Evaluation of changes in the frequency of convective precipitation days in the northwest of Iran using Mann-Kendall and Sen’s Slope method

Precipitation is one of the elements of climate that has great variability. These changes occur both in terms of time and space in many regions of Iran. Knowing the characteristics of this atmospheric element is very important for many construction plans, plans and agricultural activities. Any statistically significant change in this element can impose inevitable negative effects on natural resources, especially recoverable water reserves. Knowing the change process of climatic parameters is one of the things that has been the focus of atmospheric and hydrological science researchers in recent years. Most of the studies have been focused on investigating and analyzing the behavior of precipitation, and Mann-Kendall statistical methods have been widely used as the most common non-parametric method for analyzing the trend of precipitation time series. Since the trend finding in terms of the frequency of convective Precipitation has been less noticed by researchers, the purpose of the present study is to investigate the trend of the frequency of convective rainfall in the first half of the year in synoptic stations located in the northwest of Iran. In this research, the study area for the statistical analysis of convective precipitation includes Ardabil, East Azerbaijan, and West Azarbaijan provinces.

To carry out this research, the daily Precipitation data in the first half of the year (April, May, Jun, Jul, Aug, Sep), in synoptic stations located in the northwest of Iran (including Ardabil, Ahar, Urmia, Bonab, Bostan Abad, Tabriz, Pars Abad), Khoy, Sarab, Sardasht, Khalkhal, Marand, Mianeh, Mako, Mahabad, Meshkin Shahr and Maragheh were used during a 17-year period (2000-2016) and daily Precipitation of at least 5 mm was considered as convective Precipitation. Then, the annual frequency of convective precipitation of each station was counted and trending methods based on Mann-Kendall statistical method and Sen’s slope estimator nonparametric method were used to analyze its trend. The Mann-Kendall test does not require a normal frequency distribution or the linearity of the behavior of the data, and it works very strongly compared to the data that deviates from the linear behavior and is used to evaluate the trend. In this test, the null hypothesis (H0) and the opposite hypothesis (H1) are respectively equivalent to no trend and the presence of a trend in the time series of observational data. In Sen's method, the time series is divided into two groups, and the numbers of each time series are arranged in ascending order. Then these two series are plotted against each other on the coordinate axis. It is better to draw the first time series on the horizontal axis and the second series on the vertical axis. In the next step, the 1:1 line is drawn, if the points are above the line, there is an increasing trend, below the line, there is a decreasing trend, and on the line, the data will be without trend.

The changes of precipitation in Tabriz station have been decreasing since 2006 and because these two components outside the critical range (±1.96) have intersected each other in 2013, 2014, and 2015, this trend is not significant. This situation happened at Maragheh station in 2013 and 2014, and since 2015, a decreasing trend in rainy days has been seen, at Midane station, since 2015, there has been a change of direction in the abundance trend from a non-significant increase to 2015. In other stations of East Azarbaijan province, during the studied period, no specific mutation and trend in the frequency of convective precipitation can be seen, and in fact, these stations have no trend. In Sardasht station, the time of the beginning of decreasing and then increasing changes in the frequency of convective precipitation in the years 2000, 2002, and 2016, the two components Ui and Ui' are outside the critical range, which is not significant. A sudden change in the direction of this variable occurred in Mako station in 2001, which is incrementally insignificant. At Urmia station, several sudden changes in the frequency of convective precipitation can be seen during the studied period, which started from 2001 and continued until 2008. The trend of convective precipitation in 2003 was decreasing and after 2007 it was increasing insignificantly. In Mahabad station, the intersection of Ui and Ui components in 2001 reveals an increasing trend in the frequency of convective precipitation. In addition, in 2012, these lines intersected outside the critical range of ±1.96, which shows the increasing trend of the frequency of convective precipitation of a non-significant type. In Khoy station, between the years 2000, 2003, 2006, and 2007, a significant and sudden change began, which is insignificant. For the entire province of West Azerbaijan, a sharp and sudden change in the direction of an insignificant increase can be seen from 2003 to 2013. This sudden change has been revealed in the form of a decrease in 2013, which is not significant. In Parsabad and Meshkinshahr stations, the two components Ui and Ui' have not met each other inside and outside the critical range, and in the years 2008 and 2009, a significant decreasing trend started, which is insignificant, but in Ardabil station, these two components are within the range. In the years 2006 and 2010, they met each other, which indicates the beginning of a change of direction and a decreasing trend, albeit insignificant, in the frequency of convective precipitation at this station. This situation occurred in Khalkhal station in 2001 and 2008, which was accompanied by an increase in 2001 and an insignificant decrease in 2008. In general, the results of the analysis of the frequency of precipitation with the Mann-Kendall test showed that the variable trend, both increasing and decreasing, in the stations is not significant due to the location of the intersection point of Ui and Ui' outside the critical range (±1.96). In this way, the frequency of precipitation in the stations of East Azarbaijan province including Tabriz, Maragheh, Bonab, and Sarab shows a non-significant decreasing trend, and in Marand, Ahar, Bostan Abad, and the middle of a non-significant increasing trend, in the stations of West Azarbaijan province (including Urmia, Mako, Mahabad), Sardasht and Khoy) has had a non-significant increasing trend in the stations of Ardabil province (Parsabad, Ardabil, Meshkin Shahr, and Khalkhal). The results of trend analysis using the age slope estimator method show an increasing trend in Mako, Urmia, and Meshkinshahr stations. Also, the frequency of rainfall in Bostan Abad, Mahabad, Ardabil, and Parsabad stations has no trend and in other stations, it shows a decreasing trend without significance.