Abstract

Precipitation is a key variable in identifying climate changes in a region. In the present study, in the first stage, the homogeneity and trend test for the rainfall data of 18 stations in the watershed of Jiroft Dam, in the period from 1991 to 2020, was carried out using the Petit, Buishand, and normal standard tests and the non-parametric Mann-Kendall test. After that, in order to correct and remove outlier data, homogeneity and change-points were checked in the second stage, and by repeating the test for the intervals separated in each stage of the test, the time of occurrence of other change-points was also identified. By examining all the stages of separation, non-homogeneity in precipitation data was observed in all stations. The results of the homogeneity tests are different in different stations. In a certain station, the change-points were identified by different tests in different years. In general, 41.29% of non-homogeneity cases were diagnosed by Buishand's test, 34.83% by normal standard test and 23.88% by Petit's test. The results of the Mann-Kendall test showed that 15.86% of the data have a trend, 4.83% of the data have a deacreasing trend, and 11.03% of the data have an increasing trend. The most trending cases were observed in Jamilabad station with three deacreasing trends and one increasing trend, However, no significant trend was observed. In addition, there was no clear relationship between the significance of the homogeneity tests and the existence of trend in data series. The study results are useful for water resource management, drought reduction and sustainable agricultural planning in the region.





Keywords:

Buishand test, Mann–Kendall, nonparametric trend tests, normal standard test, Petit test



Introduction:

Precipitation is a very key variable in identifying climate changes in a region. Changes in precipitation may lead to floods, droughts, loss of biodiversity, agricultural productivity, and hydropower generation. An important and inseparable part of meteorological and hydrological studies is to ensure the accuracy of the data used (Hallegatte 2012; WMO 2017). Before ensuring the accuracy and quality of data and time series, it is not possible to use them and extract the next results (Mohammadi and BiAzar , 2017). In order to ensure the accuracy of the data series, different validation tests such as homogeneity and trend tests have been presented by researchers. Knowledge of the time range of data changes can be useful in choosing the right test, but unfortunately, this information is not available in Jiroft basin. Since the performance of tests in identifying hemoginity and change-points is dependent on environmental conditions, and it is not possible to say for sure which test has better performance, therefore, in this study, homogeneity and trend of precipitation data series was analyzed using three homogeneity tests: normal standard, Petit, Buishand and Mann-Kendall test.



Methodology

In the present study, the homogeneity and trend of average monthly rainfall data for 18 rain gauge locations in Jiroft Dam catchment from 1991 to 2020 were investigated. For this purpose, Petit, standard normal and Buishand homogeneity tests were used to check homogeneity and identify possible change-points. Since, at each step of tests, only one change-point is detected, so with each change-point detection, the desired time period is separated into years before and after the change-point, and the tests are repeated, this process continues until all change-points are detected. This process is stopped when the time period under review is less than 5 years. The non-parametric Mann-Kendall test was used to evaluate the trend of the data series.



Results and Discussion

The results of the tests and the process of separation of time intervals are given in tables (2), (3) and figure (2). In each station, the abilities of the tests were investigated separately. The results of homogeneity tests showed that the results of the tests are highly dependent on the location of the station and the time of data measurement. , so it is necessary to perform all three tests in order to identify all the inhomogeneities. In the next step, all point mutations in the data series were identified and removed by time separation of the data. In order to correct and remove mutations, the most number of separations was done for the month of Mehr (11 separations), May (9 separations), July (5 separations), June (5 separations), and April (4 separations). Petti's test showed the least amount of change-point and heterogeneity, the results of normal standard and Buishand tests were similar most of the time. Heterogeneity was observed in all stations of the basin. The most point-changes occurred in the months of March and April. The trend test results showed that none of the rainfall gauge stations had a significant decreasing or increasing trend on an annual scale, and on a monthly scale. there is an increasing trend in some months and a deacreasing trend in some months . In the studied basin, the most trend occurred in the first 6 months of the year. The largest number of inceasing trend was observed in April and at 5 rainfall gauge stations (Ismail Abad, Hossein Abad, Soltani, Maidan and Namzad). At Hanjan station, an increase of 54.49 mm of rainfall was observed in May and a decrease of 68.119 mm in March. In the month of March, a decrease in rainfall has been observed in Hanjan, Bidkardoiyeh, Jamil Abad, Cheshme Arous, Kiskan and Maidan stations, respectively 68/119, 47/56, 45/64, 161/68, 59/548 and 111/91 mm, but In the results of Kendall's test, no significant trend was observed for the precipitation data in this month. In the month of April, there was an increase in rainfall by 82.77, 54.33, 118.08, 32.68, 79.46 and 42.3 mm respectively in Namzad, Ismail Abad, Talkhe Char, Soltani, Kanarowieh and Maidan stations. By comparing with the results of the Mann-Kendall test, it can be seen that it was not just a sudden jump and the precipitation has an increasing trend. In the month of May, an increase in rainfall has been observed in Hanjan, Cheshme Arous, Raber and Glochar stations, respectively 54.409, 41.729, 6.81 and 33.3 mm, in this month only in Raber and Glochar stations there is an increasing trend was observed.



Conclusion:

The results of the tests were very different according to the station and month under investigation and the results were not the same. In general 41.29% of heterogeneous cases were identified by Buishand's test, 34.83% by normal standard test and 23.88% by Petti's test. On average, the number of point-change is 2. The results of the data series trend analysis showed that in general, 4.83% of the data have a decreasing trend and 11.03% have an upward trend. The highest number of trends was observed in Jamilabad station with three decreasing trends and one increasing trend. However, no significant trend was observed. In addition, there was no clear relationship between the significance of the homogeneity tests and the existence of trend in data series.