Monitoring of Landslide Using Photogrammetry and Remote Sensing
Lena Halounova
Remote Sensing Laboratory, Faculty of Civil Engineering, CTU Prague, Thakurova 7, 166 29 Prague 6, Czech Republic, tel. +0042 2 2435 4952, E-mail: halounov@fsv.cvut.cz
Karel Pavelka
Laboratory of Photogrammetry, Faculty of Civil Engineering, CTU Prague, Thakurova 7, 166 29 Prague 6, Czech Republic, tel. +0042 2 2435 4951, E-mail: pavelka@fsv.cvut.cz
KEY WORDS: landslide, GIS, DEM, stereo pair photos, geological data, river system
ABSTRACT
Northern Bohemia brown coal mine area is an area of more than 1500 km2. This area has been affected by open mine activities which followed ground mine activities. The history of mining in this region is longer than a hundred years. The Czech Geological Archive (GEOFOND) owns data concerning more than ninety active or passive landslides. Their study and investigation was performed by traditional method of field study and photo interpretation. The final results were presented by means of maps and reports related to individual cases. This project took into account various types of data of a selected area, such as topographical, geological, hydrogeological, and pedological maps in 1 : 50 000, or 1 : 10 000 scale, detailed landslide maps and stereo pairs of aerial photos. All data were included into a geographical information system. It was found that the most decisive phenomenum of a landslide occurrence is the existence of tailings and their slope shapes. The accurate digital elevation model showed to be necessary. It was created from digitized height contour maps and, photogrammetrically, from stereo pair photos.
The landslide risk assessment has been in detail studied within last 30 years. Various attitudes to a solution of this task have brought a large range of possible solution methods. They can be divided into five big groups.
Table 1. Landslide risk assessment analysis
type of landslide risk assessment analysis |
main characteristics |
distribution analysis |
direct field mapping of existing landslide movement |
quality analysis |
direct or half direct methods applied to geomorphologic maps |
statistical analysis |
undirect methods where statistical methods make it possible to determine potential landslides from parameter maps |
deterministic analysis |
undirect methods where parameter maps are combined with slope stability evaluation |
frequent landslide analysis |
undirect methods applying hydrological models, earthquake and rainfall data for correlation with existing landslides for evaluating threshold values with a certain frequency |
The scale choice of a risk analysis assessment is given by a purpose of analysis. The International Association of Engineer-Geological Mapping (IEG) decided to divide the analysis scale into following groups:
regional scale (< 1 : 100 000)
medium scale ( 1 : 25 000 - 1 : 100 000)
large scale (1 : 5 000 - 1 : 10 000).
The regional scale is suitable for areas 500 - 2000 km2. The smallest terrain units are tens of hectares. The medium scale is applicable for parts of districts, or for big engineering structures. The large scale is suitable for areas smaller than tens of square kilometers.
Table 2 shows an overview of data important for a relevant landslide risk assessment study.
Table 2:
data type |
|
geomorphology |
terrain map units |
geomorphological units |
|
recent landslides |
|
passive landslides |
|
topography |
digital elevation model |
slope map |
|
slope direction maps |
|
slope changes |
|
concavity/convexity |
|
engineering geology |
lithology |
derived data (DEM e.g.) |
|
sample points |
|
faults, lineaments |
|
seismic events |
|
isolines of seismic intensity |
|
land use |
present infrastructure |
previous infrastructure |
|
present land use map |
|
previous land use map |
|
cadastral data |
|
hydrology |
water streams |
catchment areas |
|
meteorological data |
|
water level |
2. STUDY AREA
The landslide risk assessment using GIS tools was methodologically processed within one map sheet in 1 : 50 000 scale of the Czech cartographic system. This map sheet covers an area which has been heavily touched by various mining activities. The main activity is represented by brown coal mines. The above mentioned region is in the vicinity of Czech towns Ústí-upon-Elbe and Bílina. The northern boundary is formed by the Erzgebirge Mountains.
3. RELATED DATA
Topographic maps in 1 : 50 000 scale, and 1 : 10 000 scale were used for digital elevation model (DEM) creation. The geological map in 1:50 000 comprised 90 geological classes. The hydrogeological map includes information about 16 hydrogeological classes, the pedological map has a long list with pedological units. The combination of all three types of data shows a large variety of natural conditions. These three maps in 1 : 50 000 scale were overlayed by the existing landslide map which comprised both recent and passive landslides from the Czech Geological Archive. It was found that nearly 90 per cent of all landslides are situated in tailings or recclamation morphological units and very often in the vicinity of water basins or streams.
Classes of these three groups were regrouped according to newly attached weight values. The values varied from 1 (no landslide susceptibility) to 4 (high landslide susceptibility) for geological, pedological and hydrogeological classes. The areas with highest landslide susceptibility (value 11, 12) were chosen to be studied in detail from the point view of slopes and directions of slopes. The creation of a digital terrain model showed to be very important for a relevant landslide risk assessment analysis made in larger scale than 1 : 50 000. The digital terrain model would have excluded areas which, even with high landslide susceptibility (>8) studied from geological maps are stable due to slope conditions on one hand, and would have stressed areas which are still untouched by landslides, but with high landslide susceptibility (>7) and of high landslide risk due to sloping on the other hand.
The digital terrain model was created from 1 : 10 000 topographical maps and from a stereo pair of aerial photos.
4. DIGITAL TERRAIN MODEL
4.1 The reference DEM
The reference DEM with a very good spatial resolution (the identical scale as the landslide changes) must be at the disposal of users in a useful raster or vector form. For the Czech Republic, a scale 1: 5000 - 10 000 is considered as sufficient. A detailed DEM was created in the past from the digital mapping in urban areas, but it is necessary to create the DEM for other areas which are endangered by landslides. It is carried out by by means of digitalization of the existing topographic maps. Elevation contours from 1 : 10 000 are vectorized and stored to files of related types of data
A derivation of the DEM from the old photogrammetric stereo images is another possibility of the reference DEM creating. In this case, the DEM is made by using a method of automatic image correlation of floating image submatrices. It is used for the creating of an orthophoto. Finally, the DEM of a selected area is stored in a raster form for a more accurate GIS analysis.
4.2 The new DEM
The new DEM of a selected area can be created by classical geodetical methods (tachymeter, total station) or by means of photogrammetry. There are two possibilities:
The full automatic process for the creating of the DEM can be used, however, it is normally applied in the case of aerial photogrammetry, by calculating orthophotos (perpendicular parallel projection). In such a case, the DEM from 1 : 7 000 aerial stereo pair photos is made.
The other possibility is the spatial (3D) measurement and data capturing on the stereoscopic model created on the analog or analytic machines or on the digital photogrammetric workstation. It is better for special defined areas (the known landslides). The scale of stereo images must be set on maximum (from a low flight, for example, the used image scale 1: 7 000).
The using of satellite stereo images depends on the size of land changes (landslides). Stereo images from the SPOT (spatial resolution 10 m), IRS (5.8m) or the scanned Russian satellite photo images with different spatial resolution (2-15m) are used. Today’s satellite data do not have ideal spatial resolution for the detection of small landslides and such a procedure is very costly. In many cases, false - colour films (infrared-diapositive) for the photo flights and aerial photo imaging are used. This type of film provides multi-purpose uses, for example, photo interpretation, land use or land cover.
4.3 Spatial analysis of DEM
If both the reference and the new DEM are in a raster form, the analysis is very simple. It is based on the difference in georeferenced raster images. On this level, the professional remote sensing software packages must be used. The most frequent use is probably the analysis on a GIS basis, where special functions for the data analysis (such as overlay function) are applicable. The long time monitoring is, at present, made only by means of GIS technique.
4.3 The testing area
There are two great landslide hazard areas in the Czech Republic. A new landslide area has appeared after the flooding in Moravia in summer 1997. The traditional landslide area is the Bohemia Brown Coal Basin. There are also landslides caused by a previous mining activity. As a testing area, a part of the Bohemia Brown Coal Basin in the North-West of the Czech Republic was prepared. In this area, testing sites were established for the landslide detection. For the creating of the new DEM, photogrammetric stereo pairs are used. In this case, the testing site was used and photographed with an aproximate scale 1: 6870 from the flight height of about 1300-1400m using MRB 15/17 camera with focus of 152mm. The photogrammetric snaps (5902,5903) were taken on 16th of April 1982. The reference DEM was created by the digitalization of contour lines from the existing topographic maps (from the year 1980).
4.4 Data processing
Both left and right images were scanned on HP scanner with geometrical resolution of about 600 DPI (1 pixel =30cm). Two methods for the processing of stereo images were used :
- The full automatic process (automatic image correlation for the DEM and for orthophoto creating); the EASI/PACE AD software were used (PCI Canada, software for processing digital images).
- The stereo image processing on the photogrammetric workstation DVP Leica; in this case, a grid of a spatial object point was manually created and processed. Processing was made by SCOP software (TU Wien, SCOP is a program system to process and maintain digital data describing the terrain or other surfaces) for the vector form or by EASI/PACE for the raster form.
5. CONCLUSION
Both types of DEM were applied. Aerial photos are more operational, they can be repeated far more frequently for time changes detection. These photos have worse accuracy in forest areas where the Earth surface is hidden under a forest cover.
References: