Extended Abstract:



Rainfall time series analysis is important in that its variability is very high, and should be considered by water resources and agricultural management planners. In this study, the annual rainfall series analysis of three stations in Kerman, Sirjan and Rafsanjan as three major pistachio growing areas in Kerman province during the statistical period of 1986-2086 (35 years) has been investigated. First, the statistical parameters of precipitation data at three stations were examined, followed by the identification of precipitation behavior and norms. To evaluate the homogeneity of mean and variance, mean homogeneity test based on Standard normal homogeneity test and homogeneity of variance based on Van Neumann test were used. Investigation and detection of trends or no trends in the annual precipitation data of the studied stations using parametric methods (autocorrelation test, Pearson correlation coefficient and least squares error test) and non-parametric tests (Mann-Kendall statistical test, test The turning points and the correlation coefficient of Spearman and Kendall Tao) were investigated. The results showed that the precipitation data series in all three stations are heterogeneous in terms of mean and variance. Based on the parametric tests of Pearson correlation coefficient and the least squares error test in Rafsanjan station, there is a decreasing trend of precipitation in the study period. In Kerman and Sirjan stations, the slope of the precipitation line was 0.439 and 0.271, respectively, and the test of the least squares of error in these two stations showed that the slope of the line is not significant. Non-parametric tests of Kendall and turning point test also showed that there is no trend in the annual precipitation data of the studied stations and the null hypothesis test should be accepted.



Introduction:

Knowledge and understanding of the behavior of climatic elements (fluctuations, trends) of rain and temperature is very important because their changes and fluctuations can affect other climatic parameters. In recent decades, human beings have witnessed the development of statistical sciences and its expansion in scientific fields (Vakili, 2014). Identification of climate change, in general and precipitation in particular, through time series analysis of the relevant element is important. Time series is a set of observations that are arranged according to time or the order of its repetition (Pourmousi, 2014). The purpose of time series analysis is to describe, and predict the future values of a process (Chatfield, 1996). Time series are divided into two types of static and non-static series. A series is static when there is no regular change in the mean and variance and periodic changes are omitted. This study aims to identify precipitation behavior and time series analysis in three main stations of the province, including synoptic stations of Kerman, Sirjan and Rafsanjan during the statistical period of 1986-1986 (35 years) to identify variability,



Methodology:

In this study, the annual rainfall series analysis of three stations in Kerman, Sirjan and Rafsanjan as three major pistachio growing areas in Kerman province during the statistical period of 1986-1986 (35 years) has been investigated. First, the statistical parameters precipitation data were considered at three stations to determine the behavior and norm of precipitation. To evaluate the homogeneity of mean and variance, mean homogeneity test based on Alexander test and homogeneity of variance based on Van Neumann test were used. To detect the trend or no trend in the annual precipitation data of the studied stations, parametric methods (autocorrelation test, Pearson correlation coefficient and least squares error test), and non-parametric tests (Mann-Kendall statistical test, turning point test) And Spearman and Kendall correlation coefficients were used.



Discussion:

The threshold value for determining the homogeneity or heterogeneity of the time series according to the Van Neumann method with a statistical period of 30 to 40 years, is between 1.49 to 1.42. The values calculated using the above method are less than the desired threshold value. Therefore, it can be concluded that the null hypothesis (homogeneity of variances) is rejected. The data series of all three stations are heterogeneous in terms of variance. At the 95% significance level, the null hypothesis (mean homogeneity) is rejected and the series mean is not homogeneous. Autocorrelation diagram, turning point test was used to calculate the randomness (independence) of the data. The results of these two tests confirmed the data independence. Parametric (Pearson correlation coefficient) and non-parametric methods (Spearman and Kendall correlation coefficient) and Mann Kendall statistical test were used to reveal the trend and significance test of the trend. According to the results of Pearson, Spearman and Mann-Kendall correlation tests, the annual precipitation trend is decreasing only in Rafsanjan station and shows a decrease in the station precipitation behavior during the study period. In Kerman and Sirjan stations, considering that the value of P statistic is greater than the significance level of 0.05, the null hypothesis is accepted. There is no trend between the time and precipitation parameters in these two stations. In this study, regression was used to reveal the precipitation trend. The above test in Kerman and Sirjan stations showed that there is no significant trend in annual rainfall. According to this test, the annual rainfall data of Rafsanjan station has a decreasing trend.



Conclusion:

The results showed that the precipitation data series in all three stations are heterogeneous in terms of mean and variance. Based on the parametric tests of Pearson correlation coefficient and the least squares error test in Rafsanjan station, there is a decreasing trend of precipitation in the study period The slope of the precipitation line was obtained based on the least squares of error for Rafsanjan station -0.23 and revealed the trend of annual decrease in precipitation in this station. Non-parametric Kendall tests and test turning point test also showed that there is no trend in the annual precipitation data of the studied stations and the null hypothesis test should be accepted. Using the autocorrelation chart, it was shown that the observations were not significantly related to each other and were independent of each other

Keywords: time series, Homogeneity of observations, trend detection, Rainfall, Rainfall, Kerman province