Study on Fog Synoptic Characteristics and Fog Forecast Method in Chongqing
重庆市区雾的天气特征分析及预报方法研究

Synoptic characteristics of fog and the vertical distribution characteristics of some meteorological elements such as temperature,humidity and so on,are analyzed in terms of 1951-2005 observed fog data in Chongqing.By means of selecting reasonable diagnostic factors,an artificial neural network model is established with dynamic learning rate BP algorithm to simulate the visibility of Chongqing.Results show that annual mean foggy days in Chongqing have an obvious descent tendency and the light foggy days are increasing sharply.This variation is likely to be mainly associated with the enhancement of urban heat island effect and air pollution.Generally in the mature phase of radiation fog,it is featured by stable inversion structure and extremely pronounced vertical variation of the temperature and relative humidity in the vicinity of the fog top.The BP neural network model is possessed of preferable adaptive learning and non-linear mapping abilities with 99% verification forecasting accuracy,wherein the forecasting accuracy of thick fog(visibility from 0 to 1 km) is 83%,the Ts grade is 69%,and the average forecast error is 0.384 km.The forecast ability of neural network to the fog(especially thick fog) is enhanced obviously due to the introduction of M-index,Richardson number,condensation nucleus,radiation condition and various physical parameters,as well as the technical processing to the network input values of some diagnostic factors in addition to conventional meteorological elements.Ts grade of thick fog with visibility lower than 0.4 km can reach 89.5% and model results can provide favorable reference to the fog forecast of Chongqing.