基于Logistic的大兴安岭雷击火预测模型

Lightning Fire Forecasting Model of Daxing'an Mountain Based on Logistic Model

利用大兴安岭雷击数据、雷击火数据、气象数据和GIS空间分析,引入加拿大雷击火预测模型和Logistic回归模型,建立了黑龙江大兴安岭雷击火预测模型。模型除了使用加拿大雷击火预测模型的固有参数外,还引入了影响火灾蔓延的地形因子作为模型参数。利用实际数据进行模型检验,结果表明:修正后的预测模型在特定区间的雷击火预测结果较为理想,总体预测正确率能达到30%以上,与加拿大预测正确率基本一致;也说明了基于logistic回归的加拿大雷击火预测模型,经过参数修正后同样适用于我国大兴安岭地区,可为该地区的雷击火预防提供可靠的数据支持。

With Canadian lightning fire forecasting model and logistic model, we established lightning fire forecasting model and checked its validation using lightning data, lightning fire data and meteorological data of Daxing＇ an Mountain. Existing model parameters of Canadian lightning fire prediction model were introduced with terrain factors influencing the spread of fire such as slope and aspect. Results drawn from actual data according to the model show that the revised model works in the specific area to predict lightning fire. The overall prediction accuracy rate reached 30%, which was consistent with that of Canadian forecasting model. The result shows that with parameter modification Canadian-based logistic regression model can equally be applied to Daxing＇ an Mountain, and it can provide real-time predictions of lightning fire, which provides reliable monitoring data to support the lightning fire prevention in Daxing＇ an Mountain.