Study on the accuracy of forest fire prediction by satellite hot spots in Daxing'an Mountains of Heilongjiang, northeastern China

Satellite monitoring is a major forest fire monitoring method at present in China, its essence is measuring and reporting the high-temperature heat source of ground. However, the coarse resolution remote sensing data can't distinguish the nature of the heat source, the fire locating is not precise, and it always make the excessive or underrated report, this can make difficult in the work of hot spot data verification. To enhance the technical level of hot spot data disposal is the research topic of forest fire protection industry. Through the data of Daxing'an Mountains satellite hot spot data and forest fire historical data from 2005 to 2015, making use of ARCGIS and statistical analysis, the conclusions were drawn as the followings: 1) Accuracy rate of satellite forest fire forecasting was 79.7%, first forecasting heat resource of varied satellites are different, the accuracy rate of NOAA-16, NOAA-18, NOAA-19 and FY-1D was all about 80%, monitoring frequency reached 1 928 times, accounted for 2/3 of the total, and they are the main satellite in forest fire monitoring. 2) The mean position error of the prediction of hot spot and the actual fire was 5 787.9 m, first forecasting heat resource of varied satellites was different, the error of NOAA-19 and Terra predicting hot spots was less, but the monitoring number was small. Although the monitoring number of NOAA-17 and NOAA-15 was more, yet the error was big, the error of NOAA-15 was 7 064.7 m, and that of NOAA-17 was more than 10 000 m. 3)Among 219 forest fires recorded from satellite and 165 fires in historical record, 147 forest fires can be recorded by satellite and historical record in the same time. The problems of multiple reports and few reports mainly concentrated in Huma County. When the satellites monitor the same forest fire, the phenomenon of overstating and missing is serious. To give advice, in the case of having no possibility to improve the performance of satellite, satellite image processing technology should be improved, also it is necessary to perfect the standards of satellite image interpretation and to improve the accuracy of interpretation. At the same time, perfecting the ground verification feedback scheme is also necessary to increase the efficiency of verification and reduce the verification cost. Based on the differences among different satellites, in forest fire monitoring, we should mainly use FY satellites which are very accurate in predicting, and the stable NOAA satellites can be used as a supplement.