Estimating Evapotranspiration within the Colorado 
Alpine Tundra with Landsat Thematic Mapper 
Claude R. Duguay 
Laboratory for Earth Observation and Information Systems 
Department of Geography 
University of Ottawa 
Ottawa, Canada, K1N 6N5 
ABSTRACT 
Evapotranspiration (ET) is a key element in climate related studies on all spatial and temporal scales. Recent studies 
have shown that ET can be estimated with some degree of precision from meteorological satellites, over flat terrain, 
using semi-empirical and analytical models. However, no method has been proposed in order to derive this parameter 
in mountainous terrain by combining remotely sensed imagery with ancillary data. This can be explained in part by 
the difficulties in estimating many of the relevant parameters which control ET rates, namely the radiation balance, 
wind speed and the aerodynamic resistance. In this paper, an approach is proposed for estimating ET in a high relief 
environment with Landsat Thematic Mapper imagery and digital terrain data. Preliminary results in the computation 
of the net shortwave radiation in the alpine tundra of Niwot Ridge (Colorado Front Range, U.S.A.) suggest possible 
solutions to the problem of estimating of ET in mountainous terrain using remotely sensed data. 
Key Words: Evapotranspiration, net radiation, alpine tundra, Landsat Thematic Mapper, digital terrain data. 
INTRODUCTION 
Developments in the study of the radiation balance and 
energy balance of the 1950s - 1960s (e.g. Pennman, 
1956) have enabled, among other things, to establish 
relationships between evapotranspiration (ET) and surface 
temperature. Equations used to estimate ET include 
different parameterizations based on the type of 
evaporating system (land cover type), its 
thermodynamical state (temperature and moisture), its 
aerodynamical behaviour (roughness and structure) and 
the atmospheric state (temperature, moisture, wind and 
stability conditions) (Becker et al, 1987) These 
parameterizations depend on the time and space scales at 
which the measurements are made, because of the non- 
linearity of the processes involved and the heterogeneity 
of the earth's surface characteristics. 
Models which were solely based on point measurements 
can now benefit from the spatial and temporal resolution 
offered by current meteorological and earth resources 
satellites. Satellite-borne sensors that record radiation 
reflected and emitted from the earth's surface are 
currently used in order to derive the surface parameters 
necessary for the determination of regional ET. Indeed, 
recent studies have shown that ET can be determined 
from complex physical models using satellite imagery 
(e.g. Meteosat and HCMM) and meteorological 
parameters (air temperature and moisture, wind speed, 
etc.). These range from statistical semi-empirical 
formulations (Idso et al., 1975; Séguin et Itier, 1983) to 
analytical and numerical methods based on sophisticated 
physical models of heat and mass transfer (Carlson et al., 
1981; Taconet et al, 1986; Abdellaoui et al, 1986). 
Results of these studies suggest that at the regional scale 
the parameterizations reproduce the general behaviour of 
experimental data taken at local scale (field 
measurements), but representative of a larger area (over 
flat terrain). 
In mountainous terrain, however, point measurements are . 
generally not adequate for representing the parameters 
required for obtaining reasonable estimates of ET over 
large areas. The nature of mountain terrain sets up such 
a variety of local weather conditions such that any point 
measurement is likely to be representative of only a 
limited range of sites. Here, point measurements of wind 
speed, air moisture and temperature need to be extended 
over surfaces of varying slope and aspect. This of 
course adds more complexity to the problem of 
estimating ET from remotely sensed data. Models based 
on extensive field campaigns (local scale) are giving us 
new insights into the determination of meteorological 
parameters over rugged terrain. In a recent study, Isard 
and Belding (1988) have shown that reasonable estimates 
of ET (local scale) could be obtained from the mid- 
latitude alpine tundra of Colorado provided that daily 
inputs of surface net radiation and ground heat flux were 
available. This paper briefly describes ongoing work 
towards the development of a model for estimating ET 
from the alpine tundra of Colorado using Landsat 
Thematic Mapper (TM) imagery and digital terrain data. 
METHODS 
Study Area 
The study area is located in the Indian Peaks section of 
the Colorado Rocky Mountain Front Range, and is one 
of the University of Colorado Long-term Ecological 
914