Geophysical and Geodetical Investigation of A Landslide Area 1 ( Koyulhisar-Sivas , Turkey ) 2

The study area is in the last section in the close south of Koyulhisar (Sivas) landslide site and the 7 study area is in the most active location where the landslide’s displacement amount is the highest. Th e landslide 8 site was examined with geophysical ( SRT-seismic refraction tomography, GPR-ground penet rating radar) and 9 geodesic (GNSS-global navigation satellite system) methods. According to the geophysical results, wit hin ~20 m 10 of investigation depth, three layers with the average seismic P-wave velocities (V P) of 0.30, 1.00 and 2.00 km/s 11 have been identified. It was determined that the thickness of the first two layers of thes e layers from top to the 12 bottom is approximately 3 and 6.5 m, and the last l yer with Vp>2.0 km/s is the bedrock. Furthermore, in 13 geophysical sections, it was determined that the depth of the sliding surface which is the upper limit of the 14 bedrock varies between ~7-10 m. The geophysical results permitted to identify th e landslide type as planar 15 sliding, with the sliding direction in S-SE, and th e tilt of the layer being orientated in the same di r ct on as the 16 topography slope (mostly bigger than 5°). In addition, according to geophysics and geodetical results, it was 17 observed that the deformations in the landslide mas s have occurred from the geological unit, the layer or 18 topography slope, and precipitation. Therefore, it was thought that landslide activity may continue in the study 19 area. These results were showed that precipitation and de formations within the layer can be effective in 20 triggering the landslide in the future. Therefore, the study area contains the risk and the natural hazards, and 21 these threaten the settlement area and the buildings and other constructions there. 22


26
Koyulhisar is also an active landslide area and for the past 17 years, there has been observed an increase in 27 landslide activity (Tatar et al., 2007;Över, 2015). The large and small landslides in Koyulhisar landslide area 28 have mostly occurred due to natural causes until today. Artificial causes mainly constitute the landslides caused  by these studies. It has been determined that the displacement amounts of the landslide velocity vary between 1-49 8.6 cm/year by topography and geological bedding and that the landslide direction is usually S-SE oriented. In 50 terms of geology, some researchers have carried out geological studies on many issues such as geological, 51 tectonic, geotechnical, geochemical and geomorphological studies at the local and regional scale in which the 52 features of the faults, water, hot water, soil and rock on the North Anatolian fault zone (NAFZ) and in the region 53 were investigated. These studies are in geology, tectonics (Toprak, 1989;Uysal, 1995

59
The geophysical studies were carried out in a limited area where the first geophysical studies took place.

60
In particular, seismic tomography (seismic refraction tomography (SRT), multi-channel seismic wave analysis 61 (MASW)) and ground-penetrating radar GPR applications are preferred methods in landslide studies. The

94
Koyulhisar is about 180 km away from Sivas city center. The study area is located in the west of Koyulhisar 95 town center and in the north of the NAFZ (Fig. 1). The geological investigation of Koyulhisar has been carried 96 out regionally or locally by various researchers (Terlemez and Yılmaz, 1980;Toprak, 1989;Uysal, 1995

98
Formation is the youngest unit in the region. It was stated that the youngest unit consisted of the talus (slope or 99 deposit) and fluvial conglomerates and was seen along the strike-slip faults (Toprak, 1989). Toprak (1989)  . When the drilling logs are examined, there is generally the second unit in east of study area 120 (Hastaoğlu et al., 2015). Furthermore, it was observed that the content of the second geological unit did not 121 change even if the depth of the drilling increased. Therefore, the second geological unit was taken into 122 consideration in the interpretation of geophysical sections.

123
As it is seen in Fig. 2, the study area is located in an active area in terms of seismicity (Fig. 2). The 124 seismological history, the magnitude (M) of which is greater than 2.5, of the examined area and its surrounding 125 were investigated for this article.

138
Large earthquakes affecting Koyulhisar district also occurred in the region. These largest earthquakes are 139 in the south of the NAFZ or Suşehri district and a total of three large earthquakes with M≥5.6 occurred there 140 (Över, 2015). Among these, the 1992 earthquake is closest to the study area with the least depth but the second 141 largest earthquake (Fig. 2). This earthquake is an earthquake with 6.1 magnitudes that occurred 10 km below the 142 ground. The large earthquakes in the south of Suşehri district which is just 13 km away from the study area 143 occurred in 1909 and 1939. 1909 earthquake occurred 60 km below the ground and is the largest and deepest 144 earthquake with a magnitude of 6.3. 1939 earthquake is also deep and the third largest earthquake that occurred 145 50 km below the ground with a magnitude of 5.6 (Över, 2015). In addition, when

154
The data regarding the rainfalls with the effects of triggering the landslides are presented in Table 1

157
According to the data obtained between 1950-2015 in Table 1, the rainy periods are generally between October-  results. On the other hand, it was understood that the precipitation increased by the decrease in temperatures. It is 192 also seen that the total annual amount of rainfall increased about 2-fold in 2014 compared to 2013 (Fig. 3a-b).

193
According to all results, rainfalls are considered to be effective in triggering of the landslide because the ground 194 of this landslide area, which is filled with loose units and old cracks, is supersaturated with water due to the The seismic refraction tomography (SRT) and ground-penetrating radar (GPR) methods are applied in 204 tomography format. The SRT method determining the seismic P-wave velocities (V P ) for seismic applications 205 and the GPR method for electromagnetic (EM) applications were used in the geophysical data collection in the 206 area (Fig. 4). The high-frequency electromagnetic waves can reach deeper in the environments with low 207 conductivity like sand. However, the conductive units such as clay and shale decrease the penetration depth of 208 the signal transmitted and lead to absorption (Annan et al., 1988;Davis and Annan, 1989). Firstly, SRT and GPR 209 data were collected along multiple transects in two different areas of the study area named A and C (see Fig. 4).

210
Then, the geophysical profiles were processed to the satellite map according to the coordinates along with the 211 topographical elevation curves and GNSS measurement locations for the ease of interpretation (Fig. 4a).

219
The profile shooting technique in the seismic study, hammer and iron plate of 8 kg weight as the source 220 P geophone of 14 Hz (the total number of geophones is 12) and Geometrics branded seismic device as the 221 receiver was used while collecting the SRT data. In all profiles, the geophone interval was 5 m, offset distance 222 was 2.5 m, the sampling interval was 256 ms and the record length was 512 ms. The geophones were 223 respectively fixed on the ground within the selected geophone range and their connections with the seismic 224 device were made. Then, seismic measurements were recorded by starting from the offset distance of 2.5 m, 225 reducing to sledgehammer plate and making at least 5 times shots between each geophone, respectively. In the 226 evaluation of the SRT data collected in the field, SeisImager program was used for displaying, processing and 227 evaluation of the seismic refraction waves. The marking of the first arrivals of the SRT data was performed using 228 Pickwin, and the evaluation of the first arrival data was performed using Plotrefa module. The GPR data were and saved in the "Correct for two layers" option. Thus, the models were converted from m to ns and the GPR 241 sections were prepared for interpretation. Thus, the collected geophysical data were converted into 2D (two-242 dimension) elevation-distance (SRT) and depth-distance (GPR) sections by assessed in the appropriate software.

243
The geophysical study area is one of the most active locations of the landslide area. As it is seen in Fig.   244 5, geomorphologically the landslide cracks on the surface, displacement traces, and structural damages in the 245 study area and its immediate surroundings can be monitored clearly in this activity area. Visibly, the damaging 246 effects of still active or old landslides on residences, roads and walls are also observed easily by field 247 observations. Therefore, none of the damaged constructions are used in the Koyulhisar.

251
SRT sections: 2D seismic cross-sections giving seismic V P -depth information are presented in Fig. 6 and   252 7. In the seismic data evaluation, the coincidence was provided with RMS (Root Mean Square) errors ranging 253 between 3.4-4.5% in 2D inversion operation. According to these sections, two or three layers were identified at 254 about 20 m depth ( Fig. 6 and 7). It was understood that the tilts of these layers were southeast oriented, and their 255 tilt was greater than 5 0 . According to the average seismic velocities (V P ) calculated, three layers with the layer 256 velocities of 650, 1200 and 2100 m/sec were defined from top to bottom. Thus, the seismic V P velocities were 257 observed that they increased towards the depth. It was determined that the depth of the sliding surface varied 258 about between 3-7 m ( Fig. 6 and 7). Therefore, these depths were defined as the layer with the risk of 259 dislocation. This area was considered to have a risk of dislocation due to these loose units, rainfall and tilt 260 conditions. The seismic velocity of the first layer is lower than V P1 <650 m/sec, but the seismic velocity of the 261 third layer may be greater than V P3 >2100 m/sec.

269
presenting a low-frequency high-amplitude sync-phase axis, which can be inferred as the sliding surface in Fig. 8 270 and 9. In other words, two layers were identified in GPR sections. The first layer is weak, loose, cracked, moved 271 and also have lost their tightness, and their seismic velocity is low. Therefore, in Fig. 8 and 9, it was thought that 272 deformations developed on the sliding surfaces due to the geology of the study area in A and C area. It was 273 identified the deformations, called sliding surfaces, landslide furrows, scarps, collapsed zones, and cracks. If the 274 areas of A and C are compared, the deformations are more in area C than in area A. Therefore, the risk of 275 landslides may be higher in area C. In Fig. 8, the EM wave velocity calculated for the reflection surface in GPR5 276 cross-section -representing the GPR profiles-was shown as an example. The picks were exported with the 277 attribute of two-way travel time and the velocity of propagation of the wave was calculated about 0.1 m/ns ( Fig.   278 8 8). This value is generally observed in dry or wet soil, dry or wet clay and sandy environments (Wilchek, 2000; 279 Cardomina, 2002). Therefore, it was thought that this velocity value was compatible with the geological units 280 and electromagnetic waves led to rapid absorption due to the silty sandy clay layer. Because the first geological 281 unit is medium-very stiff, low-high plasticity, silty sandy clay. The deformation structures as sliding surfaces, 282 landslide furrows, scarps, collapsed zones, and cracks were observed in the GPR cross-sections ( Fig. 8 and 9). In 283 other words, the geological unit, the layer or topography slope and precipitation cause deformations in the loose 284 upper unit. Therefore, these structures may develop or occur in the landslide mass, as shown in Fig. 8 and 9.

285
Additionally, the geological units were observed in DH wells in the geophysics study area (Fig. 4). These 286 are mostly silty sandy clay and they have different characteristics above and below about 10 m in DH well. The 287 topography of the study area decreases from 925 m to 840 m and the elevation difference is 85 m (Fig. 4). The 288 amount of slope in the topography increases from south to north (>5 0 -10 0 ) in the geophysical sections ( Fig. 6 and   289   7). It was determined that the landslide type in the area was planar sliding and observed that the direction of 290 sliding was SE. As this information was associated with topography and the field observations, it was observed 291 that the topography was inclined from the north to the south of the study area. The results of the various studies 292 and also the findings of this article have proved that Koyulhisar landslides are generally caused by the known 293 reasons that trigger the landslide. Therefore, it was seen that the geological bedding was compatible with the 294 topographical sloping and the groundwater was compatible with the direction of flow.

302
According to the geophysical cross-sections, it was identified that the depth of the sliding surface varied between 303 3-7 m due to the topographical differences. These depths were the depths with low seismic velocities (<650 304 m/sec) and defined as loose units which were also observed in geological drilling logs. It was determined that 305 sliding surfaces, landslide furrows, collapsed zones, scarps, cracks were observed in the landslide mass in the 306 GPR sections. It was observed that the layer tilt was generally more than 5 0 in all geophysical sections and 307 compatible with the geology and the flow direction of the groundwater. It was determined that the landslide type 308 in the area was planar sliding and the direction of sliding was SE.

309
The geophysical and other results were found to be compatible because it is known that the landslide 310 direction across Koyulhisar is in S-SW and SE. Consequently, the fact that the depth of the sliding surface over 311 the geologic unit is loose, the seismic velocity of the upper layer is low and the tilt is an excessive show that 312 there is a new risk of landslide in the area. The other factors that trigger the landslide were found to be associated 313 especially with the fact that the area is seismically active, receive heavy rain and has a poor vegetation cover. On 314 the other hand, it was thought that blasting and excavation performed by human intervention can trigger the 315 landslides due to the geologically loose unit. Hence, the landslide area can be a potential area which is open to 316 natural/artificial hazards. The identified risks and natural hazards also threaten the settlement area, the buildings 9 and other constructions (e.g. roads, walls, parks et al.) in Koyulhisar. Therefore, there is still a high landslide 318 hazard in the study area and its surroundings, and this hazard will also occur in the future.

325
We would like to thank Geological Engineer Mehmet Demirel for his contributions to the Fig. 2.