Application of Rs and GIS for Forest-Steppe and Dry-Steppe Geosystem Study
M.Ganzorig, H.Tulgaa, D.Amarsaikhan, B.Enkhtuvshin
Informatics and RS Centre of Academy of sciences
av.Enkhtaivan-54B, ulaanbaatar-51, Mongolia
Abstract
Different airspace experiments for various scientific and economic purposes have , been carried out in Mongolia using remotely sensed data taken from different ; platforms and materials of the ground measurements. The investigation of the physical condition of a geosystem in past and present time is one of the urgent tasks of many researchers dealing with earth sciences. Our aim in this study is to .investigate two different geosystems in Mongolia. To reach the goal the diverse data were compiled within the ERDAS and ILWIS-systems and various RS and GIS techniques were applied .
1 Introduction
The result of the various experiments, carried out on Mongolian territory, with the aim of exploring the environment and mineral resources by methods of remote sensing indicated indeed great opportunities for the application of remotely sensed date to solve the problems, associated with the environmental protection and earth sciences within such comprehensive area (1.565 mil. sq. km) having various natural surroundings and unique geosystems (Ganzorig et al. 1994). Among the methodical problems of RS the main attention has been paid to the thematic cartography and the study of the physical conditions of a geosystem. If the first is the traditional, then the latter is the new one related with the space monitoring of environment. In this case, the aim of RS. is to measure different characteristics of the geosystem whose investigation should be based on a systematical approach requires an implementation of a GIS which is a management tool of geographical and other spatial related information for some decision making process. Within the scope of our research, forest-steps and dry-steppe geosystems have been analyzed. As a case study the following areas have been selected :
Tsagaantolgoi, where the basic attention was paid to the investigation of the forest-steps geosystem and estimation of its condition;
Tumentsogt, where the aim was the study of the dry-steppe geosystem, its pasture degradation, soil erosion and soil saltiness;
To develop the GIS we have designed two types of databases (spatial and attribute) in different levels and for their implementation and final analyses different RS as GIS techniques were used.
2. Databases
We have designed, databases in 3 different levels: satellite, air and ground. The satellite level is based on the middle scale maps (1 : 200,000), the air level is based on the larger scale airborne data( 2 : 25, 000) The ground level is based on the data collected through field investigation. Generally the proper physical implementation of the conceptual model is very simple within the ERDAS domain because we need not to define topology to find spatial relationships or posted identifiers to join various attribute tables. Initially, in order to get the right georeference digital images in different levels were geometrically corrected to Gauss-Kruger map projection using topographic maps of the test areas corrected to gauss-Kruger map projection using topographic maps of the test areas (Ganzorig et al. 1994). All maps, except the topographic maps were converted to digital form by video scanning using a CCD camera. Then they were resampled into digital form by video scanning using a CCD camera. Then they were resampled into the geometrically corrected digital images. After that was one screen digitizing of each map using the DIGSCRN-module of the ERDAS-system. Within the GIS the entities should be uniquely identified and have a set of attributes. In our case the ordering numbers (#) of the classes of object served as the identifiers or keys. Some attribute tables were designed in Dbase-IV saturation enhancement and 3D view ere implemented within the ILWIS environment.
Source materials
Different maps for the test areas are described as follows and RS data are shown in Table 1
Tsagaantolgoi area
-
Topographic maps , scales, 1:100,00 : 50,00; 1:10,00
- Soil map, scale 1 : 200,000
- Geological map, scale 1 : 200,000
- Natural landscape (geocomplex)map, scale 1 : 200,000
- Pasture vegetation map, scales 1 : 200,00
- different ground truth data and field materials.
Tumentsogt area
- ground spectrometrical data
- Topographic maps, scales 1 : 50,00; 1:25,000
- Soil map, scale 1 : 200,000
- Natural landscape ma, scale 1:200,000
Table 1. Available RS data
| RS data |
Spectral Resolution |
Spatial Resolution |
Swath width |
Date |
| KFA-100 |
panch. |
5m |
75 km |
1985 |
| MSU-SX |
panch. |
240m |
600km |
1990 |
| MSU-E |
3 bands (vis., NIR) |
40m |
40km |
1990 |
| AFA |
panch |
1:33,000 |
1962, 1963, 1975 |
| MSK-4 |
4 bands (vis. NIR) |
1 : 25,000 |
(07-08).1990 |
| SPOT-XS |
3 bands (vis., NIR) |
20m |
60km |
1986 |
3. Test Site 'tsagaantolgoi'
Tsagaantolgoi area is included into the forest-steppe zone and its landscape is divided into 5 zones, like mountainous taiga or forest (1500m-2000m above sea level (asl), mountainous forest steppe (1300m-1500m asl), mountainous steppe (1200m-1400m asl), mountainous steppe (1000m-1200m asl), dry steppe and meadow steppe in the river valley and between the mountains (Ganzorig et al. 1995). The objectivity of the separation of the natural boundaries between the natural landscape subdivisions is defined by the availability of the information extracted from various RS data. In this area a large scale photograph taken by the 'KFA-1000' photo camera from the series of the Russian satellite cosmos' with a high spatial resolution was used as a basis for the
differentiation between the objective contours and defining the quantitative relationships
between the objective contours and defining the quantiative relationshios between the natural landscapes.
After applying some corrections to make more clear the
boundaries between the objects different gradient and high pass filterring
techniques were applied. After that, on the basis of knoweldge about the area and
other teematic information, in the enhancdd image the geocomplex features were
digitized using DIGSCRN-module and was created a natural landscape map
containing the relationship between the natural components (eg, relief, soil, vegetation, etc.) of the forest-steppe zone in Mongolia (Fig.1).
Figure 1 Natural landscape (geocomplex) map. 1. Larch-tree taiga, 2. Cedar-larch taiga, 3. Birch-larch
subtaiga, 4. forest with mezophit grasses 5-7. Forest-steppe with different stow modification, 11-21. Dry steppe with
different stow modification, 22-25 Dry steppe and bottomland meadow with different stow modification.
From the interpreted natural landscape map it is clearly seen that a set of stow
combination in the western and north-western parts of the test area is
significantly distingueshed from that in the eastern and southern parts.
Therefore, in order to find out the reason for these differences and check how
much it is real in large territory, we have used MSU-SK digital data and for its
quality improvement destriping and low and high pass filterings have been applied (Fig.2).
Figure 2 Natural rregionalization. I. Selenge forest-steppe
middle mountain and basin subprovince with regions: I-1
Buteelin forest and forest steppe, I-2 Selenge rivers dry
steppe, I-3 Buren taigaforest, I-4 Burgaltain dry steppe.
II. Orkhon-Tuula dry steppe: II-5 Enkhtal forest steppe, II-6
Tuula dry steppe and meadow, II-7 Orkhon-Kharaa dry
steppe.
The investigation of the most general tonal and textural differeces of MSU-SK
image indicated that there are along the north-western parts of the territory 6
types and subtypes of the natural landscapes, consisting of dozens and hundreds
of stows, whereas in the southern and eastern parts were only 2-3 types of
geocomplexes from the not rich combination of stows and used for husbandry mainly
in agriculture. MSU-SK data helped us to make clear the following basic natural
laws:
a) to determine the above mentioned differences related with the bioclimatic and
anthropogenic influences;
b) to identifly in detail the morphostructure in the regional level and geologicotectonic
formation clearly appeared in the relief of the territory with large area.
On the basis of these natural laws geosystems of middle taxonomical level
(natural subprovinces and rregions) have been detected. The further analyses
require the application of multispectral data. To get the best colour image
different spectral enhancement techniques have been applied to the MSU-E data and
the best rresult eas obtained by the use of the saturation enhancement. To apply
this technique, after sum normalozation, 3D RGB data was mapped to the 2D colour
triangle, removing the influence of the intensity. Within the triangle, at first
the data cloud was shifted to the origin, to make a colour balance and then
spread throughout the feature space to use all possible colours.
