**2.1.2 AR anomalies corresponding to the Ms8.0 Wenchuan EQ**

On May 12, 2008, a great EQ of Ms8.0 struck the Wenchuan county and its adjacent area in Sichuan province of China, and within one month following it, more than thirty Ms5.0~6.4

①Minjiang river fault, ②Anning river fault, ③Doujiang Weir–An county fault, ④Xianshui river fault, ⑤White Dragon River fault

Fig. 4. Distributions of station, epicenter and fault.

aftershocks occurred along the NE-strike fault belt beyond 300 km long, from the Wenchuan county to the Ningqiang county in Shaanxi province. There were six AR stations within 400 around the main epicenter, which were Chengdu station (CDU, 35km), Jiangyou station (JYO, 150 km), Ganzi station (GAZ, 331 km), Mianning station (MNI, 260) and Xichang station (XCM, 360 km) in Sichuan province, and WDH station (300 km) in Gansu province (Fig.4).

#### **2.1.2.1 Medium-term AR anomalies before the EQ**

Significant AR anomalies were recorded at four stations in the medium-term stage before the Ms 8.0 Wenchuan EQ as follows.

1. Locally concentrated anomalies

200 Earthquake Research and Analysis – Statistical Studies, Observations and Planning

observed at underground water station[2], Longjiadian station which was about 20 km from station CLH and nearby the Cangdong rupture zone. We can notice from figure 3 that immediately before and after the occurrence date of the EQ, the drop AR change at station CLH was well corresponding to the rise change in water level at station Longjiadian. The opposite changeable patterns between electric and water are quite significant, which indicated that the water in the underground medium nearby the focal region played an

It can be seen from figure 3 that nothing was recorded before the Ms6.9 Ninghe aftershock (Nov. 15, 1976) at the geo-electrical and water stations, a possible reason for which was explained by associating with the mainly active faults in/nearby the focal region and the

On May 12, 2008, a great EQ of Ms8.0 struck the Wenchuan county and its adjacent area in Sichuan province of China, and within one month following it, more than thirty Ms5.0~6.4

①Minjiang river fault, ②Anning river fault, ③Doujiang Weir–An county fault, ④Xianshui river fault,

important role in AR changes.

⑤White Dragon River fault

Fig. 4. Distributions of station, epicenter and fault.

focal mechanism of the aftershock by Du *et al*.[3-4].

**2.1.2 AR anomalies corresponding to the Ms8.0 Wenchuan EQ** 

The anomalies were recorded at CDU, JYO, GAZ and WDH during the medium-term stage before the EQ, which were in the range of 400km from the main epicenter (Fig.5, Fig.6).

Fig. 5. AR monthly mean changes observed at four stations that were located along the Songpan-Ganzi active block before the M8.0 Wenchuan EQ

Changes in Apparent Resistivity in the Late Preparation Stages of Strong Earthquakes 203

behavior was similar to previous research results. According to Du et al.[8], the mediumterm AR anomalies along a actively geological structure around an EQ focal area usually started to appear synchronously or quasi-synchronously. In fact, the medium-term anomalies in observation of groundwater chemical components and water level, geo-stress and geo-deformation along a same structure also have such behavior as seen in AR

The drop-type anomalies were recorded at stations CDU, GAZ and JYO. Only at station WDH, was a rise-type anomaly recorded, yet this was not isolated. Stations WDH and Tianshui (TSE), in Gansu province, were all located to the north of the main epicenter, and station TSE was to the north of station WDH, which was nearby the NWW-striking western Qinling rupture belt and was 452 km from the main epicenter. At station TSE, a rise anomaly beyond 1% appeared during two months period preceding the main shock, which was the most prominent AR change in the last ten years at the station. The phenomenon of mostly drop-type AR anomalies coincided with previous researches, and the spatial distribution of the medium-term anomalies at the four stations well tallied, in the range of -

For change of patterns, a drop-type or rise-type change, of the medium-term AR anomalies which were processed by using the normalized variation rate method (NVRM)[9,7], Du et al.[10,7] got the following statistic results: for Ms 7.0 EQs, about 100% of the anomalies in the range of 150 km from epicentral areas are negative (a drop-type pattern) and about 71% of the anomalies in the range of 400 km are still negative. The reasons on the change patterns of

At station CDU, which was the nearest station to the EQ epicenter, the anomaly amplitude reached up to -5.5%. At stations GAZ and WDH, which were farther away from the EQ epicenter than station CDU, the anomaly amplitudes reached up to -5.3%--4.9% and 2.9%, respectively. The mean of the anomaly amplitudes was larger than that before the1976 M7.8 Tangshan EQ, and the anomaly amplitudes of the three stations decreased with the increase of epicentral distance. The anomaly recorded at JYO was small in amplitude, only -1.1%, although it was nearby the main EQ epicenter area (according to an investigation after the EQ, the measuring instrument at the station was not in good operation at that time, with a

The relationship between AR anomaly amplitude and EQ magnitude has been studied by numerous scholars[5-7]. The anomaly amplitude before the Wenchuan great EQ was in

Two observation channels, along N58oE and N49oW directions, are employed at station CDU. The anomaly amplitude recorded through the N58oE channel was -5.5%, whereas no anomaly was recorded through the N49oW channel. According to previous works[2,11], this anisotropic AR changes roughly indicated that the underground media here had been under the action of the maximum compressive stress in the NW-SE direction during the period from about Aug. 2006 to the occurrence of the main shock. At station GAZ, two channels, along N30oE and N60oW directions, are employed. The anomaly amplitude recorded

observation[8].

3. Mostly drop-type AR anomalies

4. Large-amplitude anomalies

accordance with the previous researches.

5. Anisotropic AR changes

fixed error).

400 km, with that before EQs with magnitude Ms>7.0.

the anomalies was theoretically explained by papers[10, 7].

Fig. 6. AR daily mean changes observed at CDU station before the M8.0 Wenchuan EQ

Furthermore, the spatial distribution of the anomalies was tectonically relevant to the Songpan-Ganzi active block. The main shock and its aftershocks occurred along the Longmen mountain nappe structure of the block. Accordingly, the anomalies were recorded at the four stations along the bordering faults around the block (Fig.4), whereas such anomalies were not recorded at the two stations MNI and XCM, which were located along the Anning river fault, beyond the block. This situation is similar to that which was observed before the 1976 Ms7.8 Tangshan EQ when anomalies appeared mostly at the stations along NE- and NW-striking conjugate faults in the Beijing-Tianjin-Tangshan areas. According to previous statistical studies on numerous EQ cases[5-6] by other Chinese

scholars, the spatially distribution of medium-term AR anomalies is about 300~400 km before an EQ with magnitude of Ms 7.0. According to paper[7], for Ms>6.0 EQs most AR anomalies distribute in the range of 400 km from epicenters and there is commonly no obvious difference between the ranges for Ms>6.0 EQs and Ms 7.0 EQs. The results show that an epicentral distance of 400km can be used as a reference for identifying medium-term AR anomalies related to Ms>6.0 EQs. Du *et al*.[8] studied the relationship between the spatial distribution of medium-term anomalies of other precursor observations in China, such as AR, groundwater chemical components and water level, geo-stress and geo-deformation, and the mainly active fault around the epicentral region, as a result, it was believed that the anomalies usually appeared nearby the faults around EQ focal areas.

The spatial distribution of the medium-term AR anomalies before the Wenchuan great EQ, which were concentrated in the range of 400 km from the main epicenter and along the bordering faults around the Songpan-Ganzi active block, was in accordance with the foregoing research results.

#### 2. Synchronous medium-term AR anomalies

These AR anomalies started to appear from about Aug. 2006 at CDU, JYO, GAZ and WDH. In other wards, they appeared synchronously before the main shock at the four stations which were located along the bordering faults of the Songpan-Ganzi active block. The behavior was similar to previous research results. According to Du et al.[8], the mediumterm AR anomalies along a actively geological structure around an EQ focal area usually started to appear synchronously or quasi-synchronously. In fact, the medium-term anomalies in observation of groundwater chemical components and water level, geo-stress and geo-deformation along a same structure also have such behavior as seen in AR observation[8].

#### 3. Mostly drop-type AR anomalies

202 Earthquake Research and Analysis – Statistical Studies, Observations and Planning

Fig. 6. AR daily mean changes observed at CDU station before the M8.0 Wenchuan EQ

anomalies usually appeared nearby the faults around EQ focal areas.

foregoing research results.

2. Synchronous medium-term AR anomalies

Furthermore, the spatial distribution of the anomalies was tectonically relevant to the Songpan-Ganzi active block. The main shock and its aftershocks occurred along the Longmen mountain nappe structure of the block. Accordingly, the anomalies were recorded at the four stations along the bordering faults around the block (Fig.4), whereas such anomalies were not recorded at the two stations MNI and XCM, which were located along the Anning river fault, beyond the block. This situation is similar to that which was observed before the 1976 Ms7.8 Tangshan EQ when anomalies appeared mostly at the stations along NE- and NW-striking conjugate faults in the Beijing-Tianjin-Tangshan areas. According to previous statistical studies on numerous EQ cases[5-6] by other Chinese scholars, the spatially distribution of medium-term AR anomalies is about 300~400 km before an EQ with magnitude of Ms 7.0. According to paper[7], for Ms>6.0 EQs most AR anomalies distribute in the range of 400 km from epicenters and there is commonly no obvious difference between the ranges for Ms>6.0 EQs and Ms 7.0 EQs. The results show that an epicentral distance of 400km can be used as a reference for identifying medium-term AR anomalies related to Ms>6.0 EQs. Du *et al*.[8] studied the relationship between the spatial distribution of medium-term anomalies of other precursor observations in China, such as AR, groundwater chemical components and water level, geo-stress and geo-deformation, and the mainly active fault around the epicentral region, as a result, it was believed that the

The spatial distribution of the medium-term AR anomalies before the Wenchuan great EQ, which were concentrated in the range of 400 km from the main epicenter and along the bordering faults around the Songpan-Ganzi active block, was in accordance with the

These AR anomalies started to appear from about Aug. 2006 at CDU, JYO, GAZ and WDH. In other wards, they appeared synchronously before the main shock at the four stations which were located along the bordering faults of the Songpan-Ganzi active block. The The drop-type anomalies were recorded at stations CDU, GAZ and JYO. Only at station WDH, was a rise-type anomaly recorded, yet this was not isolated. Stations WDH and Tianshui (TSE), in Gansu province, were all located to the north of the main epicenter, and station TSE was to the north of station WDH, which was nearby the NWW-striking western Qinling rupture belt and was 452 km from the main epicenter. At station TSE, a rise anomaly beyond 1% appeared during two months period preceding the main shock, which was the most prominent AR change in the last ten years at the station. The phenomenon of mostly drop-type AR anomalies coincided with previous researches, and the spatial distribution of the medium-term anomalies at the four stations well tallied, in the range of - 400 km, with that before EQs with magnitude Ms>7.0.

For change of patterns, a drop-type or rise-type change, of the medium-term AR anomalies which were processed by using the normalized variation rate method (NVRM)[9,7], Du et al.[10,7] got the following statistic results: for Ms 7.0 EQs, about 100% of the anomalies in the range of 150 km from epicentral areas are negative (a drop-type pattern) and about 71% of the anomalies in the range of 400 km are still negative. The reasons on the change patterns of the anomalies was theoretically explained by papers[10, 7].

#### 4. Large-amplitude anomalies

At station CDU, which was the nearest station to the EQ epicenter, the anomaly amplitude reached up to -5.5%. At stations GAZ and WDH, which were farther away from the EQ epicenter than station CDU, the anomaly amplitudes reached up to -5.3%--4.9% and 2.9%, respectively. The mean of the anomaly amplitudes was larger than that before the1976 M7.8 Tangshan EQ, and the anomaly amplitudes of the three stations decreased with the increase of epicentral distance. The anomaly recorded at JYO was small in amplitude, only -1.1%, although it was nearby the main EQ epicenter area (according to an investigation after the EQ, the measuring instrument at the station was not in good operation at that time, with a fixed error).

The relationship between AR anomaly amplitude and EQ magnitude has been studied by numerous scholars[5-7]. The anomaly amplitude before the Wenchuan great EQ was in accordance with the previous researches.

#### 5. Anisotropic AR changes

Two observation channels, along N58oE and N49oW directions, are employed at station CDU. The anomaly amplitude recorded through the N58oE channel was -5.5%, whereas no anomaly was recorded through the N49oW channel. According to previous works[2,11], this anisotropic AR changes roughly indicated that the underground media here had been under the action of the maximum compressive stress in the NW-SE direction during the period from about Aug. 2006 to the occurrence of the main shock. At station GAZ, two channels, along N30oE and N60oW directions, are employed. The anomaly amplitude recorded

Changes in Apparent Resistivity in the Late Preparation Stages of Strong Earthquakes 205

And the P-axis azimuths of the three EQs were 63°, 101° and 95°, respectively, resulting in an right-lateral movement of the NNW-striking Huya fault (i.e., the focal fault) when the previous Ms7.2 EQ occurred, and then an left-lateral movements of the fault when the following Ms6.7 and Ms7.2 EQs occurred[16]. At that time, three stations were in observation within a range of ~190 km from the three epicenters: station Songpan (SPN, 45km, and it was abandoned shortly afterwards) in Sichuan Province and station Lixian (LXN, 190 km, and abandoned in the 1990s) and station WDH (110 km), in Gansu Province. There were obvious differences between AR changes recorded at station SPN and those recorded at station LXN (Fig. 8). Station SPN was located on the west side of the Minjiang River fault (MRF) with the NNE strike, whereas the epicenters were to the east of the fault. As a result, no imminent AR anomaly was recorded at station SPN when approaching the occurrence date of the first Ms7.2 EQ, though the station was only 45 km from the epicenter, whereas at LXN station, a drop anomaly with amplitude of -0.8% was recorded during 4 days before the EQ in despite of being 190 km away. Contrarily, for the subsequent Ms6.7 and Ms7.2 EQs, a drop anomaly with amplitude of -3.5% was clearly captured at station SPN during 3– 4 days before them. This anomaly amplitude was much greater than that recorded at station LXN before the first Ms7.2 EQ. No imminent anomaly appeared at station LXN for the latter EQs. At station WDH, which was located between station LXN and the three epicenters and was near to the north side of the White Dragon River fault (WDRF) with the EW strike, the AR changes recorded before the three events were similar to those seen at station LXN.

Fig. 7. Distribution of epicenters, active faults and AR stations

According to papers[4, 17], imminent anomalies recorded at underground water level and water chemistry stations before the moderate, strong EQs which occurred in the continent of China also demonstrated such spatial distribution as AR imminent anomalies, which was related to mainly active faults of epicentral areas and EQ focal mechanisms. From loading

through the N30oE channel was -5.3%, and that recorded through the N60oW channel was - 4.9%. This indicated that the media here had been under the maximum compressive stress in the NWW-SEE direction during the period. The maximum compressive stress directions that were revealed by the AR changes at the two stations basically corresponded to the Paxis azimuth of the main shock[12]. Station JYO was near the northern aftershock area of the great EQ, and two channels, along the N70oW and N10oE directions, are employed at the station. The comparison between anomaly amplitudes recorded through the two channels was still credible as the measuring instrument at the station had just a fixed error. The anomaly amplitude recorded through the N70oW channel was -1.1% and no anomaly was recorded through the N10oE channel. This indicated that the media here had been under the maximum compressive stress close to the NS direction during the period, which well corresponded to the P-axis azimuths of most strong aftershocks in the northern aftershock area.

In summary, the behaviors of these medium-term AR anomalies, such as their spatial distribution within 400 km, their tectonic relevance to the Songpan-Gazi active block, and their amplitude attenuation with increasing distance, as well as their synchronous, mostly descending, large-amplitude and anisotropic changes, strongly support that these anomalies are indeed related to the focal processes of the main shock and strong aftershocks. Such locally concentrative AR anomalies as recorded at the four stations have not been succeeded in identifying in areas beyond 400 km from the EQ epicenter.

#### **2.1.2.2 Short-term AR anomalies before the EQ**

During the short-term period before the occurrence date of the great EQ, no obvious anomaly, like these recorded at station CLH (72km) in Hebei province before the 1976 M7.8 Tangshan EQ and at station WDH (110 km) before the 1976 M7.2 Songpan-Pingwu EQs (Fig.2), was recorded at the six stations within 400 km from the EQ epicenter. Upward AR changes commenced at station CDU from March 2008 and at station JYO from April 2008 (Fig.5a-b and Fig.6). The patterns consisting of an initial medium-term drop change followed by a short-term rise change before the EQ are consistent with these of the electrical resistivity change within an EQ focal area that are forecasted by DD model[13-14]. In fact, such patterns appeared often nearby epicentral areas of previous strong EQs[9, 2,7]. However, the rise change, upon which the main shock occurred, did not satisfy the anomaly criterion in amplitude, so no sufficient precursory information could be confidently detected in the period when approaching the EQ.

Du *et al*[3, 15] studied the spatial distribution characteristics of imminent AR anomalies before the moderate, strong EQs that occurred in the continent of China. As a result, it was got that the spatial distribution of the anomalies was influenced by mainly active faults around EQ focal areas and focal mechanisms. The influences include: (1) the anomalies appear mostly along or nearby the faults; (2) most anomalies are distributed in the two areas that are symmetrical about an epicenter and that azimuthally tie to the P- or T-axis areas that correspond to the focal fault movement; (3) If there is an active fault between a station and an epicentral area, and the fault strike is along or close to the azimuth of the foregoing P- or T-axis, then, no imminent anomaly is usually recorded at the station in the period when approaching the EQ, or an anomaly is generally weak in amplitude.

For example, in the Songpan-Pingwu area in Sichuan Province, three EQs with magnitude of Ms7.2, Ms6.7 and Ms7.2 occurred successively within 8 days in August 1976 at almost the same location (Fig.7), under the action of the compressive stresses in the NE direction (for the previous Ms7.2 EQ) and near to the EW direction (for the following Ms6.7/7.2 EQs)[16].

through the N30oE channel was -5.3%, and that recorded through the N60oW channel was - 4.9%. This indicated that the media here had been under the maximum compressive stress in the NWW-SEE direction during the period. The maximum compressive stress directions that were revealed by the AR changes at the two stations basically corresponded to the Paxis azimuth of the main shock[12]. Station JYO was near the northern aftershock area of the great EQ, and two channels, along the N70oW and N10oE directions, are employed at the station. The comparison between anomaly amplitudes recorded through the two channels was still credible as the measuring instrument at the station had just a fixed error. The anomaly amplitude recorded through the N70oW channel was -1.1% and no anomaly was recorded through the N10oE channel. This indicated that the media here had been under the maximum compressive stress close to the NS direction during the period, which well corresponded to the

In summary, the behaviors of these medium-term AR anomalies, such as their spatial distribution within 400 km, their tectonic relevance to the Songpan-Gazi active block, and their amplitude attenuation with increasing distance, as well as their synchronous, mostly descending, large-amplitude and anisotropic changes, strongly support that these anomalies are indeed related to the focal processes of the main shock and strong aftershocks. Such locally concentrative AR anomalies as recorded at the four stations have not been succeeded

During the short-term period before the occurrence date of the great EQ, no obvious anomaly, like these recorded at station CLH (72km) in Hebei province before the 1976 M7.8 Tangshan EQ and at station WDH (110 km) before the 1976 M7.2 Songpan-Pingwu EQs (Fig.2), was recorded at the six stations within 400 km from the EQ epicenter. Upward AR changes commenced at station CDU from March 2008 and at station JYO from April 2008 (Fig.5a-b and Fig.6). The patterns consisting of an initial medium-term drop change followed by a short-term rise change before the EQ are consistent with these of the electrical resistivity change within an EQ focal area that are forecasted by DD model[13-14]. In fact, such patterns appeared often nearby epicentral areas of previous strong EQs[9, 2,7]. However, the rise change, upon which the main shock occurred, did not satisfy the anomaly criterion in amplitude, so no sufficient precursory information could be confidently detected in the

Du *et al*[3, 15] studied the spatial distribution characteristics of imminent AR anomalies before the moderate, strong EQs that occurred in the continent of China. As a result, it was got that the spatial distribution of the anomalies was influenced by mainly active faults around EQ focal areas and focal mechanisms. The influences include: (1) the anomalies appear mostly along or nearby the faults; (2) most anomalies are distributed in the two areas that are symmetrical about an epicenter and that azimuthally tie to the P- or T-axis areas that correspond to the focal fault movement; (3) If there is an active fault between a station and an epicentral area, and the fault strike is along or close to the azimuth of the foregoing P- or T-axis, then, no imminent anomaly is usually recorded at the station in the period when

For example, in the Songpan-Pingwu area in Sichuan Province, three EQs with magnitude of Ms7.2, Ms6.7 and Ms7.2 occurred successively within 8 days in August 1976 at almost the same location (Fig.7), under the action of the compressive stresses in the NE direction (for the previous Ms7.2 EQ) and near to the EW direction (for the following Ms6.7/7.2 EQs)[16].

P-axis azimuths of most strong aftershocks in the northern aftershock area.

in identifying in areas beyond 400 km from the EQ epicenter.

approaching the EQ, or an anomaly is generally weak in amplitude.

**2.1.2.2 Short-term AR anomalies before the EQ** 

period when approaching the EQ.

And the P-axis azimuths of the three EQs were 63°, 101° and 95°, respectively, resulting in an right-lateral movement of the NNW-striking Huya fault (i.e., the focal fault) when the previous Ms7.2 EQ occurred, and then an left-lateral movements of the fault when the following Ms6.7 and Ms7.2 EQs occurred[16]. At that time, three stations were in observation within a range of ~190 km from the three epicenters: station Songpan (SPN, 45km, and it was abandoned shortly afterwards) in Sichuan Province and station Lixian (LXN, 190 km, and abandoned in the 1990s) and station WDH (110 km), in Gansu Province. There were obvious differences between AR changes recorded at station SPN and those recorded at station LXN (Fig. 8). Station SPN was located on the west side of the Minjiang River fault (MRF) with the NNE strike, whereas the epicenters were to the east of the fault. As a result, no imminent AR anomaly was recorded at station SPN when approaching the occurrence date of the first Ms7.2 EQ, though the station was only 45 km from the epicenter, whereas at LXN station, a drop anomaly with amplitude of -0.8% was recorded during 4 days before the EQ in despite of being 190 km away. Contrarily, for the subsequent Ms6.7 and Ms7.2 EQs, a drop anomaly with amplitude of -3.5% was clearly captured at station SPN during 3– 4 days before them. This anomaly amplitude was much greater than that recorded at station LXN before the first Ms7.2 EQ. No imminent anomaly appeared at station LXN for the latter EQs. At station WDH, which was located between station LXN and the three epicenters and was near to the north side of the White Dragon River fault (WDRF) with the EW strike, the AR changes recorded before the three events were similar to those seen at station LXN.

Fig. 7. Distribution of epicenters, active faults and AR stations

According to papers[4, 17], imminent anomalies recorded at underground water level and water chemistry stations before the moderate, strong EQs which occurred in the continent of China also demonstrated such spatial distribution as AR imminent anomalies, which was related to mainly active faults of epicentral areas and EQ focal mechanisms. From loading

Changes in Apparent Resistivity in the Late Preparation Stages of Strong Earthquakes 207

On Oct. 25, 2003, two EQs with magnitude of Ms6.1 and Ms5.8 occurred in Minle-Shandan area in Gansu province of China. This year, on Jul. 21, one Ms 6.2 EQ occurred in the Dayao area in Yunnan province, and on Oct. 6, a Ms6.1 EQ occurred also in Dayao area. In fact, the medium-term AR anomalies were well discerned before the two groups of strong EQs, and the EQ locations and magnitudes were successfully forecasted on a one-year time scale, in

Within the range of ~400 km from the epicenters of the Ms 6.1/5.8 Minle- Shandan EQs, there were 6 AR stations, such as Shandan (SHD), Wuwei (WWE), Jayuguan (JYG), Lanzhou (LZL), Linxia (LNX) and Dingxi (DGX) stations, which are all set by China Earthquake Administration (CEA). Of 6 stations, the previous 5 stations, not including DGX station, were in good observation at that time. Of the 5 stations, reliable medium-term AR anomalies, drop-type AR changes, appeared at station SHD during the end of 2002. This station has always kept an observational environment which is up to the technical requirement of seismic geo-electrical station in the long-term observation. Figure 9 shows AR normalized variation rate curves of three channels of this station from Aug. of 1988 to Oct. of 2002, based on monthly AR averages, which are processed by the normalized variation rate method (NVRM)[9,7] in Nov. of 2002. It can be seen from the curves that notable medium-term or short-term AR anomalies which were up to the identification criterion for NVRM anomaly, beyond the threshold value of 2.4, appeared before several EQs with magnitude of Ms 5.0 around the station from 1990 to 2001. In 2002, two droptype AR anomalies appeared again through channels EW and NW of the station, and a risetype anomaly did through channel NS. These anomalies were new, following the rise-type anomalies of three channels in the end of 2001 which corresponded to the Nov. 2011 Kunlunshan mountain Ms8.1 EQ far-one type of anomalies which had nothing with the focal process of the great EQ[20, 7-8]. According to the past EQ cases of the station where AR anomalies corresponded to EQs around and the research results in papers [8,10], Du *et al*. forecasted in Nov. 2002 that one strong EQ will occur nearby the station in 2003 year.

**2.2 Medium-term AR anomalies discerned before EQs** 

Note: these curves were drown in Nov. of 2002

Fig. 9. NVRM curves of AR changes of SHD station (from Aug. of 1988 to Oct. of 2002)

Nov. of 2002.

experiments of rock sample[18], the crust medium nearby active faults are susceptive to stress disturbances during the loading processes, therefore, the AR and underground-water anomalies which is related to the geo-stress changes are easy recorded nearby the faults. Paper [19] calculated the strain distribution in a geologic body model with a fault using the elasto-plastic 2-D finite element method, and then the calculated strain values are converted into relative AR changes. It can be seen from the spatially non-uniform distribution of these calculated AR changes that the spatial distribution of AR changes which are related mainly active faults around epicentral areas and EQ focal mechanisms, as described in papers [3,15,7], can be well explained. In fact, the spatially non-uniform distribution of imminent anomalies in observation of underground water level and water chemistry, as described in papers [4, 17], can be well explained based on the calculated strain distribution by paper [19] also.

Fig. 8. AR daily-mean curves of SPN (a), WDH (b) and LXN (c) stations

Based on the above-mentioned research works, the lack of imminent AR anomalies before the Wenchuan great EQ can be roughly explained. The main shock and strong aftershocks occurred to the west of the NE-striking Doujiang Weir-An county fault; station GAZ was located on the southwestern side of the NW-striking Xianshui River fault, and station WDH was located on the northern side of the NWW-striking WDRF (Fig.4 and Fig.7). Hence, the reason why no anomaly was recorded at stations GAZ and WDH during the period when approaching the great EQ can be explained. The reason why no anomaly was seen at stations MNI and XCM can be explained also. Stations CDU and JYO were located to the east of the Doujiang-Weir-An county fault, near the main epicenter and in the area liable to record the anomaly related to the main shock, but at the two stations only weak upward changes were recorded during the period. The reason for this remains to be explained and requires additional research.-A possible reason is affected by secondary faults nearby the two stations.

#### **2.2 Medium-term AR anomalies discerned before EQs**

206 Earthquake Research and Analysis – Statistical Studies, Observations and Planning

experiments of rock sample[18], the crust medium nearby active faults are susceptive to stress disturbances during the loading processes, therefore, the AR and underground-water anomalies which is related to the geo-stress changes are easy recorded nearby the faults. Paper [19] calculated the strain distribution in a geologic body model with a fault using the elasto-plastic 2-D finite element method, and then the calculated strain values are converted into relative AR changes. It can be seen from the spatially non-uniform distribution of these calculated AR changes that the spatial distribution of AR changes which are related mainly active faults around epicentral areas and EQ focal mechanisms, as described in papers [3,15,7], can be well explained. In fact, the spatially non-uniform distribution of imminent anomalies in observation of underground water level and water chemistry, as described in papers [4, 17],

can be well explained based on the calculated strain distribution by paper [19] also.

Fig. 8. AR daily-mean curves of SPN (a), WDH (b) and LXN (c) stations

Based on the above-mentioned research works, the lack of imminent AR anomalies before the Wenchuan great EQ can be roughly explained. The main shock and strong aftershocks occurred to the west of the NE-striking Doujiang Weir-An county fault; station GAZ was located on the southwestern side of the NW-striking Xianshui River fault, and station WDH was located on the northern side of the NWW-striking WDRF (Fig.4 and Fig.7). Hence, the reason why no anomaly was recorded at stations GAZ and WDH during the period when approaching the great EQ can be explained. The reason why no anomaly was seen at stations MNI and XCM can be explained also. Stations CDU and JYO were located to the east of the Doujiang-Weir-An county fault, near the main epicenter and in the area liable to record the anomaly related to the main shock, but at the two stations only weak upward changes were recorded during the period. The reason for this remains to be explained and requires additional research.-A possible reason is affected by secondary faults nearby the two stations.

On Oct. 25, 2003, two EQs with magnitude of Ms6.1 and Ms5.8 occurred in Minle-Shandan area in Gansu province of China. This year, on Jul. 21, one Ms 6.2 EQ occurred in the Dayao area in Yunnan province, and on Oct. 6, a Ms6.1 EQ occurred also in Dayao area. In fact, the medium-term AR anomalies were well discerned before the two groups of strong EQs, and the EQ locations and magnitudes were successfully forecasted on a one-year time scale, in Nov. of 2002.

Within the range of ~400 km from the epicenters of the Ms 6.1/5.8 Minle- Shandan EQs, there were 6 AR stations, such as Shandan (SHD), Wuwei (WWE), Jayuguan (JYG), Lanzhou (LZL), Linxia (LNX) and Dingxi (DGX) stations, which are all set by China Earthquake Administration (CEA). Of 6 stations, the previous 5 stations, not including DGX station, were in good observation at that time. Of the 5 stations, reliable medium-term AR anomalies, drop-type AR changes, appeared at station SHD during the end of 2002. This station has always kept an observational environment which is up to the technical requirement of seismic geo-electrical station in the long-term observation. Figure 9 shows AR normalized variation rate curves of three channels of this station from Aug. of 1988 to Oct. of 2002, based on monthly AR averages, which are processed by the normalized variation rate method (NVRM)[9,7] in Nov. of 2002. It can be seen from the curves that notable medium-term or short-term AR anomalies which were up to the identification criterion for NVRM anomaly, beyond the threshold value of 2.4, appeared before several EQs with magnitude of Ms 5.0 around the station from 1990 to 2001. In 2002, two droptype AR anomalies appeared again through channels EW and NW of the station, and a risetype anomaly did through channel NS. These anomalies were new, following the rise-type anomalies of three channels in the end of 2001 which corresponded to the Nov. 2011 Kunlunshan mountain Ms8.1 EQ far-one type of anomalies which had nothing with the focal process of the great EQ[20, 7-8]. According to the past EQ cases of the station where AR anomalies corresponded to EQs around and the research results in papers [8,10], Du *et al*. forecasted in Nov. 2002 that one strong EQ will occur nearby the station in 2003 year.

Note: these curves were drown in Nov. of 2002

Fig. 9. NVRM curves of AR changes of SHD station (from Aug. of 1988 to Oct. of 2002)

Changes in Apparent Resistivity in the Late Preparation Stages of Strong Earthquakes 209

related to the two strong EQs were truly recorded before the occurrence of the two EQs. In 2005, the case for forecasting the two strong EQs was in public reported in paper [21].

Note: (1) Red solid circles are EQ epicenters that occurred in 2003; (2) Green areas are forecast areas where EQ will occur in 2003; (3) Black solid triangles are AR stations where AR anomaly appeared in 2002

Fig. 11. Distributions of epicenters, AR stations and forecast areas

Fig. 12. NVRM curves of AR changes of SHD station (from 1997 to 2004)

Note: these curves were drown in 2004

In Nov. of 2002, Du *et al*. also detected that AR anomalies appeared at Panzhihua (PAH) and Yuanmou (YNM) in Yunnan province. The AR data, monthly mean data, observed by the two stations were processed using NVRM. Figure 10 gives the NVRM curves of PAH station which were reprocessed in 2004. This station was settled in 1970s, around which many EQs with magnitude of Ms 5.0 have occurred since then, accordingly, at this station AR anomalies have been recorded before the EQs. According to the previous EQ monitoring efficiency of the station, Du *et al*. forecasted in Nov. of 2002 that the AR upward anomaly of 2002 at this station, as seen in figure 10 (the lower curve in the figure ), possibly indicated an future strong EQ nearby the station. Besides, at YNM station which was close to the PAH station, AR upward anomalies appeared in the end of 2002 also. Thus, the credibility for the expected EQ was increased.

Note: these curves were drown in Nov. of 2004

Apart from these above-mentioned, Du *et al*. processed the observation data of other AR stations in the continent of China in Nov. of 2002, as a result, in other places some AR anomalies appeared in 2002 also. Thus, Du *et al*. submitted an EQ forecast view to CEA, a formal written report in Nov. of 2002. In the report, Du *et al*. believed that several EQs probably occurred in 2003 year within 4 regions with the maximal radius being less than 100 km and the minimal radius being less than 60 km, and expected EQ magnitudes were Ms6.0±, Ms≥6.0, Ms5~6 and Ms≥5 (Fig.11), respectively. As a result, the EQs as expected in this report indeed occurred in the former two regions in 2003. It is the case that the Oct. 25, 2003, Ms6.1 and 5.8 Minle~Shandan EQs just occurred in the forecast region with magnitude of Ms6.0±, in Gansu province; and the Jul. 21 Ms6.2 and Oct. 10 Ms6.1, 2003, Dayao EQs just occurred in the forecast region with magnitude of Ms≥6.0, in Yunnan province (Fig.11). In figure 11, the two forecast regions were marked in green color, painted in Nov. 2002, and the two solid circles in red are the locations of occurrence of the two groups of EQs. In other two forecast regions with magnitude of Ms5~6 and Ms≥5, the expected EQs did not occur, and still nearby the forecast region with magnitude of Ms 5~6, in Shanxi province, one Ms5.1 EQ happened in Nov. 2003. The former two strong EQs are successfully predicted, which locations and magnitudes are properly estimated and which occurred in 2003 year, a one-year time scale prediction, therefore, it is reasonable to believe that the AR changes

Fig. 10. NVRM curves of AR changes of PAH station (from Jan. of 199 to Dec. of 2003)

In Nov. of 2002, Du *et al*. also detected that AR anomalies appeared at Panzhihua (PAH) and Yuanmou (YNM) in Yunnan province. The AR data, monthly mean data, observed by the two stations were processed using NVRM. Figure 10 gives the NVRM curves of PAH station which were reprocessed in 2004. This station was settled in 1970s, around which many EQs with magnitude of Ms 5.0 have occurred since then, accordingly, at this station AR anomalies have been recorded before the EQs. According to the previous EQ monitoring efficiency of the station, Du *et al*. forecasted in Nov. of 2002 that the AR upward anomaly of 2002 at this station, as seen in figure 10 (the lower curve in the figure ), possibly indicated an future strong EQ nearby the station. Besides, at YNM station which was close to the PAH station, AR upward anomalies appeared in the end of 2002 also. Thus, the credibility for the

Fig. 10. NVRM curves of AR changes of PAH station (from Jan. of 199 to Dec. of 2003)

Apart from these above-mentioned, Du *et al*. processed the observation data of other AR stations in the continent of China in Nov. of 2002, as a result, in other places some AR anomalies appeared in 2002 also. Thus, Du *et al*. submitted an EQ forecast view to CEA, a formal written report in Nov. of 2002. In the report, Du *et al*. believed that several EQs probably occurred in 2003 year within 4 regions with the maximal radius being less than 100 km and the minimal radius being less than 60 km, and expected EQ magnitudes were Ms6.0±, Ms≥6.0, Ms5~6 and Ms≥5 (Fig.11), respectively. As a result, the EQs as expected in this report indeed occurred in the former two regions in 2003. It is the case that the Oct. 25, 2003, Ms6.1 and 5.8 Minle~Shandan EQs just occurred in the forecast region with magnitude of Ms6.0±, in Gansu province; and the Jul. 21 Ms6.2 and Oct. 10 Ms6.1, 2003, Dayao EQs just occurred in the forecast region with magnitude of Ms≥6.0, in Yunnan province (Fig.11). In figure 11, the two forecast regions were marked in green color, painted in Nov. 2002, and the two solid circles in red are the locations of occurrence of the two groups of EQs. In other two forecast regions with magnitude of Ms5~6 and Ms≥5, the expected EQs did not occur, and still nearby the forecast region with magnitude of Ms 5~6, in Shanxi province, one Ms5.1 EQ happened in Nov. 2003. The former two strong EQs are successfully predicted, which locations and magnitudes are properly estimated and which occurred in 2003 year, a one-year time scale prediction, therefore, it is reasonable to believe that the AR changes

expected EQ was increased.

Note: these curves were drown in Nov. of 2004

related to the two strong EQs were truly recorded before the occurrence of the two EQs. In 2005, the case for forecasting the two strong EQs was in public reported in paper [21].

Note: (1) Red solid circles are EQ epicenters that occurred in 2003; (2) Green areas are forecast areas where EQ will occur in 2003; (3) Black solid triangles are AR stations where AR anomaly appeared in 2002 Fig. 11. Distributions of epicenters, AR stations and forecast areas

Fig. 12. NVRM curves of AR changes of SHD station (from 1997 to 2004)

Changes in Apparent Resistivity in the Late Preparation Stages of Strong Earthquakes 211

For example, the two observation channels, N20°E and N70°W channels, are installed at station PGU that had epicentral distances of 111 and 140 km for the 1976 Ms7.8 Tangshan EQ and Ms7.1 Luanxian aftershock; the P-axis azimuths of the two events were 75° and 297°, respectively, roughly in the EW direction. As a result, the medium-term drop-type and short-term rise-type AR changes recorded through N20°E channel (with a near NS direction) prior to the EQs were greater in amplitude than those through N70°W channel (with a near EW direction) (Fig.13). As another example, three channels, NS, EW and N45°W channels, are installed at station SHD that had an epicentral distance of 43 km for the 2003 Ms6.1 Minle- Shandan EQ; the P-axis azimuth of this EQ was 65°. As a result, the medium-term drop-type and short-term rise-type AR changes recorded through N45°W channel prior to

It is obvious that the relationship between the anisotropic AR changes and the P-axis azimuth from actual EQ cases agrees well with the relationship between the directional AR changes and the maximum loading direction in most experiments of water-bearing rock (or soil) samples. This proves that such AR changes are just related to the EQ preparation process.

Fig. 13. NVRM curves of AR changes of station PGU for the 1976 Ms 7.8 Tangshan EQ

According to DD model, the micro cracks inside the underground medium fast and nonlinearly develop immediately before the main rupture within an EQ focal area, their strikes align predominately in a certain direction and underground water fast come in them. Barsukov[29] interpreted a larger amplitude change in resistivity on the assumption that the micro crocks are tortuously linked each other, and underground water comes in them, as a result, the conductive aisles inside the medium are formed. Mei *et al*.[30] deduced that in the later preparation stages of strong EQs a number of micro-cracks inside the medium of an EQ focal region is increased sharply. The fact that in the medium-term statges before moderate, strong EQs of the Chinese mainland, for larger-magnitude EQs the AR anomaly amplitude increases fast, whereas the anomaly duration increases more slowly[10] supports Mei's deduction. According to Crampin *et at*.[31], the maximum compressional stress inside the

**3.2 Theoretical analysis on anisotropic AR changes** 

the EQ were greater in amplitude than those through EW channel (Fig. 12).

After the Ms6.1 and 5.8 Shandan-Minle EQs, Du *et al* again processed the AR data observed at SHD station (43 km) from the two epicentral areas 1997 to 2004 using NVRM, as a result, it can be seen from figure 12 that the medium-term drop-type to short-term rise-type AR changes are similar to the AR changes in the focal area as foretold by the DD model (Dilatancy-Diffusion Model)[13-14] in appearance, a pattern consisting of an initial mediumterm fall followed by a short-term rise.

People may ask why the May 12, 2008, Ms8.0 EQ was not forecasted in the medium-term period before the great EQ, which just occurred around/nearby 6 AR stations in this area? In fact, authors had no time to process and analyze the AR data observed at the 6 AR stations in those days, and the AR changes of the stations were analyzed and studied only after the great EQ[7].
