Study On The Variation Of Frost In Changzhi Area In Recent40Years

This paper have conducted a preliminary study by using data of11county stations in Changzhi area, Shanxi province.We analyze the space distribution characteristics and climate of the first and latest frost date, duration of frost-free period from1971to2007, including11counties in Changzhi area; By haar wavelet method, we analyze abrupt change of climate about the average sequence of the first frost day, the latest frost day, duration of frost-free period,in Changzhi area. Conclusions are as follows:1. The date of the first and latest frost, the length of duration of frost-free period are close with geographical features and altitude. The isoline distribution of the first and latest frost show the type of two early two late, The isoline distribution of duration of frost-free period show the type of two long two short; the county station that the first frost day is earlier and the latest frost day is later are usually located in north and east of the mountains or the higher latitudes; the county station that the first frost day is later and the latest frost day is earlier are usually located in the area that the terrain features is low or basin, or in low latitude area.2. The date of the first and latest frost, the length of duration of frost-free period are close with the climate change. The stations that the first frost day is later (earlier) are basically the same with the station that the average autumn temperature is higher (lower). The stations that the latest frost day is earlier (later) are basically the same with the station that the average spring temperature is higher (lower). The years that duration of frost-free period is longer (shorter) are basically the same with the years that annual mean air temperature is higher (lower).3. Followed along with the atmospheric background of global warming, the first frost days of11stations in Changzhi are common later, the days of frost-free period is expanding since1997, the latest frost day is finished earlier since2001. The variation of diurnal range is dissimilarity in Changzhi area by station for years, diurnal range increases in higher elevations, diurnal range changes little or decreases. The years that annual minimum temperature is greater than the threshold are increased on11station since2001, he first frost day is going to later; at the same time, the probabilities of the abnormal latest frost is reduced.4. Haar wavelet analysis display:2002is the mutation years.In2002, the high mutation point of average temperature of autumn is corresponding to the point of the first frost date, the high mutation point of average temperature of spring is corresponding to the point of the first frost date, he high mutation point of annual mean air temperature is corresponding to the point of duration of frost-free period. This is consistent with the fact that the last frost date is earlier, the first frost date is later, and frost free period is longer since2002Average first frost date occurred in Changzhi area as the predicted value,based on74circulation index as a Predictor, stepwise regression forecast model of average first frost date in Changzhi area was built applying with the double filtering stepwise regression method;Using directly the first frost day and the last frost day departure sequence in Changzhi area from1971to2000station by station and5order auto regression method to establish auto regression equation by station; then using empirical mode decomposition method with5order autoregressive methods by station to establish auto regression equation by station for the first frost day and the last frost day departure sequence. We make the prediction for early frost, late frost date prediction from2001to2007by using three methods of prediction equations. Conclusions are as follows:1. Solar activity is forecast factors that impact on Changzhi city the first frost day sooner or later, has certain relation with the emergence of the first frost date time.2. Using5order autoregressive method, EMD combined with5order autoregressive prediction method to predict the first frost date anomaly, the last frost date anomaly,the root mean square error, relative error, the arithmetic average error is not the same in different counties in Changzhi. Mountains, hills, high elevation areas first frost date of the root mean square error and the mean error is too lager.The feature highlights:EMD combined with5order autoregressive method than using autoregressive error decreases a lot. Application of EMD method to predict the result is better than the direct prediction results.