970 
tion of the reflected energy by a sensor carried on 
a data collection platform. According to energy- 
matter relationships different objects reflect ener 
gy differently in frequency, direction and inten 
sity. This permits to compose images with contrasts 
for reflected energy coming from different direc 
tions. In order to obtain object information data 
processing is required. 
As a photogrammetrist I see mainly two data stream 
types: the one for topographic information by the 
photogrammetric restitution process; and the other 
for thematic information by the image interpretation 
process. All of this information is only useful to 
me, if I can bring it into a geographic reference 
base, that is to say, if I can describe it in form 
of a map. I realize, that these definitions not 
include e.g. laser sensing methods of atmospheric 
profiles, radar altimeter and scatterometer measure 
ments of the sea surface, nor do they pertain to 
seismic, geomagnetic nor gravimetric methods, which 
some atmospheric physicists or some geophysicists do 
consider as remote sensing. 
2 DEVELOPMENT 
The development of remote sensing, according to my 
suggested definition is as old as the first photo 
graph taken by Niepce in 1839. It was the balloon as 
a platform, introduced by Nadir in the 1850's which 
brought the first military application of remote 
sensing in the form of photographic interpretation 
of the battlefield of Solferino. 
Also in the US battlefields were soon photographed 
during the civil war. World War I began to use the 
airplane as a platform for the same purpose, and 
nothing has changed much in the purpose of military 
reconnaissance since then, except the tools and 
their analysis techniques. 
The false color infrared film is such a military 
development. Another was the development of a ther 
mal infrared scanner. A third was the construction 
of an airborne radar. These sensors were classified 
in the early 1960's and therefore through the Infra 
red Laboratory of the University of Michigan, which 
later became the Environmental Research Institute of 
Michigan ERIM and through the activities of Profes 
sor Colwell at Berkeley the term remote sensing was 
introduced to data analysis with these new sensors 
including photography as well. This was still the 
time ITC established photo-interpretation institutes 
in cooperation with for example India and Colombia. 
In the meantime these institutions, like the ITC 
heavily engage in remote sensing. This has come 
about through the space activities, started in 1957 
with Sputnik in the USSR and followed shortly by the 
United States of America. Already in 1960 the US 
launched its first weather satellite Tiros and is 
operating now under NOAA two polar and two geo 
stationary satellites, which observe the earth's 
atmosphere in the visible and the infrared spec 
trum. 
The European Space Agency ESA and Japan have 
joined in with their weather satellites Meteosat 
and GMS, while the USSR has its own meteorological 
program. Remote sensing has in the 1960's in the 
USA followed the path of lunar exploration, with 
Lunar Orbiter utilized for mapping landing sites on 
an analog basis. The Lunar Landing has been the 
primary goal of exploration. 
However, the planets required for their explo 
ration from the mid-60's on imaging by digital 
systems. Landsat, launched in 1972 immediately 
provoked worldwide interest in the data, as ex 
pressed by the many receiving stations operating 
around the globe. During the last fourteen years 
remote sensing has gone with the Landsat program 
around the world as a new technology and as a new 
hope. 
Let us be reminded, that the status of world 
mapping, as expressed by UN surveys in 1980 is by 
no means adequate. Nor will it be possible to sig 
nificantly reduce the lack of maps by existing 
techniques. 
In contrast to that Landsat MSS, during the first 
eight years of its operation was able to provide a 
morefold coverage of almost every part of the 
globe. It has taken about hundred years to map the 
territory of the Federal Republic of Germany by 
classical ground survey tools at medium scales. We 
can do that today by photogrammetry in about ten 
years. 
There is hope that satellite imaging might permit 
us to do that in about one year, but it must do so 
with sufficient quality. 
The Landsat Multispectral Scanner has since 1972 
provided color composites for analog photographic 
interpretation. 
It is quite obvious that only digital processing, 
that is grey level stretch for each channel and 
contrast enhancement through local filter oper 
ations permit to look at a superior product. 
Despite of this digital processing Landsat MSS 
imagery with its 80 m pixels reaches a resolution 
limit. 
Already in 1973 it became clear, that high resol 
ution systems, such as the Skylab 190A photographic 
camera yields higher resolutions with about 20 m 
pixel equivalents. We therefore have a multitude of 
sensor systems available in remote sensing and none 
has fully replaced the other. 
Since satellite sensors are usually designed by 
industry or by governments not according to the 
requirements of practice, but according to hardware 
and funding interests and limitations, it is better 
for the user to compare the requirements by simu 
lations. Let us take the example of topographic 
mapping at 1:50.000, a key task for the economy of 
the world. A comparison showing Landsat data and 
the corresponding map 1:50.000 at the same scale 
proves that Landsat MSS in inadequate to map cul 
tural features at that scale. 
We know that 1:50.000 mapping can be done from 
1:50.000 stereo aerial photographs. If one digi 
tizes this imagery at varying intervals using a 
drum scanner, as shown here for 40 m pixels, then 
one can plot the topographic information in a 
stereo-plotter. The result of such a test in which 
pixel sizes in mono and stereo were used at 2 m, 
5 m, 10 m, 20 m, 40 m intervals, gave the indica 
tion that a 5 m pixel size is required for stereo 
restitution at the scale of 1:50.000. 
Among the existing imaging satellites of high 
resolution we must distinguish between those of 
pure remote sensing interest suitable for thematic 
cartography only, and between cartographic satel 
lites, which provide full 3D mapping capability by 
stereo imaging. 
3 STATE OF THE ART 
Landsat MSS is usually displayed as false color 
composite in 80 m pixels. The four channels permit a 
supervised multispectral classification to determine 
landuse. Unfortunately the classification precision 
for certain classes, like settlements is still too 
low. A better result is obtained when Landsat clas 
sifications are superimposed with map data. 
This is primarily so when Landsat TM data with 30 
m pixels or SPOT Data with 20 or 10 m pixels are 
combined; or if the experimental systems of space 
photography from Spacelab 1 with 8 m pixel equi 
valents and with the Space Shuttle LFC with better 
than 5 m pixel equivalents are used. 
4 FUTURE TRENDS 
Looking at the 
satellite syste 
tempted to bel: 
technology and 
sensing is in a 
In the US, 1 
National Resear 
has last year 
Sensing of th 
Crisis". This 3 
disaster and fc 
and points out 
remote sensing 
great number of 
The report ch 
ly stating, th 
world leader ir 
it has proved 
economic, socia 
the difficultie 
strative, and I 
leadership in t 
It already b« 
originally resp 
ed itself to r< 
But when NOAA' £ 
ded into gover 
be maintained ] 
land remote sen 
daily by the 
Sensing Commerc 
ing of remote 
lites is to be 
EOSAT has rec 
program in 1985 
envisaged. This 
the 10 year pr 
quired by image 
French Governme 
funded SPOT sat 
ter chance to 
ution sensors a 
Even in the 
three months of 
a replacement, 
on the Space Sh 
in 1989 be the 
with a microwav 
NASA has floi 
Shuttle once t 
the equipment : 
Metric Camera 
Republic of Ger 
The strange < 
sensing of the 
into great disa 
There are, of 
high resolution 
from sensor sen 
nology, data t: 
requirements ca 
military systen 
civilian progrs 
areas obtainabl 
mapping cameras 
the area. 
Outside of the 
and China have 
Brazil will hav 
Another decisic 
world is the ea 
TDRSS instead 
Landsat TM rec 
TDRSS was also 
that even EOSA1 
for the recepti<