Original Article Landslides (2007) 4:113–122 DOI 10.1007/s10346-006-0069-5 Received: 29 May 2006 Accepted: 26 October 2006 Published online: 1 December 2006 © Springer-Verlag 2006

Hiroshi P. Sato . Hiroyuki Hasegawa . Satoshi Fujiwara . Mikio Tobita . Mamoru Koarai . Hiroshi Une . Junko Iwahashi

Interpretation of landslide distribution triggered by the 2005 Northern Pakistan earthquake using SPOT 5 imagery

Abstract The 2005 northern Pakistan earthquake (magnitude 7.6) of 8 October 2005 occurred in the northwestern part of the Himalayas. We interpreted landslides triggered by the earthquake using black-and-white 2.5-m-resolution System Pour l’Observation de la Terre 5 (SPOT 5) stereo images. As a result, the counts of 2,424 landslides were identified in the study area of 55 by 51 km. About 79% or 1,925 of the landslides were small (less than 0.5 ha in area), whereas 207 of the landslides (about 9%) were large (1 ha and more in area). Judging from our field survey, most of the small landslides are shallow rock falls and slides. However, the resolution and whitish image in the photos prevented interpreting the movement type and geomorphologic features of the landslide sites in detail. It is known that this earthquake took place along preexisting active reverse faults. The landslide distribution was mapped and superimposed on the crustal deformation detected by the environmental satellite/synthetic aperture radar (SAR) data, active faults map, geological map, and shuttle radar topography mission data. The landslide distribution showed the following characteristics: (1) Most of the landslides occurred on the hanging-wall side of the Balakot–Garhi fault; (2) greater than one third of the landslides occurred within 1 km from the active fault; (3) the greatest number of landslides (1,147 counts), landslide density (3.2 counts/km2), and landslide area ratio (2.3 ha/km2) was found within Miocene sandstone and siltstone, Precambrian schist and quartzite, and Eocene and Paleocene limestone and shale, respectively; (4) there was a slight trend that large landslides occurred on vertically convex slopes rather than on concave slopes; furthermore, large landslides occurred on steeper (30° and more) slopes than on gentler slopes; (5) many large landslides occurred on slopes facing S and SW directions, which is consistent with SAR-detected horizontal dominant direction of crustal deformation on the hanging wall. Keywords Earthquake . Pakistan . Landslide . SPOT . Interpretation . Envisat . SAR . Kashmir . Muzaffarabad Introduction The northern Pakistan earthquake occurred in the Pakistancontrolled Kashmir region in the northwestern part of the Himalayas on 8 October 2005. The magnitude was 7.6, and the epicenter (34.402°N, 73.560°E) was 105 km from Islamabad (USGS Earthquake Hazard Program 2005). The highest numbers of aftershocks (122) were recorded on 9 October 2005 with a significant drop of aftershocks in subsequent days. The total aftershocks were 1,778 at the end of 2005 (Pakistan Meteorological Department 2006). According to the Pakistan Government, there were 87,350 fatalities as of November 8, 2005. In the Azad Jammu and Kashmir states,

which sustained great damage, the dead, the injured, and the number of damaged houses were 43,421; 31,489; and 249,251, respectively (Azad Jammu and Kashmir Government 2006). Damaging earthquake-induced landslides have been documented from at least as early as 1789 BC in China and 373 or 372 BC in Greece (Keefer 1994). According to Chigira and Yagi (2006), earthquakes that triggered landslides in the past decade or so include the 1999 Chi-Chi earthquake in Taiwan (10,000; Wang et al. 2003), the 1995 Hyogoken-nanbu earthquake in Japan, the 1994 Northridge earthquake in the USA (11,000; Harp and Jibson 1996), the 1989 Loma Prieta earthquake in the USA (1,280; Keefer 2000), and the 2004 mid-Niigata prefecture earthquake in Japan (1,353; Sato et al. 2005a). The northern Pakistan earthquake also induced many landslides. Sato et al. (2005b) interpreted the distribution of landslides around Muzaffarabad for the earthquake by comparing two 1-mresolution Ikonos color images that were taken on 22 September 2002 before the earthquake and 9 October 2005 after the earthquake (images are published on the Web; Space Imaging 2005). This study showed 100 landslides in an area 11 by 10 km. However, it was difficult to identify landslides in each unit slope because Ikonos images were not stereoscopic. In the present study, landslides triggered by the earthquake were interpreted using the 2.5-m-resolution System Pour l’Observation de la Terre 5 (SPOT 5) stereo image. This is a black-and-white image, and the resolution of SPOT 5 image is coarser than that of the Ikonos image; however, the coverage area of SPOT 5 is wider than that of the Ikonos, and it is suitable to map landslides. In the monsoon season, heavy rainfall may cause secondary landslides, and many residents who live on the slopes may suffer additional damage. To document the disaster, it is necessary to know the wider distribution of landslides in detail. Furthermore, it is important to map landslides to analyze their geomorphologic and geological characteristics. Study area The study area shown in Fig. 1a is on the northwestern part of the Indo-Australian plate that is being driven northward beneath the Eurasian plate from a spreading center in the Indian Ocean. Plate convergence began in Cretaceous time (before about 150–65 Ma) and continues today (Lefort 1975). The study area is in the Himalayan orogenic belt within fold and thrust zones between two plates. Rugged mountains and deep valleys characterize most of the area. Local relief is extreme, from an elevation of 650 to 1,050-m valley along the Jhelum River to over 4,500 m in mountains between branches of the river.

Landslides 4 • (2007)

113

Original Article Fig. 1 Index map showing study area (red rectangle) in the context of tectonics of wide area (Drewes 1995, modified). a 1 Thrust fault movement dominant, 2 strike–slip fault movement dominant, 3 spreading center, 4 tectonic transport direction. b N Nanga Parbat syntaxis, J Jhelum River (Hazara) syntaxis, I Indus river (Kalabagh) syntaxis, P Potwar plateau, MFT main frontal fault, MBT main boundary thrust, and MMT main mantle thrust

Geologic setting In Fig 1b, the study area covers two geological areas: one is Hazara district, north of the main boundary thrust (MBT); another is south of the MBT. The geology in the study area ranges in age from Precambrian to Quaternary and includes sedimentary, igneous, and metamorphic rocks, and unconsolidated material. In the Hazara area, rock units of Precambrian and Paleozoic ages are mainly clastic. A long period of mainly carbonate deposition extending from Carboniferous (before about 360 Ma) to the Eocene (before about 45 Ma) is recorded in a nearly complete sequence of rock units totaling about 1,700 m in thickness. Clastic

114

Landslides 4 • (2007)

deposits became dominant in the middle and late Tertiary (before about 23–5.2 Ma), reflecting the collision of the two plates (beginning of the Himalayan orogeny). In the south area of the MBT, Miocene (before about 15 Ma) deposits exist. They mainly consist of sandstone transported from the rising mountains in the north and deposited in a marginal sea environment (Calkins et al. 1975; Drewes 1995). Tectonic setting In Fig. 1b, the dominant structural feature is not straight lined but is the nearly 180° bend of thrust fault. This structure is called Nanga

Parbat, Jhelum, and Indus syntaxes in this paper. The origin of Nanga Parbat syntaxis is thought to be formed in Miocene. Before the end of the Miocene, the compressive stress field spread farther south. Compressive deformation of the rocks north of the MBT fault were intensified, and deeper-level features and rocks were eventually brought to the surface. As the stress field propagated still farther south, fault propagation advanced more rapidly than it did elsewhere, forming the Salt Range and the lobate pattern of the Potwar Plateau. The MBT plate did not cease its southward movement before 1.8 Ma. The Indus River syntaxis was growing, and the nearby thrust faulting took place as recently as 0.6 Ma (Drewes 1995). Crustal deformation of the earthquake The crustal deformation caused by the earthquake was spatially mapped with the synthetic aperture radar (SAR) data from the European Space Agency’s environmental satellite (Envisat). Fujiwara et al. (2006) used the SAR data collected from the earthquake-damaged area from descending acquisitions (17 September 2005 and 22 October 2005)that result in an east–southeast line-of-sight (LOS) direction from the ground target to the satellite. The measured crustal deformation was the change in length along the radar LOS. In calculating crustal deformation, Fujiwara et al. (2006) used the interferometric SAR in the area where the deformation gradient is not large. However, in the area where the deformation gradient is large such as the area near faults, they used a subpixel-level offset estimation technique (Tobita et al. 2001), which is an enhanced SAR-image matching method using the fast Fourier

transformation. The result shows that the newly deformed area occupies a 90-km-long northwest–southeast trending strip. Furthermore, Tobita et al. (2006) calculated the components of displacement in each direction (north–south, east–west, and up– down), determined by the combination of the three pairs of SAR data from both descending and ascending (i.e., the Envisat moves toward north and south, respectively) acquisitions (19 September 2005 and 24 October 2005, 3 September 2005 and 12 November 2005). The result is an image (Fig. 2) based on a traditional Cartesian map. The measurement accuracy is ∼1 m in each direction. The figure shows the horizontal component of the crustal displacement by the red vectors and the vertical displacement by color shading. In Fig. 2, a 90-km-long northwest– southeast trending strip is shown as reported by Fujiwara et al. (2006), and the maximum vertical deformation (6 m) was observed north of Muzaffarabad (Tobita et al. 2006). Furthermore, south– southwest horizontal crustal displacement (∼1 to 2 m) is shown from Balakot to Muzaffarabad. Known active faults (Muzaffarabad and Tanda faults) stretch to the northwest and southeast from the epicenter (Nakata et al. 1991). They reported that Pleistocene fan surfaces located in southeast Muzaffarabad are vertically dislocated by the Tanda fault. Observations revealed some uplift (on the northern side of Tanda fault) and dextral (right-lateral) strike–slip displacement. Recently, Kumahara and Nakata (2006) gave a redefinition of the Tanda fault and the Muzaffarabad fault, and they comprehensively renamed them as the Balakot–Garhi fault.

Fig. 2 Detected crustal deformation using Envisat/SAR data with a wavelength of 5.6 cm, C-band (Tobita et al. 2006, modified). Color bar and vector indicate up–down and horizontal component of the SAR-detected crustal displacement, respectively. M Muzaffarabad, B Balakot. The star mark is the epicenter

Landslides 4 • (2007)

115

Original Article In Fig. 2, the Balakot–Garhi and Jhelum faults (pink and black lines, respectively) are mapped. The SAR-detected crustal deformation was along the Balakot–Garhi fault. The northeast side of the fault is the hanging wall, and all observations were consistent with previously known directions of past fault movements. Method Landslide interpretation using stereo images To obtain stereoscopic view from the satellite images, two images were used. One was taken on 20 October 2005, and the other was taken on 27 October 2005. Figure 3a shows the stereo image in north Muzaffarabad where ∼6 m of earthquake uplift was seen by SAR data. The landslides are the bright white slopes on the east side of the river in Fig. 3a. In this figure, the ellipse, #5, shows not landslides but the flexure cliff along the surface rupture of the earthquake fault. In the landslide sites, it was difficult to interpret geomorphic features in detail because of SPOT 5’s resolution and the high-

contrast image. As a result, the outline of each landslide was delineated as separate neighbor landslides. As was stated previously, the purpose of this study is to map landslides triggered by the earthquake. However, some of the identified landslides may have occurred before the earthquake. Furthermore, interpretation may confuse land-use characteristics as landslides. When agricultural crop fields and logging sites (bright whitish images) were surrounded by the forest (blackish image), they tended to appear similar to shallow landslides. To remove such the interpretation errors, two images, taken on March 2, 2004 (before the earthquake) and October 20, 2005 (after the earthquake), were compared. Preparation of orthographic image To precisely identify the position of the landslides, SPOT 5 images were geometrically corrected using ground-control points (GCPs), and orthographic (ortho-) images were prepared. The 34 mountain summits’ location and elevation data were gathered from 90-mresolution shuttle radar topography mission (SRTM) grid data (USGS Earth Resources Observation and Science 2006) and were used as the GCPs. As a result of geometric correction calculations, root mean square errors at the 34 GCPs showed about 72 m in the east–west direction and about 62.3 m in the north–south direction, respectively. This error indicates that the location of ortho-images randomly shift one-grid long at most. Finally, the landslide outlines were drawn on the ortho-images. Result and discussion

Fig. 3 a Black-and-white SPOT 5 stereo image. The area is shown as the small rectangle in Fig. 4. b Large landslide (shallow disrupted rock slide) at north Muzaffarabad

116

Landslides 4 • (2007)

Landslide distribution As shown in Fig. 4, the interpreted landslides were overlaid on SRTM data. The total number of the landslides was 2,424. The landslides were mapped in the overlap area of the pre- and postearthquake images, 55 by 51 km. However, it was difficult to recognize landslides in a high mountain area covered with snow. Figure 5 indicates the area of mapped landslides as pale blue and the noninterpreted area as dark blue. In Fig. 5a, 2,424 landslides are shown. About 79% or 1,925 were small (<0.5 ha in area) landslides; 207 (about 9%) were large (≥1 ha in area). In Fig. 5b, 207 large landslides are shown. According to our field survey on 20–21 January 2006, most of the small landslides were shallow rock falls and slides, and large landslides were shallow disrupted rockslides. Figure 3b, taken at north Muzaffarabad, shows an example of a large landslide where dry, highly disaggregated material cascaded downslope to flatter areas. The comparison of Fig. 4 with Fig. 2 shows the landslides were concentrated on the hanging wall (northeast side) of the Balakot–Garhi fault. Furthermore, many landslides occurred along the Balakot–Garhi fault. The relation between the landslides and the distance from this fault is shown in Fig. 6a. It was found that about 36% (875) and 47% (1,139) of the total landslides occurred within 1 and 2 km of the fault, respectively. Figure 6b shows the correlation between landslides and elevation based on SRTM data. Over 200 landslides occurred in each 100m-interval zone from 800 to 1,500 m, and the highest numbers of 295 landslides occurred in the 1,300 to 1,400-m zone. This simply indicates that many landslides occurred at the elevation

Fig. 4 Distribution of landslides triggered by the earthquake, shown as red dots. Elevation data are based on the SRTM. Small shaded rectangle at north Muzaffarabad indicates the area shown in Fig. 3a. The star mark epicenter, M Muzaffarabad, B Balakot. Orange line Balakot–Garhi fault, and blue line Jhelum fault. The interval of gray and black contour lines are 100 and 500 m, respectively. Furthermore, the colored counter lines are as follows: palegreen 1,000 m; green 2,000 m; orange 3,000 m; purple 4,000 m. Darkblue frame illustrates SPOT 5 imagery coverage area

where the faults exist, and it should not be attributed to a dependency on elevation. Relation between geology and landslides To investigate the relation between geology and landslides, the landslide distribution was overlaid on the geological map (Geological Survey of Pakistan 1993). A description of the geologic units in the legend is shown in Table 1. As shown in Fig. 7, the area of mapped landslides covers geology from Precambrian rocks to Quaternary sediments, and locations of both the Balakot–Garhi and Jhelum faults are not always correspondent to the location of

the MBT. According to Fig. 7, Precambrian and Triassic rocks thrust and converge over Neogene rocks, and MBT is formed in a cusp shape. Near Muzaffarabad, Cambrian and Paleogene rocks thrust over Neogene rocks. According to our field survey, many landslides occurred along the Balakot–Garhi fault. In Fig. 8a, the number of landslides within each geologic unit was tabulated. Almost half of the landslides occurred in Miocene sandstone and siltstone (Tmm). Tmm occupies 47.3% of the mapped area. Furthermore, the Balakot–Garhi fault passes through the Tmm. These are the

Landslides 4 • (2007)

117

Original Article Fig. 5 Area of mapped landslides triggered by the earthquake using SPOT 5 stereo image, M Muzaffarabad, B Balakot. a Total landslides, b the 207 selected large landslides. The large landslides delineated by a dotted ellipse in Fig. 5b occurred on gentler slopes and at a distance of about 18 km or further. See Fig. 9b

reasons why the remarkable highest numbers are recorded in Tmm. Figure 8b shows the landslide density, namely the number of landslides in each geologic unit divided by the area of each unit (pale blue area shown in Fig. 5). Precambrian metamorphic and

118

Landslides 4 • (2007)

sedimentary rock (pCt) near Balakot has the highest density of 3.2 landslides/km2. The second highest density of 2.9 landslides/km2 was recorded in Eocene and Paleocene limestone and grey fossiliferous shales (Tep) near north Muzaffarabad. One of the reasons why the pCt and Tep recorded contained high landslide

slope change at each grid cell. Higher positive values means the shape of the slope profile is more convex; lower negative values means it is more concave. Zero means it is linear. Figure 9a shows the frequency of landslide with the slope category. The minimum and maximum slope where landslides occurred was 8.4 and 52.3°, respectively. The highest slope category with the number of landslides that occurred was 30–35°; 87 landslides occurred in the slopes 35° and higher, and 76 landslides occurred at slopes of 30° or less. There is a slight tendency that the large landslides occurred at steeper slopes than on gentler slopes. Figure 9b shows the relation between the slope and the distance from the Balakot–Garhi fault for large landslides triggered by the earthquake. It infers that large landslides occurred not only at a wide spectrum of slopes near the active fault. Most of the landslides at 10–15 km occurred on steeper slopes (40–45°). However, at a distance of about 18 km or further, large landslides also occurred at gentler slopes in the shaded area in Fig. 9b. Landslides within the shaded area in Fig. 9b and also delineated by a dotted ellipse in Fig. 5b may have been caused by locally amplified strong shaking. Figure 9c shows landslides compared to vertical curvature. The highest number of landslides occurred at the curvature of +0.1 to +0.2. One hundred twenty-eight landslides occurred at the curvature greater than zero (convex slope), and 79 landslides occurred at a curvature of less than zero (concave slope). As shown in this figure, the tendency is that large landslides favored convex slope profiles than concave slopes. Fig. 6 Investigation of a distance of landslides from the active fault and b elevation on the 2,424 landslides

density is that the Balakot–Garhi fault passes through or near these geologic units. However, Triassic limestone and dolomite (TR) contained a relatively high density of 2.4 landslides/km2, although it is located far from Balakot–Garhi fault. Figure 8c shows the landslide-denuded area ratio, namely the area denuded by landslides in each geologic unit divided by the area of each geologic unit (pale blue area shown in Fig. 5). This figure indicates that Tep has the highest ratio (2.3 ha/km2). The second highest ratio is in the Cambrian limestone and so on (Cat) with 1.9 ha/km2 and, the TR, third highest at 1.3 ha/km2. Relation between slope and vertical curvature Two parameters of slope in the dip direction and vertical curvature were calculated at each grid using the nearest three by three grid cells (180 by 180 m in area SRTM data), and the large landslides were overlaid on them. The reason why large landslides were used is that the two slope parameters are not applicable to small-scale features. The vertical curvature was calculated by the method of Zevenbergen and Thorne (1987). This index shows the rate of

Relation between slope aspect and horizontal direction of crustal deformation The relation between the slope aspect of large landslides was investigated using SRTM data in the nearest three by three grid cells and the horizontal component of crustal displacement shown in Fig. 2. The result is shown in Fig. 10a where red dots illustrate the slope aspect of the large landslides. It was found that the 151 landslides (about 73%) occurred at the slope facing S and SW direction. In Fig. 10a, the blue dots show the horizontal direction of displacement, which is also concentrated in the S and SW directions. Figure 10b shows the frequency of all slope aspects in the area on the hanging wall within 4 km of the Balakot–Garhi fault, which illustrates that they are concentrated in the direction from SW to NW. The frequency of S slope aspects accounts for over half of the slopes. However, slope aspect is not concentrated in S and SW directions. Figure 10c indicates the slope aspect (red) and the horizontal direction of displacement (blue) on the 123 large landslides in the same area as Fig. 10b. By comparing these figures, it is concluded that the aspect of landslides is not controlled by the general aspect of the slope but the direction of the displacement of the earth surface, i.e., the direction of crustal deformation. Final remarks We mapped 2,424 landslides using SPOT 5 satellite images. Using SPOT 5 stereo imagery, we mapped landslides with

Landslides 4 • (2007)

119

Original Article Table 1 Legend and explanation of the geological map (Fig. 7)

Legend pCm pCs pCt Cat

Cg

Dm Mz

TR TRCp Tep Tmm Q

Explanation Precambrian rocks Low-grade metasediments, argillites and subgraywacke sandstone, and thick algal limestone units Precambrian metamorphic rocks Schistose to phyllitic quartzite, phyllite, carbonaceous slate, sericite schist, and marble Precambrian metamorphic and sedimentary rocks Medium to coarse-grained meta-quartzite, micaceous metaspammites, and subordinate garnet-mica schist Cambrian rocks Lower part: limestone, dolomite, red sandstone and shales. Upper part: mostly siltstone, sandstone, volcanic rocks and hematitic mudstone, contains Early Cambrian hyolithes Cambrian igneous rocks Mostly granite and leucocratic siliceous gneiss ranging from finely foliated to coarse augen gneiss; absolute ages for Mansehra granite of 516 plus–minus 16 Ma with mafic intrusions dated 284 to 262 Ma Devonian rocks White, massive, and hard quartzite with dolomite intercalations Mesozoic rocks Includes Cretaceous, Jurassic, and Triassic rocks: dominantly marine limestone, shales, and sandstone: fossiliferous, mainly ammonoides, foraminifera, and bivalves Triassic rocks Limestone, dolomite, and shale Triassic to Carboniferous rocks Lava flows of augite–andesite composition, and agglomerate slate with minor quartzite, limestone, and porphyries Eocene and Paleocene rocks Shallow marine forminiferal limestone and grey fossiliferous shales, divided into several formations Miocene rocks Red and purple, hard sandstone and siltstone, contains thin intraformational conglomerate and lenses (Murree formation) Holocene Surficial deposit Unconsolidated deposits of clay, silt, sand, gravel, and moraines

Fig. 7 Overlay of the landslide distribution on the geological map (Geological Survey of Pakistan 1993). M Muzaffarabad, B Balakot, 1 MBT, and 2 Panjal thrust. Black solid line indicates SPOT 5 imagery coverage area

120

Landslides 4 • (2007)

Fig. 8 Comparison of triggered landslides with geology. a Numbers of landslide, b landslide density, c denuded area

respect to landforms. We confirmed that most of the landslides occurred along the active faults, that they were concentrated on the northeast (uplift) side of the faults, and that the earthquake-induced large landslides were triggered by the main shock. This study showed the usefulness of satellite data. These analyses using satellite data can be used not only to estimate damaged areas caused by the earthquake but also to evaluate the later effects of rainfall. In the monsoon season, heavy rains may cause secondary landslide damage. The interpreted

Fig. 9 Investigation of slope and vertical curvature on the 207 large-scale landslides. a Frequency with respect to slope, b relation between the slope and the distance from the active fault, and c frequency of landslides with respect to vertical curvature. Large landslides within the shaded area in Fig. 9b occurred on gentler slopes and at a distance of about 18 km or further. These large landslides are delineated by a dotted ellipse in Fig. 5b

landslide data were provided to the Geological Survey of Pakistan through the Japan Landslide Society, and they will help such disaster prevention measures. However, to prevent and mitigate future damage, further monitoring may be important.

Landslides 4 • (2007)

121

Original Article 3 Fig. 10 a Slope aspect (red) and horizontal direction of the crustal deformation (blue) of the 207 large landslides, b slope aspect of the hanging wall within 4 km from the Balakot–Garhi fault, and c relation between the slope aspect (red) and the horizontal crustal deformation direction (blue) on the 123 large landslides on the hanging wall within 4 km from the fault

References Azad Jammu and Kashmir Government (2006) Statistics of damage done by the earthquake. http://www.ajk.gov.pk/site/index.php?option=com_content&task= view&id=2651&Itemid=153. Cited 2 May 2006 Calkins JA, Offield TW, Abdullah SKM, Ali ST (1975) Geology of the southern Himalaya in Hazara, Pakistan, and adjacent areas. USGS Professional paper 716-C, p 29 Chigira M, Yagi H (2006) Geological and geomorphological characteristics of landslides triggered by the 2004 mid Niigata prefecture earthquake in Japan. Eng Geol 82:202–221 Drewes H (1995) Tectonics of the Potwar Plateau region and the development of syntaxes, Punjab, Pakistan. USGS Bulletin 2126 (22 pp) Fujiwara S, Tobita M, Sato HP, Ozawa S, Une H, Koarai M, Nakai H, Fujiwara M, Yarai H, Nishimura T, Hayashi F (2006) Satellite data gives snapshot of the 2005 Pakistan earthquake. EOS Transact Am Geograph Union 87:73–77 Geological Survey of Pakistan (1993) Geological map of Pakistan 1:1,000,000 Harp EL, Jibson RW (1996) Landslides triggered by the 1994 Northridge, California earthquake. Bull Seismol Soc Am 86:319–332 Keefer DK (1994) The importance of earthquake-induced landslides to long-term slope erosion and slope-failure hazards in seismically active regions. Geomorphology 10:265–284 Keefer DV (2000) Statistical analysis of an earthquake-induced landslide distribution— the 1989 Loma Prieta, California event. Eng Geol 58:231–249 Lefort P (1975) Himalayas—the collided range. Present knowledge of the continental arc. Am J Sci 275-A:1–44 Nakata T, Tsutsumi H, Khan SH, Lawrence RD (1991) Active faults of Pakistan. Research Center for Regional Geography, Hiroshima University, Japan, p 141 Kumahara Y, Nakata T (2006) Active faults in the epicentral area of the 2005 Pakistan earthquake. Research Center for Regional Geography, Hiroshima University, Japan, p 54 Pakistan Meteorological Department (2006) Earthquake/aftershocks report. http://www. pakmet.com.pk/EQ-Report.html. Cited 2 May 2006 Sato HP, Sekiguchi T, Kojiroi R, Suzuki Y, Iida M (2005a) Overlaying landslides distribution on the earthquake source, geological and topographical data: the mid Niigata prefecture earthquake in 2004, Japan. Landslides 2:143–152 Sato HP, Une H, Koarai M (2005b) Ikonos imagery interpretation on the disaster of northern Pakistan earthquake. http://cais.gsi.go.jp/Research/geoinfo/geoinfo_e.htm. Cited 2 May 2006 (in Japanese) Space Imaging (2005). http://www.spaceimaging.com/gallery/asiaEQViewer.htm. Cited 2 May 2006 Tobita M, Murakami M, Nakagawa H, Yarai H, Fujiwara S, Rosen PA (2001) 3-D surface deformation of the 2000 Usu eruption measured by matching of SAR images. Geophys Res Lett 28:4291–4294 Tobita M, Nishimura T, Ozawa S, Fujiwara S (2006) Crustal deformation of 2005 northern Pakistan earthquake detected by SAR (2) SAR image matching and 3D deformation map. Abstracts of Japan Geoscience Union Meeting 2006. http://www.jpgu.org/ publication/cd-rom/pdf/D124/D124-017_e.pdf. Cited 16 October 2006 USGS Earthquake Hazard Program (2005) Magnitude 7.6—Pakistan-usdyae. http:// earthquake.usgs.gov/eqcenter/eqinthenews/2005/usdyae/. Cited 2 May 2006 USGS Earth Resources Observation and Science (2006) Shuttle radar topography mission (SRTM). http://www.edc.usgs.gov/products/elevation.html. Cited 2 May 2006 Wang WN, Wu HL, Nakamura H, Wu SC, Ouyang S, Yu MF (2003) Mass movements caused by recent tectonic activity: the 1999 Chi-Chi earthquake in central Taiwan. Isl Arc 12:325–334 Zevenbergen LW, Thorne CR (1987) Quantitative analysis of land surface topography. Earth Surf Processes Landf 12:47–56 H. P. Sato ()) . H. Hasegawa . M. Tobita . M. Koarai . H. Une . J. Iwahashi Geography and Crustal Dynamics Research Center, Geographical Survey Institute, Tsukuba, Ibaraki 305-0811, Japan Fax: +81-29-8642655 S. Fujiwara Planning Department, Geographical Survey Institute, Tsukuba, Ibaraki 305-0811, Japan

122

Landslides 4 • (2007)