Two multi-temporal datasets to track debris flow after the 2008 Wenchuan earthquake

We provide two datasets for tracking the debris flow induced by the 2008 Wenchuan Mw 7.9 earthquake on a section of the Longmen mountains on the eastern side of the Tibetan plateau (Sichuan, China). The database was obtained through a literature review and field survey reports in the epicenter area, combined with high-resolution remote sensing image and extensive data collection and processing. The first dataset covers an area of 892 km2, including debris flows from 2008 to 2020 (an updated version). 186 debris flows affecting 79 watersheds were identified. 89 rainfall stations were collected to determine the rainfall events for the post-earthquake debris flow outbreak. The second database is a list of mitigation measures for post-earthquake debris flows, including catchment name, check dam number, coordinates, construction time, and successful mitigation date. The datasets can aid different applications, including the early warning and engineering prevention of post-earthquake debris flow, as well as provide valuable data support for research in related disciplines. Measurement(s) debris flow • water-based rainfall • check dam Technology Type(s) satellite imaging • weather station Sample Characteristic - Organism earthquake induced debris flows and mitigation measures Sample Characteristic - Environment earthquake Sample Characteristic - Location Wenchuan County Measurement(s) debris flow • water-based rainfall • check dam Technology Type(s) satellite imaging • weather station Sample Characteristic - Organism earthquake induced debris flows and mitigation measures Sample Characteristic - Environment earthquake Sample Characteristic - Location Wenchuan County

especially source-to-barrier distance, on debris flows velocity and volume using smoothed particle hydrodynamics (SPH). Dai, et al. 49 analyzed the impact force of debris flow on the check dam after the Wenchuan earthquake by numerical simulation method. However, the monitoring and control measures of debris flow require a lot of time and effort because few data and records of post-earthquake debris flows can be freely available 41 .
The debris flow in Wenchuan area was active before the 2008 earthquake 50 . Many debris flows occurred in the Longmen Mountains area after the 2008 Wenchuan earthquake 13 (Fig. 1a). By the end of 2010, more than 440 debris flows happened in the earthquake-stricken area 6,51 . Such as the "9.24" catastrophic debris flow event in 2008, the "August 13" event in 2010, the "July 10" catastrophe event in 2013 30,[52][53][54] , and the "August 20" debris flow event in 2019 [55][56][57] . Following the Wenchuan earthquake, many researchers are working on the mechanism, prediction, and early warning of post-earthquake debris flow 6,27,30,31,35,41,43,51,[56][57][58][59] . How to prevent and control the post-earthquake debris flow has become a prominent and urgent research topic 22 . Targeted reconstruction work has been carried out in the earthquake area, effectively preventing debris flow in some areas and ensuring the safety of residents' lives and property 60 . According to the field investigation results and literature, most of these catchments remain natural without disposal. Successful prevention and control cases are important to learn from and summarize 46 . These debris flow events and mitigation measures provide valuable data for studying post-earthquake debris flow early warning and mitigation measures 61 .
However, existing public data has many time gaps that need to be filled 41,62 , and there is still a lack of accessible public data on debris flows after the Wenchuan earthquake and no multi-temporal disaster mitigation measures datasets that are freely available. In this study, we focus on the watershed from Yingxiu Town to Wenchuan County, along the bank of the Min River after the 2008 Wenchuan earthquake. Two datasets were supplied that track debris flows events. Through data collection, interpretation of high-resolution remote sensing images, and field investigation, 186 debris flows were identified in the first dataset that affected 79 catchments. The total area of these catchments is approximately 892 km 2 (Fig. 1). 89 rainfall stations were collected, covering an epicenter area of 1566 km 2 . The second dataset contains a list of debris flow mitigation measures from 2008 to 2020, including the catchment name, dam number, construction period, and coordinates. The flow chart of the database construction is shown in Fig. 1b. Our datasets are freely available at https://doi.org/10.5281/ zenodo.6891244 63 . We also encourage other scholars to share their relevant data, which can help improve the current dataset to facilitate post-earthquake debris flow research.

Methods
The study area is along the Minjiang River from Yingxiu Town to Wenchuan County and the Longchi and Yuzixi river basins, including 79 watersheds with an area of 892 km 2 (Fig. 1). Combining literature review, field investigation reports, and remote sensing images, we constructed two datasets of debris flow events and mitigation measures after the 2008 Wenchuan earthquake. The first dataset covers an area of 892 km 2 , including debris flows from 2008 to 2020. 186 debris flows affecting 79 watersheds were identified. 89 rainfall stations were collected to determine the rainfall events for the post-earthquake debris flow outbreak. The second database is a list of mitigation measures for post-earthquake debris flows, including catchment name, check dam number, coordinates, construction time, and successful mitigation date. The detailed database build flow chart is shown in Fig. 1b.
According to the recorded debris flow events, we went to the field to investigate and interview residents every year after the rainstorm. GPS and laser range finders were used to measure the coordinates, and debris flows fan thickness. According to the field survey reports, the volume of the material washed out by the debris flow is calculated. The hourly rainfall data was from The Meteorological Administration of China and the Meteorological Bureau of Sichuan Province, recorded by an automatic rain gauge. The data obtained from the rainfall station are usually time-continuous series, which cannot be directly used for analysis. Therefore, it is necessary to pre-process the data to extract the rainfall events. In this study, a standard proposed by Zhou and Tang 64 was used to divide the rain events. This standard regards the hourly rainfall of >1 mm as the beginning of rainfall and <1 mm for 6 consecutive hours as the end of the rainfall 64 (Fig. 2).
As for the triggering rainfall of each debris flow event, the coordinates of rain gauges were used to calculate the closest distance to the debris flow event catchment. To provide more data for researchers and consider the distribution density of rainfall data we collected, the rain gauges within 8 km (the mean mainstream length) of the debris flow were selected 41 . Most debris flow events in the first dataset were collected from literature review and survey reports (see the file named "data references" in the debris flow repository). For the bigger and most catastrophic events, we conducted field investigations and interviews with the residents. Remote sensing imagery validated debris flow events (Table 1).
Dataset of multi-temporal debris flow events after the wenchuan earthquake. The first dataset is about the recorded debris flow events after the 2008 Wenchuan earthquake. Figure 3 shows the detailed number of these debris flow events from 2008-2020. The structure of the dataset is summarised in Table 2. The first dataset contains information about recorded debris flows (DF_DATA) events and their triggering rainfalls (DF_R_DATA). The debris flow (DF_DATA) events are stored in the ".xlsx" file format in the dataset. It includes information such as DF_ID, gully name, coordinates, date and time, deposition volumes, and data references. Rain gauge data includes the rain gauge ID (RG_ID), coordinates, altitude, amount of rain, temporal resolution, units, and other information. The debris flow triggering rainfall data (DF_R_DATA) are stored in ".txt" file format and includes DF_ID, RG_ID, debris flow date and time, rainfall, and units. Not only the catastrophic debris flows  www.nature.com/scientificdata www.nature.com/scientificdata/ from 2008-2020 7,27,31,43,53,56,57 , but also some small events are included in this dataset. Some small debris flows usually lack records in the previous literature because of their long distance, or do not cause serious damage to the downstream residential areas or buildings.

Triggering rainfalls of multi-temporal debris flows. Our first dataset contains 89 rainfall stations from
Wenchuan country to Yingxiu Town, covering an area of 1566 km 2 , including the catchments along the Minjiang River, Longchi, and Yuzixi River, with the period from 2008 to 2020 (Fig. 4). Table 2 shows the structure of the first dataset.
The name of the saved folder corresponds to the ID number of debris flow events (X) in the dataset, in a format like RG_FD_ID_X. All the segmented rainfall data is in ".txt" file format and stored in folders numbered by debris flow events. The ".txt" file in the folder name format is AB_CD_EF, where "AB" is the number of calculation iterations (sorting function; The smaller the "AB" value under the same directory, the closer the rainfall station is to the debris flow catchment area. "CD" is the corresponding rainfall station number, and "EF" is the date and time of the debris flow outbreak. Figure 5 shows the rainfall events and the initiation time of the debris flow. We chose to provide rainfall data for a time window starting from 7 days before and ending 1 day after the outbreak of the debris flow event. This facilitates the subsequent research on the triggering conditions of debris flow caused by early rainfall. Readers can get more detailed rainfall data from the author if needed. Mitigation measures of debris flow after the wenchuan earthquake. The catchment ID (C_ID), catchment name, coordinates, the number of check dams, construction time, and date of successful debris flow mitigation were included in the database ( Table 3). The mitigation measures dataset is available in csv and shapefile formats in 79 watersheds after the earthquake in the study area. Figure 6 shows the successful case of mitigation measures and debris flow prevention and control in the study area. The results showed that effective mitigation measures were taken in 17 of the 79 watersheds in the study area during the subsequent rainstorm.
It can be seen that some mitigation measures have blocked the movement of debris flow through high-resolution remote sensing images (Fig. 6c). Figure 6(a,b) shows the mitigation measure in the Qipan catchment after the 2013 rainstorm.   47,65 . They are usually built in densely populated areas and catchments connecting river channels. Since the debris flows move quickly and can carry boulders, the impact pressure on the inspection dam is extremely high. For example, the estimated peak impact pressure of the Wenjia catchment debris flow is about 2.4 MPa 47,49 , The loose deposits are transported to the channel gradually with time, and the back of the dam body is gradually filled with debris, which will reduce the Fig. 3 The catastrophic debris flows in the study area and the corresponding daily rainfall. www.nature.com/scientificdata www.nature.com/scientificdata/ effectiveness of the check dam 47 (Fig. 6c,d). Therefore, it is necessary to dredge the debris behind these check dams in time before the rainy season to prevent debris flows and dam breaking.

Technical Validation
Due to the lack of adequate monitoring equipment in the study area, the initiation time of debris flows is usually difficult to measure 41,66 . In addition to catastrophic debris flows that have been well studied and reported, there are still much small-scale debris flows in the study area that may not have been confirmed 36,41 . Therefore, the actual number of debris flows in the study area is much higher than the recorded number in our dataset. Rainfall stations are selected based on the distance between the rain gauge and the catchment, and due to differences in rainfall spatial distribution, this can make the selected data very different from the real data in debris flow initiation. The rainfall monitoring equipment of the government meteorological bureau is generally set near the  www.nature.com/scientificdata www.nature.com/scientificdata/ ditch mouth or villages with low altitudes, but it is rarely installed near the source area. Therefore, when debris flows initiation, observed rainfall may be significantly less than actual rainfall. Due to the limitation of image resolution and time, there are errors in determining the time of disaster interpretation and mitigation measures.

Usage Notes
We presented a multi-temporal debris flow and triggering rainfall in the Wenchuan earthquake-affected area from 2008 to 2020 and a list of debris flow mitigation measures from 2008 to 2019. The two available datasets can be used to investigate the temporal patterns of accelerated mass wasting produced by a strong earthquake and assess the effectiveness of debris flow prevention methods.

Code availability
There is no custom code produced during the collection and validation of this dataset.

author contributions
Lei Wang designed the work and methodology, manuscript, and figures; Ming Chang drafted the review, editing, and supervision; Jian Le actively took part in the fieldwork and data collection; Lanlan Xiang and Zhang Ni performed the mapping of multi-temporal remote sensing images; All authors revised and approved the datasets and the manuscript files.