Modeling of Some of Effective Climatic Indices in Ardebil Synoptic Station For Natural Hazards Management

Changes in climate indices and their risky behavior in each geographical area could lead to the occurrence of some natural disasters in the region. Environmental crises always leave a lot of damage to human societies. Natural disasters have related with changes and behaviors of climate elements. By Modeling and forecasting of climatic factors, an important step will be taken in the management of natural disasters. Statistical methods such as time series have better performance to predict future climate indices. Ardebil city is always faced with natural disasters, especially climate crisis. Therefore, by the modeling of climate indices in this city we can perform some management schemes to reduce or eliminate the adverse effects of natural disasters.
In this study, some climatic indices in Ardabil synoptic station include average of maximum daily temperature, average of minimum daily temperature, speeds of gusty winds, the highest daily precipitation, heating degree days, dusty days, days with maximum temperature below zero and zero days with minimum temperature below zero and zero snowy days with small hail, thunderstorm days and days with visibility less than or equal 2km during 1976- 2005 were modeled using winters, autoregressive and ARIMA models respectively. For data modeling, the sequence of data was drawn. According to the non-static data, to perform modeling, a series difference method was used. for investigation of significance of residuals, Kolmogorov-Smirnov test was used. In addition, ACF and PACF graphs were plotted. In implementation of model that supplies a combination of autoregressive and moving average delays were considered. In choosing the best model among the models for time series prediction, prediction error rate was a useful framework for identifying model. There are several indices for the validation of the study, the indices of the coefficient of determination (R2), mean absolute error, mean absolute percentage error and root mean square error are used.
The results of this study suggest that amounts of 12 climatic variables in Ardebil synoptic station, have a specific model for future conditions. In reversible process for all the variables a specific model (2, 0, 0) has been used but the results of the model error rate for any of the studied variables were not approved. Average of maximum daily temperature and the highest daily precipitation variables also have (2, 4, 3) model. But average of minimum daily temperature, speeds of annual gusty winds and heating degree days have different models. Dusty days, days with maximum temperature below zero and zero days with minimum temperature below zero and zero snowy days with small hail and thunderstorm days prove (2,2,2) model. Based on the results, variable of average of maximum daily temperature, speeds of annual gusty winds, heating degree days, days with minimum temperature below zero and zero and days with visibility less than or equal 2km have positive coefficients. Variables of average of minimum daily temperature, dusty days, snowy days with small hail and thunderstorm days have negative coefficients.
Investigation of sequence diagrams of series of each variable showed more fluctuations in those variables. The Kolmogorov Smirnov test was performed for the remaining differential. The results of this test for all variables with 0.95 confidence, confirmed the assumption of normality of the model residuals. The results of the final model accompanied with an acceptable probability for all variables. Some of the parameters in the final model were similar which indicate a clear and consistent trend of changes of these variables. Some variables such as annual fastest wind direction and speed and heating degree day, have more uncertain and fluctuating trend during the statistical period. The final model results showed that number of days with dust, number of days with maximum temperature equal 00 and below, number of days with minimum temperature equal 00 and below, number of days with snow or sleet and number of days with thunder storm have (2, 2, 2) model and average of maximum daily temperature and greatest daily of precipitation follow (2, 4, 3) model.