The parameters and uncertainties of the C4 catalogue were determined by methods and practices developed and used by experts over 20 years ago, using their experience and personal judgment, and often without documenting the reasoning. In interpreting the catalogue, it is therefore necessary to be aware of this lack of documentation (Min et al., 1995). Here we address some issues and opportunities that have developed since 1995, notably:
(a) Improved understanding of the historical context, notably the practices of local administration of data collection, documenting and zoning (Tan 1996a; Tan 1996b),
(b) Better models for the attenuation of macroseismic intensity (Wang et al., 1998; Wang et al., 2000),
(c) Use of topographic maps to interpret the historical records for relative spatial information (Wang, 2011; Wang et al., 2020). For example, in some cases, there were the descriptions in original records such as “heavy damage near a mountain or a river”,
(d) The recognition that seismic density index (defined in section 3.1) can retain a long memory of historical events in China (Wang et al., 2017) as well as elsewhere.
In this section, we apply these four post-C4 developments to reassess the locations of historical earthquakes in the study area, within their stated uncertainties. As a result, we have re-assigned epicentres and/or magnitudes to four historical events, as indicated in bold in Table 5, but could find no reason to do so for the other two listed in Table 1. Of those with re-assigned parameters, we start with the 1829 earthquake, because it has the greatest potential for re-examination of local historical data with a large and more detailed archive of contemporary accounts of historical earthquakes (Xie et al., 1987; Min et al., 1995), including archives not considered completely in compiling the National C4 catalogue.
4.1 The 19 November 1829 M6¼ earthquake
This earthquake occurred in the Qing Dynasty, and has an epicenter in the C4 catalogue at the mid-point between the regional capitals of Yidu and Linqu (blue circle in Fig. 5), with a reported error in location of ± 25km and an estimated magnitude of 6¼ (Table 1, Min et al., 1995). Noting the practice of assigning intensity data at local capitals, we tested the hypothesis that the C4 estimate was optimal. First, we scrutinized the original records for critical observations of damage that could be assigned with confidence to a particular location. For example, “The damage is relatively heavy in Yidu and Linqu, especially for some villages where houses are built close to the mountains. Yidu: total 28 persons died and 7 047 houses collapsed or were destroyed in 27 villages. Linqu: total 89 persons died and 9 544 houses collapsed or destroyed. Another key constraint is “Boshan: cracks occurred in the mountain. Zouping: some old houses were damaged” (Xie et al., 1987, p. 811–818).
In Yidu, the archives report some 27 villages that suffered damage, as well as the total number of fatalities and damaged houses, including those in the 27 villages. While it is possible to estimate a macro-seismic intensity from the records, it is not possible to assign these to a particular village, because neither their names nor their precise locations are reported in the original records. Hence, the compilers of C4 assigned the maximum observed intensity values of Ⅷ to the capital towns of Yidu and Linqu. Meanwhile, a lower intensity value of Ⅵ was assigned to Boshan and Zouping (Min et al., 1995, p. 339). In the absence of other information, the epicenter was then estimated in C4 at the midpoint between Yidu and Linqu (blue circle in Fig. 5).
Figure 5. Topography around the epicenters and the model isoseismals of the 1829 earthquake using the updated best estimate for the location (red star) and magnitude of the event. The C4 epicenter is shown for reference (blue circle).
The primary records highlight the importance of topography, also illustrated in Fig. 5, as a reference point. The mountain ranges lie to the south-west of the C4 epicentre. According to “The Historical Atlas of China, Vol. 8—Qing Dynasty”, there was a town name as “Wujing” located at the boundary of the mountain (Tan, 1996b) whose reported damage would likely have been included in the records for the regional capitals. There is also a present-day seismic density anomaly just west of Wujing on Fig. 5, also near an active fault as shown on the map. If we assume the seismic density index represents a memory of past events, then it may be better to use the center of the modern day seismic density anomaly near Wujing as a candidate alternate epicenter. This location is more consistent with the available information from the re-examined historical data and the topography than the C4 epicentre, located in the mountains south-west of the C4 epicentre, but within its uncertainty of ± 25km, as shown by the red star on Fig. 5.
For a given candidate epicentre, the macro-seismic intensity contours can then be estimated based on the available historical data and a calibrated (isotropic) attenuation law for a range of trial source magnitudes (e.g. Wang et al., 1998). This unavoidably leads to concentric circular intensity contours on Fig. 5 rather than the more complex structures that might result with a greater density and spatial coverage of primary intensity data points. The parameters of the specific calibrated attenuation law used are listed in Table 4 for different candidate source magnitudes (after Wang et al., 2020).
Table 4
Equivalent radius for the isoseismal area as a function of intensity (columns) in three magnitude bins (rows)
Magnitude bin | | I(Ⅵ) | I(Ⅶ) | I(Ⅷ) | I(Ⅸ) | I(Ⅹ) |
6.0-6.4 | Radius (km) | 40.2 | 16.2 | 7.8 | | |
6.5–6.9 | Radius (km) | 69.8 | 35.5 | 16.8 | 9.9 | |
7.0-7.4 | Radius (km) | 158.1 | 76.3 | 32.7 | 16.7 | 8.5 |
By trial and error, a source magnitude in the range 6.5–6.9 is found to be more consistent with the historical data and the independent constraints mentioned above than the other magnitude bins shown in Fig. 5, notably the reported maximum intensity of Ⅷ assigned in C4 to Yidu and Linqu. The preferred model has an intensity Ⅷ covering the eastern periphery of the mountain, and intensity VI assigned to Boshan and Zouping. Given the new information from the attenuation law, we can rule out a source magnitude of 6.0-6.4, which would not be consistent with the Ⅷ intensity observed on the plain, where 27 villages suffered heavily damage over a wide area near the mountains. Similarly, a trial magnitude of 7.0-7.4 would be too high to explain the damage observed at Boshan and Zouping. Accordingly, we re-assigned the epicenter of 1829 earthquake to the center of the seismic density anomaly near Wujing, estimated its magnitude as 6¾, and reduced the estimated intensity at the regional capitals Yidu and Linqu to Ⅶ (Fig. 5). The re-assigned magnitude and miezoseismal intensity are both one category higher than that assigned in C4 (which had M6¼, I0=Ⅷ, Table 1).
The parameter uncertainties in C4 were determined by a previous generation of experts and there are often no explanations for the reasons. However, our understanding is that the distances between the individual intensity data points, or equivalently the density of locations recording ground shaking, is the main factor in controlling the uncertainty. In this case, the most importance intensity data points include Linqu and Yidu, a distance of about 20 km apart. The half distance is then 10 km. If there were only the two points, the uncertainty would have been given as ± 10km in C4. However, the intensities from other intensity data points, such as at Boshan and Zouping, require a higher epicenter uncertainty of ± 25km for this event.
4.2 The 1 June 70 B.C. M ≥ 7 earthquake
This earthquake occurred in 70 B.C. in the Han Dynasty, about 2090 years before present. In C4, the epicenter was put at the mid-point between the capital towns of Yingling and Dongwu (blue circle in Fig. 6), with a reported uncertainty of ± 100km. The miezoseismal intensity (I0) was estimated to be ≥Ⅸ and magnitude estimated only as the constraint M ≥ 7 in C4 (Table 1, Min et al., 1995). We now follow the same procedure as in section 4.1 to re-evaluate these estimates.
One key excerpt from the original records translates as: “There were 49 shires affected by an earthquake. In the Beihai shire and Langya shire, ancestral temples, town walls and houses were destroyed. Landslides occurred in the mountains and water gushed out. More than 6000 persons were killed.” (Xie et al., 1983, p. 8). This account is of typical of the historical earthquake record of the time in China, with damage descriptions that are concise and relative simple. It implies the felt region is large (49 shires affected), but that only two shires suffered heavy damage (Beihai shire and Langya shire). Unfortunately, there was no name of a specific town or village in the shires in the original record for this event. According to the conventional practice at the time, the damage would have been assigned to the capital town of a shire. Hence, the C4 epicenter was put at the mid-point between the two capital towns, which could introduce a large error in the estimated epicenter location (Min et al., 1995; Wang et al., 2020), represented by the large quoted uncertainty above. The distance between the capital towns is about 69 km, which is too far to guarantee both sites suffered an intensity as large as Ⅸ, because the equivalent radius for intensity Ⅸ in the magnitude 7.0-7.4 range (see below) is only about 16.7 km (Table 4, Wang et al., 2020). This rules out the possibility of an epicenter at the mid-point.
According to “The Historical Atlas of China, Vol. 2—Qin Dynasty and Han Dynasty” (Tan, 1996a), the administrative areas of the Beihai shire and Langya shire occupied the areas shown along with other shires of the time in Fig. 6. The figure also shows the locations of Han dynasty towns (purple triangles) and modern cities and towns (filled circles) for reference. The capital town of Beihai shire was named “Yingling” and the capital town of Langya shire was named “Dongwu”. From this map we would expect the damage associated with an event of this size at the C4 location to be significant at the capital town of Gaomi, in the shire of the same name, to the east of the C4 epicentre, similar to that reported for the shires of Beihai and Langya assigned to Yingling and Dongwu respectively. The original record did not mention Gaomi, despite it being a regional capital, implying the source is more likely to be further away and towards the west of Gaomi town than the C4 epicentre.
There is a seismic density anomaly to the west of the epicenter of C4 on Fig. 6, so we simulate the intensity pattern that would have arisen if the center of the anomaly is taken as a candidate epicenter instead of the C4 estimation. After the trial calculation, we find that magnitude of 7.0-7.4 is more reasonable (Table 4), and apply modern estimates of the intensity attenuation law to reconstruct the intensity contours. From the results of the simulation (Fig. 6) Wucheng town in the shire of Langya is located within the area of intensity Ⅹ consistent with the estimation of C4 that miezoseismal intensity ≥Ⅸ (Table 1, Min et al., 1995). The town of Zhuxu is just outside the circle of Ⅹintensity, also in Langya shire. There are other four towns, in the region of Ⅷ intensity in this shire on Fig. 6. All of these reports of damage would also have been assigned to Dongwu, the capital town of Langya Shire. The simulation consistent with the original record that heavy damage in Langya Shire, but the simulated intensity in Dongwu is only Ⅶ.
Apart from Yingling, the only known Han dynasty town in Beihai shire was Anqiu, the closest of all to the C4 epicentre on Fig. 6, due east of the candidate epicentre. Its damage reports would have been assigned to Yingling, so it is not possible to constrain the candidate epicenter from the east on the map. Both Yingling and Anqui are located close to the line of Ⅸ intensity on the simulation, representing significant damage. The simulated Ⅸ intensity region covers a more mountainous region to the south of the candidate epicentre. The simulated Ⅷ intensity circle just touches the western boundary of the shire of Gaomi, almost all of which is located within the intensity Ⅶ contour, and thus would not have experienced heavy damage. On the balance of evidence, we conclude that the candidate epicenter near Wucheng is more likely than the C4 estimate.
In this case, the only two intensity data points are available at Yingling and Dongwu, some 69 km apart. Accordingly, the uncertainty estimated in the ± 100 category in C4 seems too conservative. Hence we estimate the uncertainty estimated in the ± 50 category.
Figure 6. Administration zoning around the epicenters and the modelled isoseismals of the 70 B.C. earthquake.
4.3 The 26 July 1668 M6¾ earthquake
This earthquake occurred as an aftershock, just one day after the great Tancheng M8½ earthquake. In the C4 catalogue, the epicenter was put at Anqiu (blue circle in Fig. 7), with a reported error in location of ± 50km. The magnitude was estimated at M6¾.The earthquake was felt in a large area: the distance from the south point to the north point is about 900 km, but distance from east to west is difficult to estimate, because the eastern part is in the sea. All descriptions of the level of destruction were assigned to Anqiu in the original records. The description of the destruction quoted in (Min et al., 1995, p. 221) is “Anqiu: the earthquakes occurred day after day, and houses collapsed intermittently. Shandong: total 25 Xian affected, including some in the meizoseismal area of the main shock; Jiangsu: 8 Xian affected; Anhui: 2 Xian affected; Zhejiang: 2 Xian affected; Hebei: at least 3 places affected”. As a consequence, it is not possible to estimate the miezoseismal intensity I0. In contrast, the estimate of the felt area quoted above is well constrained by the list of towns affected. Anqiu is not located in the center of the felt region, so it is very unlikely to be the epicentre. The southernmost recording of felt effects was at Shaoxing about 600 km from Anqiu, and the northernmost at Tianjing, some 300 km from Anqiu. Moreover, no significant damage was recorded at the regional capitals of Weixian and Gaomi (Fig. 7), as might be expected from the C4 epicentre. Together, these observations imply the epicenter is more likely to be to the south of Anqiu.
Figure 7. The epicenters and isoseismals (blue dashed lines) of the earthquake on July 26, compared with those for the mainshock (black).
In the south of Anqiu, there is a seismic density anomaly centered near the historical town Guanshuai, so we tested a candidate epicenter at that location. After the trial calculation, we find that magnitude of 6.5–6.9 is more reasonable (in this case the same as the C4 magnitude), as illustrated in Fig. 7. Almost all of the area of intensity Ⅶ and some parts of the area for intensity Ⅵ for this event occur within the area of intensity Ⅹ of the M 8½ main shock, whereas the regional capital of Anqiu was outside this area. Peak intensity Ⅹ for the main shock implies all houses would have collapsed completely in this zone in the mainshock, which could explain the absence of any further destruction recorded for the later event there, except at Anqiu, which recorded intensity VI during the later event. The original epicentral location is biased towards the northern part of the felt region. While this is a possible solution, we conclude the revised location of 35.90°N and 119.06°E illustrated on Fig. 7 is a more likely epicenter than the C4 one also shown, but there is no need to revise the magnitude in this case. Unfortunately, there are no observations from outside Anqui to confirm this inference.
4.4 The 17 August 462 M6½ earthquake
This earthquake occurred in the Nanbeichao Dynasty. Its epicenter was placed near Weishan with a reported error in location of ± 100km, and its magnitude was estimated at M6½.The miezoseismal intensity was estimated as Ⅷ (Table 1, Min et al., 1995).
Excerpts from the original records translate as: “earthquake, noise came from north. In the Lu shire, the hills tottered and the ground shook. In Pengcheng (nowadays named as Xuzhou), the parapet of the town wall fell down and the houses toppled or collapsed. In Yanzhou, cracks appeared on the ground and water gushed. The shaking lasted on and off for more than two years”. There were only two places mentioned in the records. One is Pengcheng (117.20°E, 34.26°N) and another is Yanzhou (116.83°E, 35.54°N). According to the conventional method in China, the epicenter would have been assigned at the midpoint between these two capital towns. Independently, the Atlas of the Historical Earthquakes in China (referred to as ‘Atlas’ below) provides another valuable catalog for reference (Institute of Geophysics in State Seismological Bureau & Institute of Chinese Historical Geography in Fudan University, 1990). The original records quoted in C4 and in the Atlas are exactly same for this earthquake, but the epicenter for this earthquake given in the Atlas is around 10km further eastern than the C4 estimate (Table 5). In this case, we prefer the Atlas epicenter, because it is closer to the midpoint, but could see no reason to revise the magnitude or miezoseismal intensity. In this case, Yanzhou and Xuzhou are the key intensity data points, separated by a distance of about 148 km. Given half of this distance is larger than 50km, an estimated uncertainty in the ± 100 category seems reasonable.
To conclude this section, Table 5 summarizes where the catalog parameters have been updated (lower case label, e.g. ‘a’) or not (capital label, e.g. ‘B’ only) based on considering the new information presented above. The candidate zones identified by the seismic density anomalies on Fig. 8 are labelled by number (e.g. ‘Z1’) ordered with the oldest event first. The definition of the zones, based on the outlines of seismic density anomalies, is given in the next section.
Table 5
Parameters of strong historical earthquakes (M ≥ 6 in the C4 catalogue) near the Tan-Lu fault in eastern China associated with modern-day zones of seismic clustering, including those amended from the C4 catalogue in the present work (highlighted in bold).
Zone Number | Event Label | Date Y-M-D | Latitude | Longitude | M | I0 | Location error (km) | Geographical Location | Source |
Z1 | A | -70-06-01 | 36.3° | 119.2° | ≥ 7 | ≥Ⅸ | ± 100 | Northwest of Zhucheng | C4 |
a | -70-06-01 | 36.37° | 118.90° | 7¼ | Ⅹ | ± 50 | West of Anqiu | Amended |
Z6 | B | 462-08-17 | 34.8° | 117.0° | 6½ | Ⅷ | ± 50 | South of Yanzhou | C4 |
| b | 462-08-16 (Julian calendar) | 34.8° | 117.1° | 6½ | Ⅷ | ± 50 | North of Pengcheng | Atlas |
Z4 | C | 1668-07-25 | 34.8° | 118.5° | 8½ | ≥Ⅺ | ± 25 | Tancheng | C4 |
Z2 | D | 1668-07-26 | 36.4° | 119.2° | 6¾ | Ⅷ | | Anqiu | C4 |
| d | 1668-07-26 | 35.90° | 119.06° | 6¾ | Ⅸ | | Guanshuai | Amended |
z8 | E | 1668-09-18 | 36.2° | 117.1° | 6 | | | Northeast of Taian | C4 |
| e | 1668-09-18 | 36.2° | 117.1° | ≈ 6 | | | Northeast of Taian | Amended |
Z3 | F | 1672-06-17 | 35.6° | 118.8° | 6 | | | Juxian | C4 |
Z7 | G | 1829-11-19 | 36.6° | 118.5° | 6¼ | Ⅷ | ± 25 | Yidu | C4 |
| g | 1829-11-19 | 36.45° | 118.30° | 6¾ | Ⅸ | ± 25 | Southwest of Linqu | Amended |