Comparative Analysis of Rainfall Trends in the Jinghe River Basin of China During 1959-2014

8 Trend analysis is widely applied in hydrometeorological research. Considering that Innovative 9 Trend Analysis (ITA) and Innovative Polygonal Trend Analysis (IPTA) can detect small variations 10 on annual and smaller scale, rainfall trends at 14 hydrometeorological stations in the Jinghe River 11 Basin were analyzed by ITA, IPTA and Mann Kendall test (MK). The results showed that the 12 rainfall trends are subsistent from 1959 to 2014. Comparing the results of ITA and MK on annual 13 level, it was determined that trends are consistent, but only two stations passed the 90% 14 significance test through MK, while all stations passed the significance test through ITA. 15 Accordingly, the ITA method proved to be better than MK in detecting small changes in time 16 series. Changes in high and low values, obtained by the ITA method, reflected flood and drought 17 trends in the basin. In addition, IPTA is an improved ITA method that is suitable for a relatively 18 short time span. Through the IPTA method for analyzing the monthly precipitation trends, the 19 results showed that rainfall at 14 stations increased in January, February, March, June and 20 December, and decreased significantly in September. Therefore, the methodology applied in this 21 study can provide detailed recommendations for hydrometeorological research.


26
Rainfall is a basic component of the global water cycle (Michaelides et al., 2009), and also one of 27 the most important input variables in climate and hydrological research. For arid and semi-arid areas, 28 the change in precipitation is directly related to the change of surface runoff and regional dry and wet 29 conditions. According to the Fifth Assessment of the Intergovernmental Panel on Climate Change 30 (CMIP5) (IPCC, 2014), the global mean surface temperature has increased 0.74 ℃ over the last century.

31
The increase in global temperature is affecting and will continue to affect the water cycle, leading to 32 changes in rainfall distribution (Chen et al., 2014;Zhang, Gao,&Zhao, 2003). Previous research (Wen, (Serinaldi, Chebana,& Kilsby, 2020) have shown that the ITA method is influenced by sample 49 size, distribution shape, and serial correlation, its advantages can be applied in different aspects.

50
Kisi (Kisi & Ozgur, 2015) found that the graphs obtained by the ITA method can better reveal the 51 hidden trends in pan evaporation, compared to the MK and SR tests. Once the ITA method was 52 proposed, it is not only applied in different aspects and regions, yet many scholars have improved the 53 ITA method. Şen (Şen, 2017) further improved the ITA method and developed a calculation approach to 54 derive a monotonic trend and a significance test, making it easier to obtain the trend behavior for 55 different time series categories (low, medium, and high) (Güçlü, 2018). The new ITA 56 visualization (Güçlü, 2020) shows clearly a number of data unlike the ITA applications. In addition, the 57 change point on difference series is identified by universal Pettitt test, and then two subcategories are 58 objectively defined as "high" and "low" values. The ITA-change boxes (ITA-CB) approach(Alashan,

96
The Mann-Kendall test(Alashan, 2018; Kendall, 1990;Mann, 1945) was used in this study to 97 explain sequences in precipitation trends and variability. The basic principle of this method is based on 98 a comparison of the time series (Z) in itself   1 2 , , , n z z z . If the examined data is larger (smaller) 99 than the following, then −1(+1) is added to the MK statistics (S) (Eq. 1). Here, the variable (i) varies 100 from 1 to n−1, and the variable (j) varies from i+1 to the data length (n). This process is repeated for 101 all data element and S statistics are calculated and summed (Eq. 2).

102
It is assumed that S has a normal probability distribution function (PDF) with zero mean and some 122 3) The first sub-series   1 S are located on the x-axis, and the other sub-series           Zhenyuan are relatively small indicating that the monthly rainfall at these stations is relatively 207 consistent with more stable occurrence of hydro-meteorological events.

208
In Jingyang, rainfall remained above the 1:1 straight-line during five months (January, February,

209
March, June, and December), which indicates that there was more rainfall during these months in the 210 recent investigated period (1987-2017) compared to the previous investigated period