5.1 Near-fault strong motion directionality and caused damage pattern
Two recordings are observed close to the epicenter with a distance of fewer than 10 km. We investigate near-fault strong motion directionality by comparing the observation in the fault parallel (FP) and fault normal (FN) direction. Figure 8a-b show the FP and FN components of acceleration (Fig. 8a) and velocity (Fig. 8b) time series and response spectra for 5% damping ratio at these two stations of 53YBX and L2203 which are located in the intensity VIII and VII zone respectively in Yangbi county (see Fig. 1). The fault parallel PGV of the 53YBX record is over 2 times larger than the fault normal, and the fault parallel PGA and response spectra at periods over 0.4 s are much higher than the fault normal. The L2203 record observed in the intensity VII zone also shows an apparent discrepancy between the FP and FN components, with higher PGA and PGV in the FP direction. The comparison of ground motions observed in the FP and FN direction demonstrates strong directionality in the near-fault during this event. Meanwhile, the acceleration response trace shown in Fig. 8c-f indicate a clear predominant direction along the FP for near-fault ground motion in Yangbi county town, especially at periods of 0.5, 1.0 and 3.0 s.
From June 6 to 10, 2021, a joint scientific research team of the Institute of Geophysics, CEA, Beijing University of technology, and Yunnan Earthquake Administration conducted an earthquake damage survey on the most severely damaged Yangbi county and surrounding villages, as well as the ancient city and urban area of Dali. It mainly investigates the types of housing structures, including brick concrete structure, frame structure, earth-wooden structure, and other rural dwellings. Figure 9 shows the location of seismic damage investigation sites and deployed topographic array in the most severely damaged Yangbi county. Figure 10a-b show two teaching buildings (RCF structure) in a primary school of Xiajie village in Yangbi county town. The teaching building along the FN direction is basically intact and able to be used normally. In contrast, the other one along the FP direction is damaged with obvious cracks and wall spalling and is suspended for use. Figure 10c shows a comparison of two earth-wooden houses in Huaian village north of Yangbi county town. The tiles of the houses facing FN direction are relatively intact, and the houses facing FP direction are seriously damaged with shuttle tiles and walls. The strong directionality of near-fault ground motion was observed in this earthquake; that is, ground motion in FP direction is greater than that in FN direction. This phenomenon can explain the typical seismic damage differences in the FN and FP direction and might be related to the rupture direction of the seismogenic fault. We also compared the two horizontal response spectra of stations 53YBX and L2203 observed during the aftershocks. We find no clear directionality in the near-fault ground motion; that is, ground motion amplitudes are very close between the two horizontal components. We consider that the strong directionality observed in the mainshock can not be caused by the local site effect, while it should be related to the focal mechanism of the mainshock.
5.2 Comparison of near-fault response spectra with building code in China and observations of structure damage
Figure 11 illustrates the comparison of observed near-fault response spectra (5% damping) with code design spectra in China. The dominant periods of horizontal response spectra of record 53YBX and L2203 are small and about 0.1 s and 0.2 s, respectively. In the intensity VIII zone, the FP and FN component spectra of the 53YBX record are even higher than the code design spectra for the intensity IX zone at period around 0.1 s. However, at periods of 0.5-3.0 s close to the fundamental period of building structures, both the FP and FN components are lower than the code design spectra for the intensity VIII zone. In the earthquake intensity VII zone, the L2203 record has spectra values larger than the code design spectra for intensity VII zone at period around 0.3 s. Still, at periods over 0.5 s, the spectra are much lower than the code spectra.
Figure 12 shows the typical earthquake damage phenomena in (a-e) Xiajie village and (f-g) Laojie of Yangbi county town and (h-i) Huaian village, such as collapse and shear failure of filled wall of earth-wooden structures (Fig. 12a-b), overall deviation of the brick-concrete house caused by foundation failure (Fig. 12e-f), shear failure of brick concrete courtyard wall (Fig. 12h). Among them, Fig. 12c-e shows the overall offset of a brick concrete house (left side) in Xiajie village, colliding with adjacent houses due to foundation failure. A distance of 50 cm was reserved between the two buildings, but they touched each other after the earthquake. The earthquake also caused the building on the left to rotate. This damage may result from the shallow foundation and slope in the foundation of building, and the asynchronous displacement occurred in the earthquake.
We find that most of the gables of local earth-wooden houses are rammed walls, which are seriously damaged in this earthquake. The damage includes peeling and cracking of gables, cracking at the junction of vertical and horizontal walls, flash collapse of some gables (Fig. 12a-b, i). The wooden structures of civil buildings are basically intact, but some wooden columns are damaged due to the decline of bearing capacity (Fig. 12g). The seismic damage of frame and brick concrete buildings is relatively light, and the main damage is tile falling, wall peeling and infilled wall cracks. The brick concrete structure with ring beam and structural column structure generally has good seismic performance with less damage and fewer cracks on the wall.
Figure 13 shows the typical earthquake damage in Xiuling village near the epicenter. Xiuling village is located about 15 km northwest of Yangbi county town. The survey indicates that the Xiuling village suffers the hardest during this earthquake. Figure 13a-b show the overall collapse of the adobe wall of two earth-wooden houses, with residual wooden load-bearing pillars, which are basically intact. The wall thickness reaches 50-60cm, made of local cohesive soil and water without adding traditional cohesive media such as rice straw or glutinous rice. The adobe wall itself has a large self-weight and lacks shear capacity. In addition, the wall is located outside the column and lacks a restraint connection with the column, so it is easy to collapse in an earthquake. Figure 13c shows the damage of a brick-concrete building of Xiuling village neighborhood committee. The aged wooden frame on the roof of the building collapsed and a large number of tiles fell, but the wooden load-bearing pillars were basically intact without fracture and collapse. Figure 13d reflects the impact of the strong polarity of near-fault ground motion in Xiuling village. The tiles of the front house along the FN direction are relatively intact, but serious shuttle tiles appear in the rear houses along the FP direction. In general, the earthquake damage in Xiuling village is more serious than that in Yangbi county town. Xiuling village is located on the ridge near the top of the mountain, and the mountain inclination is about 30°, resulting in an obvious topographic amplification effect.
According to the earthquake damage investigation in Yangbi county and its surrounding areas, the newly-built brick concrete and frame structures constructed following the building code have good performance. In contrast, old brick concrete structures with a serious decline in bearing capacity and earth-wood houses without following the regulation are seriously damaged. Characteristics of observed near-fault response spectra can explain the damage pattern observed in the near-fault for different building structures.
The 21 May 2021 Yangbi M 6.4 earthquake was well recorded by a dense network of strong motion stations and intensity stations in China. The observation shows that the spatial variability of near-source strong ground motions is strongly affected by the earthquake source and local site effects. The recordings clearly demonstrate the characteristics of short duration (< 15-20 s) and small (0.1-0.5 s) dominant period of strong ground motion in the near-source. Observed PGVs and long period SAs over 1.0 s are relatively weak and attenuate rapidly with distance, showing a good coincidence with the damage spatial variation pattern.
The anomalies in the distribution of near-source strong ground motions are mainly caused by local site effects due to topography and soft sediments. Local topography mainly affects PGA and short-period response spectrum values below 1.0 s, which are observed by intensity stations co-located in communication base stations. The reason for strong topography effects is that these communication base stations are deployed on hills to get good signals. Meanwhile, the soft covering soil has strong amplification effects in the long-period response spectrum values over 1.0 s. The attenuation of ground motion IMs also reveals two kinds of different local site effects.
Near-fault ground motion shows strong directionality with a clear predominant direction, leading to different building damage patterns along the FN and FP directions. The observed near-fault motions at short period around 0.1 s can be higher than the code design spectra in China, but they are much lower than the design spectra at periods close to the fundamental periods of building structures. Characteristics of near-fault strong motion provide a good explanation for the damage phenomenon observed in Yangbi county town. The earth-wooden houses are seriously damaged, with rammed earth gable and outer longitudinal wall totally collapsed, and it is difficult to repair after the earthquake. However, the newly-build fortification buildings of brick-concrete and frame structures performed well in the earthquake. We consider improving the seismic fortification of engineering structures to be the most effective means to reduce earthquake disasters.
Data and Resources
The strong motion dataset consists of data provided by China Earthquake Networks Center (CENC), available by application through the website of National Earthquake Data Center (http://data.earthquake.cn) (last accessed on 20 August 2021). The geophysical maps produced in this study are produced using Generic Mapping Tools (GMT; Wessel and Smith, 1991).