The mainland China surface rupture database (MCSRD): development, structure, and descriptive analysis

Earthquake surface rupture zones serve as the most ideal windows for directly observing and studying seismogenic faults. Their structural characteristics-such as geometric morphology, combination patterns, and the distribution of coseismic deformation-provide crucial information for deeply understanding various scientific issues. These include the complexity of seismogenic faults (e.g., Fletcher et al., 2014;Hamling et al., 2017;Guo et al., 2020) and the dynamic processes of rupture propagation (e.g., Sibson, 2003;Wesnousky, 2006;Elliott et al., 2009). Furthermore, they shed light on the slip history of paleoearthquakes (e.g., Benedetti et al., 2013;Klinger et al., 2011;Salisbury et al., 2012), and the mechanical mechanisms of seismic faulting (e.g., Johnson et al., 2002;Griffith et al., 2009;Shaw, 2011;Wesnousky, 2008).During an earthquake, the direct destruction resulting from surface rupturing and fault dislocation exhibits a distinct spatial control over severe seismic disasters (Xu et al., 2016;Nurminen et al., 2022;Baize et al., 2019). Moreover, the length and coseismic displacement of surface ruptures are vital parameters for assessing the seismic hazard of active faults. Previous researchers have established empirical relationships between earthquake magnitude and rupture geometric parameters (e.g., Bonilla, 1988;Wells andCoppersmith, 1993, 1994;Biasi and Weldon, 2006;Petersen et al., 2011), which require robust regional datasets for calibration. Numerous seismic events indicate that earthquakes with M 6.5 are likely to generate surface ruptures. Weldon and Biasi (2015) suggested a probability of 50-60% for Mw 6.5 and 80% for Mw 7.0 events, respectively.Mainland China is one of the regions with the most frequent continental earthquakes globally, experiencing 35% of the world's continental earthquakes of magnitude 7 or greater during the 20th century (Zhang et al., 2003;Zheng et al., 2022). Since the 1980s, post-earthquake geological surveys have accumulated a wealth of high-quality surface rupture data in mainland China. This accumulation has been significantly accelerated by modern high-resolution topographic technologies like Unmanned Aerial Vehicle (UAV) photogrammetry, Light Detection and Ranging (LiDAR), and high-resolution optical satellite imagery (e.g., Oskin et al., 2012;Middleton et al., 2016;Pan et al., 2022;Liu et al., 2022;Bi et al., 2020;Xiong et al., 2022). However, these data are mostly scattered across various publications with inconsistent formats and contents, which to some extent restricts innovative big data research. Nurminen et al. (2022) and Baize et al. (2019) established the first global surface rupture database with a unified structure for 50 earthquakes, but it only included one case from mainland China: the 2008 Wenchuan Ms 8.0 earthquake. Moreover, recent studies on complex continental earthquakes, such as the 2023 Mw 7.8 and Mw 7.6 Kahramanmaraş earthquake sequence in Turkey (Meng et al., 2024), the 2024 Mw 7.0 Wushi earthquake in Xinjiang, China (Li et al., 2026), and the 2025 Mw 7.7Mandalay earthquake in Myanmar (Xu et al., 2025), have further highlighted the critical global need for high-resolution, standardized datasets to investigate multi-fault ruptures. Therefore, establishing a unified and comprehensive earthquake surface rupture database for mainland China is imperative to facilitate broader big data research and seismic hazard mitigation.China possesses a nearly 4000-year-long earthquake record, with the earliest documented event dating back to 2300 B.C. The Catalogue of Chinese Historical Strong Earthquakes (2300 B.C. to A.D. 1911), the Catalogue of Chinese Present Earthquakes (A.D. 1912to 1990) (e.g., State Seismological Bureau, 1995;China Earthquake Administration, 1999), and the online catalog by the China Earthquake Networks Center (CENC) (A.D. 1970 to 2025) were utilized in this study. Surface rupture data were compiled from published peer-reviewed scientific articles (both Chinese and English literature), Ph.D. theses, and academic books. A total of 632 articles were reviewed.The inclusion criteria for the database were: (1) events occurring within mainland China; (2) events with definitively identified and reported coseismic surface ruptures; and (3) availability of measurable geometric or kinematic parameters. The final dataset encompasses 72 historical and instrumental earthquakes (comprising 52 historical events prior to A.D. 1970 and 20 instrumental events up to 2025)..To ensure the accuracy, consistency, and usability of the collected data, the construction of the MCSRD followed a rigorous two-step workflow: Data Standardization and Data Visualization (Figure 1).Step 1: Data Standardization. The core challenge was integrating heterogeneous multi-source data. We first evaluated whether the collected literature contained original vector data of the rupture zones. If vector files were available, they were directly imported. If not, the geological maps and rupture distribution figures in the publications were geo-referenced and manually digitized into ESRI shapefiles using ArcGIS. Once the spatial polylines were established, we systematically extracted key quantitative parameters, primarily checking the availability of the Surface Rupture Length (SRL) and Maximum Coseismic Displacement (MD). All spatial data in the database are uniformly projected and stored using the China Geodetic Coordinate System 2000 (CGCS 2000) coordinate reference system.Step 2: Data Visualization. To maximize practical application, a Web-GIS solution was developed. The finalized geospatial database is hosted via ArcGIS Server, allowing users to intuitively browse, query, and interact with the spatial distribution and attribute details. The MCSRD contains two primary spatial datasets: a point shapefile providing the geographic location and basic parameters of each source earthquake, and a line shapefile providing the geometry and attribute information of the surface ruptures. Because the source publications exhibit large variability in detail and data format, we systematically reviewed the full text of each source publication to properly transfer the relevant parameters into the unified database.The attribute information of the source earthquake point shapefile is comprehensively structured as follows:IdE: Concatenation of earthquake date numbers (yyyy/mm/dd) based on local Beijing Time (BJT).NameE: Usual name of the earthquake, typically including the date of occurrence and location (e.g., 2008 Wenchuan earthquake).The designated name of the fault that ruptured (e.g., Haiyuan Fault, Xiaojiang Fault).Magnitude: Earthquake magnitude (from CENC catalog).Depth: Earthquake depth (from CENC catalog) (m). (Note: If the depth data is unavailable, the value is recorded as "-999" to indicate NoData).The attribute information of the surface rupture polyline shapefile is comprehensively structured as follows:IdE, NameF, Magnitude, and NameE are consistent with the source earthquake point shapefile.TLength: The total measured length of the surface rupture segment in kilometers (km).The maximum coseismic vertical displacement (throw) measured along the rupture segment, recorded in meters (m).The maximum coseismic horizontal (strike-slip) displacement measured along the rupture segment, recorded in meters (m). (Note: For MaxVOffset and MaxHOffset, if specific measurements are unavailable in the literature, the value is recorded as "-999" to indicate NoData).The dominant kinematic nature or faulting mechanism of the surface rupture (e.g., normal fault (N), reverse fault (R), sinistral strike-slip fault (SS), and dextral strike-slip fault (DS)).A data quality classification level (ranging from A to C) assigned to the rupture segment, indicating the confidence in the mapped geometry based on the investigation methods (detailed in Section 2.4).The primary literature sources, authors, and publication years from which the rupture geometry and displacement data were extracted.Because data availability and investigation methods vary across different historical periods, the reliability of individual surface ruptures differs significantly. We established a grading standard based on the investigation methods used.Level A reliability, the highest level, is assigned to cases where direct observations and measurements of the entire surface rupture were obtained immediately after the earthquake using one or more highprecision geological and geophysical techniques. This includes systematic post-earthquake Unmanned Aerial Vehicle (UAV) photogrammetry or high-resolution remote sensing interpretation covering the full extent of the rupture segment.Level B reliability refers to surface rupture data derived from geological surveys and geomorphological measurements (e.g., UAV, high-resolution remote sensing, or LiDAR) of historical earthquakes, provided that the investigation covers the entire rupture segment.Level C reliability is assigned to data derived from geological surveys and measurements of historical earthquakes where only partial segments of the rupture were investigated or well-preserved.Surface ruptures compiled in the MCSRD are predominantly distributed along the margins of major active tectonic blocks (Figure 2). The descriptive attributes within the database reveal significant regional variations in kinematic types:The database contains records for 9 historical destructive earthquakes along the rift basins surrounding this block. The recorded features are dominated by normal fault ruptures. For instance, the 1739 Pingluo M 8.0 earthquake and the 1673 Tianzhen M 7.0 earthquake manifest as linear normal fault scarps and stepped composite scarps. Of the 72 earthquakes documented in the database, 24 are strike-slip faulting events, 11 are normal faulting events, 11 are reverse faulting events, and the remainder exhibit complex dip-slip or oblique kinematics. Statistical extraction from the database reveals a strong correlation between the kinematic type (recorded in the Features field) and the maximum rupture length. Strike-slip faults consistently exhibit the maximum recorded surface rupture lengths. The longest continuous record in the database is the 2001 Kunlunshan Ms 8.1 earthquake, which produced a 426-km-long left-lateral strike-slip rupture. Other notable long strike-slip ruptures include the 1920 Haiyuan Ms 8.5 earthquake (240 km) and the 1997 Manyi Mw 7.5 earthquake (200 km). In contrast, typical pure normal and reverse fault ruptures tend to be shorter, mostly confined to lengths between 30 km and 100 km, bounded by the physical dimensions of the specific rift basins or mountain fronts they control.The database structure specifically separates maximum vertical displacement (MaxVOffset) from maximum horizontal displacement (MaxHOffset), allowing for a detailed kinematic analysis. The displacement statistics demonstrate clear tectonic dependencies.The largest horizontal displacements are predominantly associated with major strike-slip events on block boundaries. For instance, the database records maximum horizontal offsets of ~8.0 m for the 1920 Haiyuan earthquake, 8.4 m for the 1833 Songming earthquake, and 7.6 m for the 2001 Kunlunshan earthquake.The maximum vertical surface displacements in the database are predominantly associated with large-scale continental reverse faulting or major normal faulting. The most extreme vertical deformation recorded is from the 2008 Wenchuan Ms 8.0 earthquake, where the main Beichuan-Yingxiu thrust segment exhibited a MaxVOffset of 9-10 m alongside a MaxHOffset of 4.9 m. Similarly, historical normal faulting events, such as the 1556 Huaxian M 8.0 earthquake, also recorded exceptional vertical offsets reaching up to 9.0 m. These extracted parameters confirm that the MCSRD accurately captures the diverse faulting mechanisms of mainland China, providing a reliable quantitative basis for establishing regional faultscaling relationships.An explicit requirement of a Data Report is to elucidate how the scientific community can utilize the presented dataset. The standardized parameters within the MCSRD offer a robust foundation for the following geoscientific applications.Empirical scaling relationships between earthquake magnitude, surface rupture length, and maximum displacement are the cornerstone of Probabilistic Seismic Hazard Assessment (PSHA) and fault displacement hazard analysis (FDHA). However, historically, these global regressions have been heavily weighted toward data from western North America and specific highly active plate boundaries. Readers can extract the Length, MaxVOffset, and MaxHOffset attributes to calibrate and refine these scaling laws specifically for the tectonic framework of eastern Eurasia. Preliminary observations from our dataset indicate that large strike-slip faults in the Tibetan Plateau (e.g., the Xianshuihe, East Kunlun, and Haiyuan faults) may exhibit distinct length-to-displacement ratios compared to global averages, likely reflecting their role as primary boundaries accommodating the lateral extrusion of lithospheric blocks. By making this region-specific data publicly available, the MCSRD will significantly reduce epistemic uncertainties in seismic hazard modeling. This is particularly crucial for the engineering design and risk mitigation of major infrastructure projects (e.g., trans-continental railways, pipelines, and hydropower stations) traversing the active fault zones of western China.A critical insight derived from the MCSRD is the pronounced structural complexity of continental surface ruptures. Although the database records 72 distinct source earthquakes, it contains over 2500 individual rupture segments. This discrepancy quantitatively highlights that large continental earthquakes rarely rupture as a single, simple linear trace. By explicitly archiving complex branching, conjugate faults, and step-over geometries, the MCSRD challenges the traditional "single fault segment" paradigm. Researchers can reuse these spatial polylines as essential observational constraints for dynamic rupture simulations and Maximum Credible Earthquake (MCE) evaluations.Investigation of Triggering Thresholds: The compilation of the MCSRD provides a unique opportunity to investigate the behavior of continental intraplate and diffuse plate-boundary earthquakes, which often differ significantly from those on mature, simple plate boundaries (e.g., the San Andreas Fault or the North Anatolian Fault). The database documents significant variability in surface rupture triggering thresholds in continental intraplate settings. For instance, the database includes the 2024 Wushi Mw 5.7 aftershock, which generated an unexpected 4.7-km surface rupture.Users can correlate these rupture dimensions with local crustal rheology and strain rates to investigate rupture propagation efficiencies.The complete datasets presented in this Data Report have been deposited in a public repository and are strictly version-controlled.Name of the dataset: Mainland China Surface Rupture Database (MCSRD).Link for confidential peer-review: https://doi.org/10.5281/zenodo.19198199.(Note: As per journal policy, the link providing the representative sample dataset will be updated to the fully public open-access link containing the entire database upon the acceptance and publication of this Data Report).Interactive Web-map: To facilitate immediate visualization, the spatial distribution of the ruptures can be intuitively browsed via https://www.activefault-datacenter.cn/map.Recognizing that fault investigation is a continuously advancing field, the MCSRD is designed as a dynamic, "living" database. To ensure its long-term utility and accuracy for the geoscience community, we have established the following ongoing maintenance and updating mechanisms.As new surface-rupturing earthquakes occur in mainland China (typically M ≥ 6.5), our team will systematically collect post-earthquake field survey data, UAV/LiDAR measurements, and InSAR interpretations to digitize the new rupture traces and append them to the database.With the rapid development of high-resolution topography (e.g., airborne LiDAR) and paleoseismology, the geometric and kinematic parameters of historical ruptures are constantly being refined. We will periodically review newly published literature to update existing polylines, adjust maximum displacement values (MaxVOffset, MaxHOffset), and upgrade the Reliability levels of specific segments when higher-quality data become available.The database will be maintained using a strict version control system on the public repository. Major updates will be released as new versions (e.g., v2.0, v3.0) with accompanying release notes detailing the modifications. Furthermore, we encourage the broader seismological and geological communities to provide feedback, point out discrepancies, or contribute their own newly mapped rupture data to further enrich the MCSRD.