摘要
采用时间序列加法模型分析气温的长期趋势、周期波动及不规则变动.利用快速傅里叶变换对西安泾河站点日平均气温进行低频去噪,确定了气温全年变化的主频周期波动,并进行傅里叶函数拟合,精度98.64%;对气温的高频信息部分采用时间序列模型建立AR(2)模型,拟合优度54%,气温序列实现独立化分离过程.就全年气温变化而言,西安气候舒适,全年的平均气温为14.7℃,气温变化主频为365.5天,按时间呈余弦波动形态,振幅为±14.65℃,一年内最低气温日均值约为0.1℃,最高气温日均值约为29.3℃;全年最低气温95%的概率出现在-5.7℃,最高气温95%的概率出现在35.1℃;而可能出现的极端低温与高温分别为-9.6℃、39℃.短期的气温不规则变动主要影响其后五天的天气,一天的记忆程度为85%,逐渐衰减,至五天的记忆程度为9.44%,7天后可以认为记忆消失(2.59%).
The time series additive model is used to analyze the long-term trend, periodic fluctuation and irregular change of air temperature. Fast Fourier transform is used to denoise the daily average temperature of Jinghe station in Xi’an at low frequency. The main frequency periodic fluctuation of the annual variation of temperature is determined and fitted with Fourier function with an accuracy of 98.64%. Time series model is used to build AR(2) model for high frequency information of temperature, and the goodness of fit is 54%. The independent separation process of temperature series is realized. As far as the annual temperature change is concerned, the climate in Xi’an is comfortable, with an average temperature of 14.7 ℃. According to the model, the main period of temperature change is 365.5 days, which is predicted to be cosine fluctuation in time. The amplitude is (+14.65 ℃). The average annual minimum temperature is about 0.1 ℃. The average annual maximum temperature is about 29.3 ℃. The lowest temperature 95% occurs at -5.7 ℃ with 95% probability;the highest temperature occurs at 35.1 ℃ with 95% probability.And the extreme minimum and maximum temperatures are -9.6 ℃ and 39 ℃. Short-term irregular changes in temperature mainly affect the weather in the following five days. The degree of memory in one day is 85%, and gradually decreases. The degree of memory in five days is 9.44%. It can be considered that memory disappears after seven days (2.59%).
作者
戴婷
汤雯茜
DAI Ting;TANG Wen-xi(College of Science, Hunan Institute of Engineering, Xiangtan 411104, China)
出处
《湖南工程学院学报（自然科学版）》
2019年第3期61-65,共5页
Journal of Hunan Institute of Engineering(Natural Science Edition)
