عنوان مقاله [English]
نویسندگان [English]چکیده [English]
For the past several years, the severity and frequency of extreme events like flooding or prolonged droughts seem to be on the rise, globally. Precipitation is the most vital element of climate which almost influences many aspects. The effect of precipitation on different human activities including farming, industry and services is obvious (directly, indirectly, or with intermediate). In this regard, the relation between precipitation and farming, especially dry farming, is considerable and somehow distinct from other activities. It is clear that the most natural kind of using precipitation at farming section is dry farming. Identification of rainfall events are important in designing the related water structures, farming, weather modifying, policy making and planning and also in monitoring climate change. Methodology identification of trend or persistence in the rainfall series is essential to present the hydrological information in a condensed form for decision making and planning in water resources of any region. Effective rainfall (ER) is defined as that portion of rainfall which is useful directly and/or indirectly for crop production at the site where it falls. Effective rainfall is influenced by the factors such as quantity and intensity of rainfall, evapotranspiration (ET) and deep percolation losses, and irrigation management practices. Estimating the effective rainfall is extremely useful for operational planning and management issues.
This paper exemplifies a study involving non-parametric statistical method of Mann-Kendall test for identification of trends in annual rainfall series in Iran. The nonparametric Kendall test was applied to find trends in a number of climatic and hydrologic variables. This test was selected because it can handle non-normality, censoring, or data reported as values "less than", missing values or seasonality and because it has a high asymptotic efficiency. Mann Kendall nonparametric test is widely used for the analysis of trends in meteorological and hydrological series. One of the advantages of this method is its applicability for a time series distribution, which does not follow a typical statistical distribution. This method of analysis adopts two parameters such as the Kendall statistic, S and the normalized test statistic Zs, which are used to determine the nature and level of the significance of trends exhibited by the variables. Generally, a positive value of S is an indication of an upward trend, while a negative value indicates a downward trend. Also, the value of Zs -greater than 1.96 at a selected confidence limit of 95%- shows that the trends can be interpreted as statistically significant or otherwise. The data must be serially independent in the case of the non-parametric tests. Based on the performed studies, the existence of serial correlation will increase the probability for significant trend detection. This leads to a disproportionate rejection of the null hypothesis of non-trend, whereas the null hypothesis is actually true. Therefore, the influence of serial correlation must be eliminated. In this regard, different methods such as pre-whitening, variance correction, and TFPW have been proposed.
The TFPW procedure presented here provides a better assessment of the significance of the trends for serially correlated data than the other approaches and several researchers have used this procedure. The trend changing of total and effective rainfall of the entire country of Iran on three scales (annual, monthly and seasonally) was analyzed using statistical tests. The data of 33 stations was used in the study area. In this analysis, the trend changes of total and effective precipitation in the period of 40 years (1971-2010) and (1961-2000) and a 50-year period (1961-2010) were studied. The slope of the linear trend of data was estimated by TSA and then, if there was a correlation between data, the autocorrelation coefficient between data removed using TFPW method, and the time series of precipitation pre-whited. Then, the trend channeling of total and effective rainfall time series was analyzed, using the Mann-Kendall test. The results showed that the median of effective rainfall in the selected stations were negative and therefore they had decreasing trend. Also, the median of total rainfall in the selected stations for a period of 40 years (1961-2000) was approximately equal to zero and for the 40-year period (1971-2010) the trend was significant at the level of 10%. While, for the period of 50 years (1961-2010) the median was negative, but not significant.