Study Of The Correction Effect Of Ensemble Kalman Filter Algorithm On FARSITE Prediction Of Forest Fire Propagation

Forest fire is a global disaster,there are a certain number of forest fires breaking out every year in different places.These fires are often large in scale and have recently shown the potential to reach colossal dimensions which are very difficult to control.Forest fires will not only destroy the ecosystem,but also release carbon gas to cause large-scale environmental pollution,what's more,it may even threaten human life and property.Many countries put a lot of energy on forest fire management and fire-fighting every year.Computer simulation technology,an important way of forest fire prediction,can help us grasp the change of fire in advance and be well prepared for the crisis.However,the combustion mechanism of forest fire is complex,and its occurrence and development are uncertain.Forest fire behavior is affected by a variety of environmental parameters,such as fuel,terrain,weather,etc.The accuracy of these parameters will have a great impact on the prediction accuracy.In this paper,FARSITE is selected as the forest fire simulator,and the ensemble Kalman filter algorithm is used as the data assimilation method to carry out the dynamic data-driven simulation of forest fire propagation in a given fire case,which provides a new idea for the accurate prediction of forest fire.And the two parts of the main work in this paper are as follows.The first is the influence of wind,which is one of the main meteorological factors in forest fire propagation,on the correction effect of ensemble Kalman filter algorithm.To solve the problem of inaccurate fire perimeter prediction of FARSITE that arises from the input error of fire source position,we take fire perimeter as the variable to be corrected and assimilate the observation data of true perimeters by ensemble Kalman filter algorithm for dynamic correction.In this process,wind data is dynamically added to study its influence on the correction effect.The results show that the existence of wind will make the topology of the predicted fire perimeters more complicated.However,with wind or not,the algorithm perform well in reducing the prediction deviation and correcting the position and shape of the fire perimeters.In addition to the fire source,when there is also an error in wind data,the correction capability of ensemble Kalman filter will decline and the root mean square error curve changes from continuous decline to fluctuation.The correction results can not reflect the true state of the system ideally.The second is the influence of the parameters of the ensemble Kalman filter algorithm on the correction effect.Several key parameters in the algorithm,including the number of ensemble members,the standard deviation of observation data and the assimilation frequency,were studied respectively with the control variable method.By changing the values of these parameters,the differences of the results before and after assimilation are compared from the output the fire perimeters,the root mean square error of the state variable,and the distance index.From the relevance discussion based on the results of multi-dimensional analysis,the functionary mechanism of the algorithm parameters in the process of the assimilation is summarized,and the rule of the influence of parameter values on the correction effect is obtained.This study can provide a reference for parameter analysis and assimilation scheme design of ensemble-based assimilation.