Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986 
525 
Comparison of SPOT-simulated and Landsat 5 TM imagery 
in vegetation mapping 
H.T0mmervik 
Institute for Biology and Geology, University of Tromse, Norway ' 
ABSTRACT: In this paper I will present some results from work done on both SPOT-simulated imagery and Landsat 5 
TM imagery. The testarea is situated in Central Troms, Northern Norway, and has a fairly wide variation in eco 
logical niches and comprises both forest landscape and mountain areas (Attic alpine vegetation). The ground 
truth program did include both a traditional vegetation mapping program and a test-site program. Digital image 
processing was carried out both on the SPOT-simulted imagery and the Landsat imageries, but the scenes were 
taken too early in the spring or too late in the autumn, to give a sufficient basis for a good classification. 
The investigation has shown that the Landsat scene from the springtime almost had the same accuracy by super 
vised classification as the SPOT-simulated imagery, due to the better radiometric resolution for the TM-sensor 
compared to the simulated HRV-sensor on the SPOT satellite. But classification on the SPOT-simulated imagery 
showed that the vegetation units within small areas, were better picked out due to the better spatial resolu 
tion, compared to the TM-sensor on the Landsat 5 satellite. Visual interpretation on the autumn-scene of Land 
sat 5 TM-sensor gave a successful detection of swamp forests at the riverbanks by using the channel combina 
tions of CH 456 and CH 543. 
1 INTRODUCTION 2.3 Data acquisition 
In Norway there are very small areas that have been 
vegetation mapped, and there are in certain areas a 
very strong need for vegetation-maps as a basis for 
land use planning. Methods for mapping natural re 
sources has been strongly develloped the last decades, 
and one of the most used methods is Remote Sensing. 
In comparison with personalintensive and conventio 
nal methods, Remote Sensing from satellite give a good 
ability for mapping and monitoring the natural re 
sources, as a result of the great ceverage in area and 
the ability for repetetive passes. 
The aim of the present investigation is a comparison 
of SPOT-simulated and Landsat 5 TM imagery for vegeta 
tion mapping purposes. 
2.0 The study area 
In this present study I have selected one of the test- 
areas which was included in the SPOT-simulation prog- 
ram in Norway 1982. This area called Habafjell-Skrub- 
ben (Cavarre in Sami language) is situated in the lar 
gest valleysystem in Central Troms, Northern Norway, 
and comprises parts of Rostadalen and Dividalen val 
leys (figure 1). This area has a fairly wide variation 
in ecologicalniches and most of the common vegetation 
types in the central part of Troms are represented in 
the study area. Forests covers about 60 % and alpine 
vegetation 30 % of the land area (Elven & Vorren 1980) 
2.1 Ground truth program 
The ground truth program did include: 
1. Traditional vegetation mapping of the whole study 
area based both on IR-aerial- and BW-aerial photo 
graphs . 
2. A special training-site program. 
3. Phytososiological investigation of the several 
vegetation types within the study area. 
2.2 Fieldwork 
The fieldwork was carried out during June to September 
1983 with supplementary visits to the area in 1984. 
Two Landsat 5 TM scenes were ordered, the first from 
3. june 1984 and the second from 2. October 1984. 
Landsat 5 TM PASS 197/11 1984-06-03 Time: 09.50.00AM 
Sunelevation: 42 degrees. 
Landsat 5 TM PASS 197/11 1984-10-02 Time: 09.46.00AM 
Sunelevation: 17 degrees 
Figure 1. The position of the study area, and the most 
important geographical names mentioned in the text. 
(Based on Cappelen's map 9-10, 1:400 000).