基于BP神经网络拟合四川南部山区焚风过程气温变化的方法

Method of Fitting Temperature Changes during Foehn Process in Mountainous Areas of Southern Sichuan Based on BP Neural Network

针对发生在四川南部山区的焚风天气,利用BP神经网络良好的函数模拟能力,将过程中的气温、时间、经纬度和海拔高度等要素作为训练数据,构建BP神经网络模型并做误差检验。结果表明:建立的BP神经网络模型可以很好地模拟焚风过程中气温变化,可用于甄别焚风现象中气温升高是否属于异常或疑误,为气温质量控制提供了一种新思路;训练后的BP神经网络可应用于气温变化的插值分析,制作高分辨率实况网格产品,为气象预报、预测和服务业务提供数据支撑。

Using the good function simulation ability of BP neural network,the temperature,time,longitude,latitude,and altitude of the foehn process occurring in the mountainous areas of southern Sichuan are constructed as the suitable training dataset for BP neural network training.After error analysis of the trained BP neural network,the analysis results show that the established BP neural network model can well simulate the temperature changes during the foehn process,and can be used to identify whether the temperature rise in the foehn phenomenon is abnormal or suspect,providing a new idea for temperature quality control.The trained BP neural network can be applied to interpolate and analyze temperature changes,produce high-resolution live grid products,and provide data support for weather forecasting,forecasting,and service businesses.