Real-time Monitoring And Early Warning Of Grassland Fire Based On Geostationary Meteorological Satellite

Timely detection,early warning and accurate simulation of fire spread are very important to block and extinguish grassland fires.Mongolia is an important neighbor in the north of China.It shares a border of 4676.8km with China,and the terrain,climate and vegetation types on both sides of the border line are similar.In recent years,grassland fires have frequently occurred in Mongolia,which often spread to the border by the wind,and seriously threaten the ecological security of grassland in northern China and the safety of people's lives and property in border areas.Taking the border area of China and Mongolia as the research area,using the data of HIMAWARI8/9,a new generation of geostationary meteorological satellite with high time resolution,to carry out the work of rapid identification of fire point,real-time monitoring of the development of grassland fire,extraction of burned area,and to improve the existing fire spread cellular automata model,the meteorological,topographical,and combustible data are used to simulate the spreading trend of grassland fires.The results show that:(1)Based on HIMAWARI8/9 data,the research of grassland fire fast identification and real-time monitoring was carried out.The grassland fire that occurred in Mongolia on April 29 to 30,2019 was dynamically monitored.It was found that the grassland fire lasted for 11h30min and was monitored by HIMAWARI8/9 satellite for 70 times,with a total burned area of 3098.09km~2.The initial detection time of the fire on the HIMAWARI8/9 image was 14:50(UTC)on 29th,24 km from the border line,located at 115°49?E,49°2?N.Through real-time tracking and monitoring,it was found that the fire continued to spread to the southeast,until the fire head no longer spread outward at 02:20on the 30th,and gradually extinguished.During the spread of fire,the spreading speed showed an increasing-decreasing-increasing trend,and the average spreading speed was about 6.08 km/h,which was a rapid surface fire.The standard deviation ellipse analysis method was used to evaluate the accuracy of the fire point spatial distribution.The results showed that the center of gravity of the standard deviation ellipse of fire points extracted from MODIS?S-NPP/VIIRS and HIMAWARI 8/9 was very close,all around 49.03°N,115.96°E,and the distribution direction of the fire points was from northwest to southeast,consistent with the moving direction of fire line.(2)Compared with the four indexes of Normalized Difference Vegetation Index(NDVI),Enhanced Vegetation Index(EVI),Global Environment Monitoring Index(GEMI)and Burned Area Index(BAI),the Normalized Burn Ration(NBR)index had better separation ability for the burned area,and the separation index M was greater than 1.GEMI?BAI?NBR three vegetation indexes were used to extract burned areas from HIMAWARI8/9 satellite data and used Sentinel-2 satellite to evaluate the accuracy,it was found that the NBR index had a better ability to identify burned areas,with an overall accuracy of 96.25%and Kappa coefficient 0.91,and the potential of BAI and GEMI index to extract burned areas had little difference,with the overall accuracy of 84.86%and 83.96%,respectively,and Kappa coefficient is both 0.60.(3)On the basis of mastering the principle of cellular automata and Wang Zhengfei-Mao Xianmin forest fire spread model,the forest fire spreading model based on geographic cellular automata proposed by Zhang Feifei was improved,and a grassland fire spread simulation model was constructed.In the new model,the influence of cell combustion was introduced,the transformation rule function of cell was redefined,the simulation software was developed with Python language,and the fire point information extracted from HIMAWARI8/9and other data such as meteorology,topography and combustibles were input,and the fire spread simulation experiment was carried out.According to the observation frequency of HIMAHARI8/9 and the acquisition time of wind data,the grassland fire spread simulation was carried out in three ways,and the simulation accuracy was verified by the actual burned area.It was found that the simulated burned area was always larger than the actual burned area after the simulation was performed in various ways,among which the best results show that 92.24%of the actual burned area was successfully simulated by the model,and the model accuracy reached 73.29%.In this paper,the results show that the grassland fires can be extracted and tracked in real time by the high time resolution geostationary meteorological satellite.Based on this,combined with grassland fire spread models,the fire spread direction and speed can be accurately simulated.In the case of cross-border fire,it can provide the information of the location and time of fire crossing the border to the fire prevention management department,and provide effective information for the work of blocking,extinguishing and emergency management.