FROM SATELLITE IMAGE TO MAP - A NEW GUIDE MAP FOR FINNISH RURAL AREAS
Kari Tuukkanen
Helsinki University of Technology
Institute of Geodesy and Cartography
Otakaari 1, SF-02150 Espoo, FINLAND
Abstract
In the research project called "A Guide map for Finnish Rural Areas" we have developed an entirely new map
production process as well as the map output. The goal was to develop a map for different distribution purposes,
for emergency vehicle use as well as for general transportation. The source information was gathered as follows:
land use information was interpreted from Landsat satellite images, height zones were generated from a DEM,
roads and names were digitized manually on the screen, residential information was retrieved from the nationwide
building register. In the production process, both vector and raster techniques were used. The final map output was
designed to match the quality and type of source information. The Intergraph MGE-system (Map Publishing
application) on Unix-workstation was utilized in the project.
KEY WORDS: Satellite image, DEM, Map design, Cartographic theory
1. INTRODUCTION
In Finland, the new addressing system for rural areas obliges
communes to update their road addresses and address-maps.
lisalmi is a town of about 24 000 inhabitants in the central part
of Finland. A new address-map for rural areas came into being
in lisalmi in spring 1991. Already before that year, some ideas
had appeared about more versatile use of the guide map.
Emergency services, post, other distribution activities, tourism,
and general transportation need a proper map when navigating
in the countryside.
In lisalmi a digital map database with Intergraph's hardware
and software has been under construction for many years. This
project has been quite successful and it has given some good
practical experience to the staff.
The Finnish National Board of Survey produces thematic map
data from satellite images. The existence of this data and its
usefulness for digital map production contributed to starting the
project described in this paper. The Guide Map for Rural
Areas’ is a totally new map type. The aim of this six-month
project was to define the information contents of the map as
well as the layout and production process for the new map
type.
The following partners were involved in the project: Geodata
Ltd. as an expert in raster data manipulation, the town of
lisalmi, the National Board of Survey (NBS) as the producer of
digital data, and Helsinki University of Technology (HUT) as the
performer of the research. A pilot project producing the guide
map for lisalmi was included in this research project.
This article gives an overview of the data sources used in the
project, the production process of the map, and the design of
the map.
166
2. DATA CAPTURE
2.1 Land use information
NBS produces thematic maps interpreted digitally from Landsat
and Spot satellite images. Landsat 4 and 5 Thematic Mappers
are the most commonly used data sources. The resolution of
Landsat images is 30 meters. After rectification, resolution is
improved to 25 meters. There are seven channels available for
different wave lengths of light.
Satellite data is manipulated by a segmentation process.
Segmentation generates regions from pixels that have
homogeneous densities within a certain tolerance. Segmen-
tation is performed digitally using the Australian Disimp
software. Mode filtering method is used in internal examination
of the segments.
Different channels provide different information. One channel is
used to interpret waters. Reliability for water interpretation is
high, 95-98%. Swamps are difficult to interpret, so a swamp
mask taken from a Topographic Map 1:50 000 is used as
additional information. Field is the most difficult theme to
separate from images. Reliability for field areas is about 50%.
A field mask of the Topographic map 1:50 000 is therefore
used in interpretation. Percentage rates for reliability are taken
from tests made in NBS. There are altogether 55 different
classes in land use and forest interpretation. Classes can be
merged with each other to achieve the ideal result for
cartographic presentation. There are the following classes in
the original interpretation: water, field, open area, clear felling,
sapling stand, open swamp, pine swamp, spruce swamp, pine
stand, spruce copse, broadleaf forest and mixed stand. Forests
are divided into subclasses by their timber volume.
Interpreted data is delivered in digital raster files or as
photographic material in the required coordinate system
(National Board of Survey, 1991).
Data was input into the Intergraph's system as a raw raster file.
It was
modul
is dis
table.
22.D
NBS
contol
the siz
the r
preset
Heigh
repres
Zones
and ti
was |
model
junctic
junctic
name
road
inform
the B;
the de
somel
Digitiz
In our
first, t
so-cal
graph
roads.
Fourtl
screel
were |
map 1
24 R
There
includ
detail:
Interg
HUT.
eleme