable, additional spatial in-
formation for the dependent atmospheric variable can be deduced. Where radiation is concerned,
sunshine duration, cloudiness, degree hours of temperature, relative humidity, precipitable wa-
ter content, and composition, concentration and size distribution of aerosols are among the in-
dependent variables which have been used for parameterisation. The usefulness of a parameteri-
sation scheme depends very much on the spatial behaviour of the chosen variables. Some authors,
for instance, selected temperature and precipita- tion as independent variables which can cause
problems, if applied in complex terrain. The pa- rameterisation schemes might work at the few
places where they were developed for but not necessarily at other sites. In order to apply a pa-
rameterisation scheme as it was developed by Bristow and Campbell 1984 or Lexer 1997,
empirical coefficients would have to be interpo- lated in complex terrain. As the spatial behaviour
of temperature and precipitation in complex terrain is as complex as the terrain, such an
approach appears impractical. 4. As a fourth point, general knowledge about the
physics of an atmospheric variable can add substan- tial information: laws describing the relative posi-
tion of the sun on a tangential plane on the earth’s surface and shading through topographical features
as a function of time help to calculate global radi- ation at a specific point in complex terrain. Or the
idea of global radiation as a sum of direct radiation, diffuse radiation and radiation reflected from the
earth’s surface greatly supports the reconstruction of global radiation from interpolated transmittance
values.
This work aims at a straightforward and easily ap- plicable procedure to produce radiation data in com-
plex terrain based on the four above mentioned sources of information. Although the procedures which will
be introduced show certain limitations, they can nev- ertheless prove their usefulness for a number of appli-
cations.
As it was impossible to find a suitable procedure which could solve the stated problems, two approaches
were developed and compared. One was the interpo- lation of measured global radiation sums daily and
monthly values via atmospheric transmittance and the other, a parameterisation scheme with subsequent in-
terpolation of the parameterised atmospheric transmit- tances. For parameterisation, cloud cover observations
and the height of the cloud observing station were selected as independent parameters.
Comparing advantages and disadvantages of the pa- rameterisation and interpolation approach, following
factors have to be considered: 1. Station density for global radiation measurements
more than 1000 km
2
station in Austria. 2. Station density for cloud observations less than
300 km
2
station in Austria. 3. Quality loss through parameterisation.
4. Quality loss through interpolation. Parameterisation has a higher station density at its
disposal, but suffers quality loss through a two-step process including parameterisation and interpolation.
Interpolation of observed global radiation has to cope with a station density which is about a third of that for
cloud observations, but saves the quality reducing step of parameterisation. The quality of both procedures
will be compared quantitatively at each step.
H. Scheifinger, H. Kromp-Kolb Agricultural and Forest Meteorology 100 2000 127–136 129
2. Radiation in complex terrain