Sichuan spans China's first and second-level landform steps. With mountains and hills widely distributed, frequent seismic activities, complex and changeable climate, it is the province with the largest number of geological disasters in China.
In order to strengthen the monitoring and early warning of natural disasters and protect local people’s heath, the Sichuan Provincial Institute of Land Space Ecological Restoration and Geological Disaster Prevention and Control adopted technologies of micro-service architecture and distributed storage and analysis, and developed the geological disaster and meteorological risk early warning system (hereinafter referred to as "the system").
According to different geological backgrounds and environmental conditions, the system divides the province's early warning range into three major regions and six subregions, and builds an early warning model for imminent geological disaster and meteorological risks in the province. By combining the data of actual rainfall and forecast rainfall provided by the Provincial Meteorological Bureau, it can also dynamically generate geological disaster and meteorological risk warning and forecast information. By sending the information to the person in charge in a timely and accurate manner through the Geological Disaster Intelligent Prevention APP and early warning text messages, it provides contingency plans for on-site disposal and saves valuable time for transferring the masses.
Refining the early warning area to improve the warning accuracy
At present, due to factors such as errors in equipment monitoring source data, rough division of early warning areas, unreasonable selection of early warning models, and unscientific setting of early warning thresholds, the accuracy of early warning of land disasters is generally low. In order to improve the accuracy, the system verified the rainfall data of meteorological stations in Sichuan Province, and eliminated the rainfall data of obviously abnormal stations; the early warning area was refined to the level of administrative villages, and administrative villages with the same geological environment are divided into the same warning area, which reduced the influence of different geological environments in the same early warning area on the early warning results.
In addition, according to the historical rainfall distribution in the early warning area, the user can select the data of real rainfall and forecast rainfall of the meteorological station in different periods and input into models for calculation. On the basis of the early warning thresholds in the three major regions and six sub-regions, the thresholds are set considering the actual local rainfall conditions to further improve the accuracy of early warning and forecasting of land disasters.
Improving the model computing to ensure timely warning
In the battle of geological disaster prevention and control, the early warning of geological disasters plays the role of a radar sentinel. It is not only necessary to identify accurately, but also to warn quickly. In order to improve the calculation efficiency of the early warning model, the calculation data of the system is stored in the MongoDB database, the data is processed through Python, and distributed computing is performed. Combining multi-thread computing and multi-cluster deployment, the calculation time of the early warning model of the system is reduced from the initial 30 minutes to about 7 minutes, which ensures that the warning results can be sent in a timely manner and buys more time for persons in charge to implement disaster prevention plans.
Classified early warning for dual control of geological disasters
In view of the fact that heavy rains in Sichuan Province can easily lead to geological disasters such as landslides and mudslides, the whole range of the province is divided into extremely high, high, medium, and low four-level risk areas based on the susceptibility and danger of the area, and the extremely high and high-risk areas are also included in the monitoring, early warning and protection system to achieve dual control of geological disasters.
In the prevention and control of geological disasters, it is necessary not only to consider the consistency of regions of different risk levels, but also to customize the early warning and protection measures of geological disasters based on the actual situation of the region. The system is divided according to the scope and responsibilities of users of four levels: province, city, county, and grassroots. Users at certain level can set and send imminent warning SMS content to persons in charge at this level and lower levels, playing the role of prompting and supervising.
Coordination between PC and mobile Apps to form a closed-loop management
The system can generate imminent warning results and send them to the PC and mobile APPs. Through user rights management, users of different levels can only view the warning information within the scope of their own responsibility. When the early warning is triggered, the system will compare the warning area and level with that of last time, and send early warning text messages every 6 hours, informing the corresponding person in charge of the early warning information and pre-planned measures.
In the system, users at city and county level in Sichuan Province are mainly responsible for the overall management, supervision, and command and dispatch of disaster protection. Grassroots full-time monitors will strengthen inspection after receiving early warning text messages, and adopt corresponding emergency plans according to the early warning level. When the early warning message is sent, if the grassroots full-time monitors have not read the early warning message within 10 minutes, the system will automatically voice dial and play the full content of the early warning message to ensure response from frontline personnel, which will form a closed-loop management mechanism from early warning information generation, transmission, reception to early warning response.
Results display in GIFs
Based on the geological environment and meteorological station rainfall data, the system dynamically calculates and generates rainfall and early warning information at the back end. Through SuperMap GIS platform, data such as basic data, rainfall distribution, early warning areas, hidden danger points and risk areas in the early warning areas are loaded onto the map for visualization rendering, which will realize functions such as layer management and map-attribute linkage. Through the visual map, users can quickly learn about the distribution of rainfall, the distribution area and level of the early warning area, the distribution of hidden danger points and risk areas in the early warning area, and quickly locate and query the basic attribute information of relevant hidden danger points and risk areas.
The system also answers to the query about the actual rainfall of meteorological stations in Sichuan Province and the query about the 24-hour real-time rainfall trend of the station. Through the dynamic display of the 24-hour real-time rainfall distribution data of the whole province, users can quickly grasp the changes in rainfall distribution and heavy rainfall in the whole province in the past 24 hours and the change trajectory of heavy rainfall. This will provide comparative analysis data for the verification of the threshold value of the early warning model, gradually improving the threshold value of the model, and continuously improving the accuracy of early warning.
