Tatsuo Sekiguchi* and Hiroshi Sato*

Transcription

1 by Tatsuo Sekiguchi* and Hiroshi Sato* ABSTRACT Landslides induced by heavy rainfall and earthquakes may result in disaster by destroying homes and buildings. High-fluidity landslides caused by liquefied ground are characterized by rapid movement and long run-out distance. Photogrammetory has been used until now for producing maps and investigation of landslides in areas where a disaster occurred by landslides. Recently, Airborne laser scanning technology is expected to become a promising technique for making topographical maps and investigating the microtopographic classification and disaster prevention on slopes around urban areas. In this study, airborne laser scanning was carried out in a hilly area with landslides, and the effectiveness of airborne laser scanning was tested in comparison with a laser contour map and an aerial photo-interpretation result. KEYWORDS: Airborne laser scanning, Laser contour map, Microtopographic classification map, Aerial photo-interpretation 1. INTRODUCTION Landslides induced by heavy rainfall and earthquakes may result in disasters by destroying homes and buildings in urbanized areas. High-fluidity landslides caused by liquefied ground, which are most dangerous and damaging due to debris flow, are characterized by rapid movement and long run-out distance (Wang and Sassa 2002). In the 1995 Hyogo-ken Nanbu Earthquake, a landslide occurred in the Nikawa area because of the earthquake in Nishinomiya City, Hyogo Prefecture, Japan. On 20 July 2003, a landslide and a debris flow occurred because of a heavy rainfall in Minamata City, Kumamoto Prefecture, Kyushu Island, Japan. Great damage occurred because of these disasters and many people were killed. It is important to identify the geometrical and geomorphological conditions of slopes in which landslides occurred in the past. Photogrammetry has frequently been used for measurement. In addition, it has become possible to use a method called airborne laser scanning for three-dimensional measurement in detail. Airborne laser scanning has begun to be used to identify landslide characteristics (Hasegawa and Okamatsu, 2001). It is expected that airborne laser scanning will accelerate not only high-precision mapping but also landform analysis (Ackermann 1999). Using the acquired data, high-precision topographic maps, inclination classification maps and shading maps can be produced more efficiently. Furthermore, microtopographic mapping, analysis for landslide simulation and unstable slope identification will be actively performed. The purpose of this study is to show an example of the mapping of hilly terrain and microtopography by airborne laser scanning data in the Tama Hills near Tokyo. 2. STUDY AREA The Tama Hills are located near the boundary between the western Kanto Mountains and the Kanto Plain (Fig. 1). The Tama Hills lie on the southwest side of the Tama River, and their elevation gradually increases from the southeast to northwest. More specifically, the elevation increases from 80 m a.s.l in the eastern part of Yokohama City to 200 m a.s.l at Hachioji City. Though many rivers dissect the hills, the profile is generally gentle. The northeastern part of the Tama Hills, which is closest to Tokyo, has rapidly urbanized and it is difficult to find the original landscape. * Geographical Survey Institute

2 3. AIFRRBONE LASER SCANNING An airborne laser scanning uses an active sensor that measures the distance from the sensor to the ground on which the laser beam is reflected (Fig. 2). Aircraft positions are calculated using a combination of Global Positioning System (GPS) data both on the aircraft and on the ground, aircraft acceleration and three-axial attitude (,,) data measured by an Inertial Measurement Unit (IMU). Furthermore, the direction data of the laser beams are measured by a sensor onboard the aircraft. These data are combined to calculate the three-dimensional position (X, Y, Z) on the ground. In this study, the measurement specifications were as follows: laser instrument, EnerQuest RAMS system; laser wavelength: 1064 nm ; pulse rate: Hz ; scanning frequency: 24 Hz ; flying altitude, normally, 2600m above the ground ; flight speed: km/h ; swath width: 881 m. The interval of the measurement points was 2.5m in the track

3 direction and 2.0 m in the cross direction. 4. FILTERING PROCESSING To show only the surface of the ground, i.e. a Digital Terrain Model (DTM), it is necessary to remove the ground features from the DSM by filtering (Fig. 3). on a Triangulated The manual processing was repeated a few times to produce a more realistic DTM on the superimposed image (Fig. 5).. Laser measurement points in the study area were approximated by quadratic polynomials, which gave a threshold value from 2 m to 9 m. Finally, laser measurement points over the threshold value were eliminated from the search area (Fig. 4). 4. CONTOUR MAP Figure 4 Image map of filtering processing. Solid line shows approximated ground surface (A). Dashed lines are threshold value(b). Furthermore, manual processing was performed Irregular Network (TIN) calculated by the DTM, and the contour map was superimposed on an orthogonal color aerial photo. When the contour lines were dense on buildings or trees on the superimposed image, laser measurement points which had dense contour lines were removed from the DTM. to obtain a smoother DTM. A contour map was drawn in advance, A contour map of the study area was drawn on the DTM. First, the TIN was calculated by the DTM. At this stage, the DTM consists of randomly scattered laser measurement points. Next, 1 m, 2 m and 5 m grid DTMs were obtained from the TINs. At this stage, the DTMs consisted of regularly arranged points, namely grid data. Then, contour maps of 1 m, 2 m and 5 m contour intervals were drawn on the respective grid DTMs. Finally, nine kinds of contour maps (three kinds of grid intervals by three kinds of contour intervals) were produced. In terms of smooth contour lines, microtopography expression and ease landform

4 interpretation, comparing these maps with, it was found that a 2 m interval contour map made from the 2 m grid DTM was best in this study area. This contour map of the Hachioji area, called a laser contour map in this study, is shown in Fig. 7. Fig. 8 is a photogrammetric contour map. These two maps show the same place at the same scale. We made an aerial photo-interpretation of the microtopography in this area. It was found that the contour lines of the laser contour map, such as ridges, valleys and landslide are very clear, and more realistic than the photogrammetric contour map. 4. LANDFORM IDENTIFICATION The landform of the study area can be classfied into various microtopographies using aerial photo-interpretation. These microtopographies showed good correspondences to the contour lines of the laser contour map from the airborne laser scanning. Microtopographic classification map and the legend are shown in Fig. 9 and Table 1. Ridges and valleys from north to south are shown in the study area. There are gentle slope surfaces in the ridges and the steep slope on the downward side for the boundary has formed a knee line. In these steep slopes, there are many landslides. In the valleys, alluvial deposit surfaces are formed caused by landslide and debris flow. There are two types of landslide on the slope. One type, of small-scale and steep slope, we called "disturbed landslide" in this paper. Another type is on a little larger scale and relatively gentle slope inclination compared to the former type, which we called a " less disturbed landslide". The number of the type of "disturbed landslide" was distinguished respectively at 3, 12 and 14 sites in 1956, 1961 and 1974, by aerial photo-interpretation. Also, the type of "less disturbed landslide" was 12 sites. It was found that the landslides occurred on 32 sites (60%) in the vicinity of the knee line, but they occurred on 21 sites (40%) away from the knee line. Thus it was concluded that more landslides occurred in the vicinity of the knee line. 4. THE EFFECTIVE OF AIRBONE LASER SCANNING DATA FOR LANDSLIDE STUDY

5

6 As shown in Figs. 7 and 9, laser contour maps indicate microtopographic characteristics clearly and precisely. Laser contour maps are also more useful than photogrammetric contour maps to place and delineate landslides and other microtopographic characteristics interpreted on aerial photos. However, laser contour maps do not always indicate actual landforms : a few errors were confirmed on valley bottoms or near ridges (cf. the footnote of Fig. 10 for details). It was difficult to filter out bamboo grasses, because the grasses are low in height (10-20 cm). However, trees were successfully filtered out, because conifer trees are tall in height (10-20m). landforms such as knee lines and landslides more precisely than photogrammetric contour maps. It was also found that the laser contour map at the contour interval of 2 m, produced by a 2 m grid DTM, was useful for interpreting microtopographic classification maps. With multi-temporal aerial photo-interpretations, microtopographic characteristics, and landslides were identified. The identified features were placed or delineated on a laser contour map to produce a microtopographic classification map. This classification map indicated that landslides have mainly occurred at knee lines. In the future, it will be necessary to study how to identify relatively high-risk places for landslides using laser data, microtopographic classification maps and field surveys. It will also be necessary to examine the relationship between ground cover ratios and the distribution of laser measurement points of DTM. 5. REFERENCES A cross section of the data sample removed by the filtering is shown in Fig. 14. As a result of investigating this place with the aerial photo-interpretation, it became clear that it was a part in which conifers such as cedar forests grow thickly. In this region, most of the vegetation consisted of deciduous trees. Therefore, accurate laser data could be acquired, since the laser beams easily reached the ground surface in winter. In the future, there will be the necessity of investigating the relationship between airborne laser scanning data and vegetation such as conifer trees and bamboo forests. 4. CONCLUSIONS Ackermann F (1999) Airborne laser scanning present status and future expectations. ISPRS Journal of Photogrammetry Airborne laser scanning was carried out in the Tama Hills to produce laser contour maps. It was found that laser contour maps indicate