Intraseasonal evolution and climatic characteristics of hourly precipitation during the rainy season in the Poyang Lake Basin, China

Abstract In the key flood control area of the Poyang Lake Basin (China), over 80% of annual precipitation is concentrated in the rainy season. This study investigated the intraseasonal evolution and climatic characteristics of precipitation during the rainy season in the Poyang Lake Basin under the background of climate change. On the basis of the precipitation pattern, the rainy season was divided into three stages: the spring rainy season, Meiyu season, and summer rainy season. Generally, total precipitation and precipitation hours showed no significant trends of change, whereas precipitation intensity had an obvious trend of increase. In the spring rainy season, the trend of total precipitation was downward but not significant; however, owing to substantial reduction in precipitation duration, precipitation intensity was enhanced significantly. In the Meiyu season (summer rainy season), total precipitation increased significantly because of increased precipitation duration (precipitation intensity). The spatial distributions of total precipitation in the spring rainy season and Meiyu season were similar, i.e., precipitation occurred mostly in the east and less in the west; however, the spatial distributions of the change trends were opposite. In the summer rainy season, precipitation was concentrated primarily in the surrounding mountainous areas.


Introduction
The monsoonal rainy season in eastern China has obvious regional and temporal characteristics (Ding 1994;Ding and Chan 2005;Wang and Ding 2008). The development and establishment of the summer monsoon over the South China Sea forms the First Flood Period in South China, the Meiyu in the basins of the Yangtze and Huaihe rivers, and the Rainy Season in North China (Wang 1981;Tao and Chen 1987;Ding 1992;Huang et al. 1992). The Poyang Lake Basin, located in the middle and lower reaches of the Yangtze River in China, has a subtropical warm and humid monsoon climate (Figure 1). The rainy season in the Poyang Lake Basin has long been considered part of the First Flood Period in South China (Chi et al. 2005;Chen et al. 2006;Huang et al. 2014). However, it is recognized that a precipitation peak occurs in the Poyang Lake Basin before the First Flood Period in South China (Chen 1991;Chen et al. 1991). This early precipitation peak is an indicator of the onset of the rainy season in East China (Zhan et al. 2017). This precipitation period, which is called the spring rainy season in Jiangnan, is a type of hazardous weather that is a unique climatic phenomenon in East Asia (Wan and Wu 2008). He et al. (2008) suggested that the spring rainy season in Jiangnan is the "breeding stage" of the subtropical monsoonal precipitation, and that the Meiyu season that occurs after the onset of the South China Sea monsoon represents the peak of the subtropical monsoonal precipitation. Owing to its unique geographical location, the rainy season of the Poyang Lake Basin comprises three distinct periods: the spring rainy season, the Meiyu season, and the summer rainy season.
Although no significant change has been observed in the long-term trend of annual precipitation in China, the interdecadal, interannual, and regional variations are remarkable (Zhai et al. 2005;Shang et al. 2019). Regionally, extreme heavy rainfall has shown a significant trend of increase and light rainfall has shown a significant trend of decrease over central eastern China. However, in northeastern China and the Sichuan Basin, the precipitation amount in each precipitation intensity category has shown a trend of decline (Zhai et al. 2005;Yao et al. 2008;Yu et al. 2010). Seasonally, precipitation has increased during winter and summer but has decreased during spring and autumn (Huang and Wen 2013;Wen et al. 2015;Li et al. 2016). It has been established that maximum daily precipitation is increasing in southern river basins but decreasing in northern river basins, which leads to no discernible trend of increase or decrease in maximum daily precipitation across China as a whole . Investigation of the interannual variability of rainfall has also revealed that the number of heavy precipitation days has increased since the 1980s in most of China, particularly in regions of southern China and areas to the south of the Yangtze River Basin (Liu 1999;Yang et al. 2013). Under the background of global climate warming, precipitation in the Poyang Lake Basin has changed significantly (Wang et al. 2009;Zhan et al. 2011). Since the 1960s, the annual number of precipitation days (daily precipitation ! 0.1 mm) in the Poyang Lake Basin has shown a significant downward trend, whereas the number of heavy rain days (daily precipitation ! 25.0 mm) has increased (Zhan et al. 2019). The trend of change of extreme precipitation is more obvious than that of annual precipitation (Gao and Ren 2016).
Recent research on precipitation in the Poyang Lake Basin has focused mainly on the climatic characteristics of the entire rainy season or the entire year, neglecting variations in different periods of the rainy season. In fact, total rainy season precipitation in the Poyang Lake Basin can often be "normal" but with drought in one period or flooding in another. Notably, precipitation in different periods has different varying degrees of impact in the Poyang Lake Basin. In spring (Mar-May), persistent rainy days have greatest influence on agriculture. For example, long periods of rain might prevent the germination of seeds. In the Meiyu season, continuous heavy rain can lead to landslides, debris flows, and basin flooding. In summer (Jul-Aug), precipitation is caused mainly by strong convection in the afternoon that can often lead to urban waterlogging. Therefore, in undertaking comprehensive research, it is very important that the intraseasonal evolution and climatic characteristics of hourly precipitation in the rainy season in the Poyang Lake Basin be analyzed.

Study area
The Poyang Lake Basin (28 22 0 -29 45'N, 115 47 0 -116 45'E, area: 1.62 Â 10 5 km 2 ; Figure 1) holds the largest freshwater lake in China. The region has a subtropical humid monsoon climate, with annual precipitation of the order of 1662 mm and annual mean temperature of 18.1 C . The heavy rainfall that occurs during the major rainy season (March-September) contributes approximately 80% of the regional annual precipitation.

Meteorological data
The quality-controlled hourly precipitation data used in this study were recorded at 79 meteorological stations during 1978-2019 (black triangles in Figure 1). Prior to further analysis, this dataset was obtained from the National Climatic Reference Network and National Weather Surface Network of China, and it was collected and quality-controlled by the National Meteorological Information Center of the China Meteorological Administration. The data were tested for homogeneity and the average annual rate of missing data for the selected study period was 3.01%.

Intraseasonal evolution of the rainy season
Precipitation in the Poyang Lake Basin, which occurs mainly during March-September (the rainy season), accounts for approximately 80% of the total regional annual precipitation ( Figure 2). Precipitation in the Poyang Lake Basin has well-characterized stages during the rainy season. The main driving factors of precipitation in the different stages are not the same. In the first stage (from March to early May), the spring continuous rain, which a unique weather and climatic phenomenon in East Asia, is the result of dynamic and thermal forcing by the high terrain of the Qinghai-Tibet Plateau. The precipitation during this stage is mainly frontal precipitation (Wan and Wu 2008). In the second stage (May-June), from the 4th to 5th pentads in May, the South China Sea monsoon breaks out. At this time, a belt of water vapor transport is established that extends from the low-latitude ocean in the Southern Hemisphere to southern China via the Arabian Sea, Indian Ocean, and Bay of Bengal (He et al. 2008). The precipitation during this stage is mainly monsoonal precipitation. In the third stage (July-September), the rainband of the East Asian subtropical monsoon extends northward to North and Northeast China. The precipitation during this stage is mainly convective precipitation (Jiang et al. 2015). Precipitation duration is an important feature of any precipitation process. Compared with volume, frequency, and intensity, the duration of a precipitation event is more closely related to the characteristics of the precipitation process (Li et al. 2011). Yu and Zhou (2007) proposed that duration is the key factor for distinguishing different types of precipitation process. Precipitation events persisting for 1-6, 7-12, and !13 h are called short-, medium-, and long-duration precipitation events, respectively (Wu and Zhan 2020). It can be seen in Figure 3 that rainy season precipitation in the Poyang Lake Basin has evident characteristic stages. In the first stage (13th to 27th pentads), i.e., the spring rainy season, the volume of precipitation begins to increase, and most precipitation events are of short and medium duration. In the second stage (28th to 36th pentads), i.e., the Meiyu season, precipitation increases and events with rainfall of >1.6 mm can be of short, medium, or long duration. In the third stage (37th to 49th pentads), i.e., the summer rainy season, precipitation events are mainly of short duration. The precipitation duration distribution of each stage has distinct characteristics.

Theil-Sen's slope estimator
The Theil-Sen's slope estimator, which is an unbiased nonparametric estimator of the true slope in simple linear regression, was used in this study to examine the longterm trends. If a given time series has a linear trend, then the level of increase or decrease per unit time can be described by the Theil-Sen's slope (Sen, 1968). Unlike least squared linear regression, the Theil-Sen's slope estimator is insensitive to outliers and thus the estimated linear trend is significantly more accurate and robust for skewed data. The slope of N pairs of data points can be estimated using the following relation: where x l and x j are data values at time l and time j, respectively. Owing to its robustness for estimating the magnitude of a trend, this method is applied widely to hydrological and climatic time series (Cunderlik and Burn, 2003;Tabari and Talaee, 2011;Jhajharia et al. 2012;Wang et al. 2013).

Mann-Kendall trend test
The Mann-Kendall (M-K) test statistic S (Mann, 1945;Kendall, 1975) is calculated as follows: where n is the number of data points, x i and x j are data values in time series i and j (j > i), respectively, and sgn(x j . x i ) is the sign function, which is given as follows: The variance is computed as follows: where n is the number of data points, m is the number of tied groups, and t i denotes the number of ties of extent i. A tied group is a set of sample data having the same value. In cases with sample size n > 10, the standard normal test statistic Z S is computed using the following relation: Positive (negative) values of Z S indicate increasing (decreasing) trends. Testing of trends is done at a specific a significance level. When jZSj > Z 1Àa/2 , the null hypothesis is rejected and a significant trend exists in the time series; Z 1Àa/2 is obtained from the standard normal distribution table. In this study, significance levels of a ¼ 0.01 and a ¼ 0.05 were used. At the 5% (1%) significance level, the null hypothesis of no trend is rejected if jZSj > 1.96 (jZSj > 2.576).
The M-K statistical test has been used frequently to quantify the significance of trends in hydrometeorological time series (Douglas et al. 2000;Yue et al. 2002;Partal and Kahya 2006;Modarres and Silva 2007).

Diurnal variation of precipitation
In the Poyang Lake Basin, the diurnal variation characteristics of precipitation in the three different stages are distinct (Figure 4). In the spring rainy season, precipitation presents a single-peaked distribution, and it mainly occurs during 06:00-09:00 (all times expressed as Beijing time), although afternoon precipitation does gradually increase as the rainy season progresses. In the Meiyu season, precipitation shows a bimodal distribution, with precipitation occurring mainly during 06:00-10:00 and 14:00-16:00. After the Meiyu season, precipitation starts to decrease. In the summer rainy season, precipitation is mostly convective and concentrated during 16:00-18:00.

Intraseasonal variation of precipitation
Rainy season precipitation in the Poyang Lake Basin has obvious stage characteristics. In terms of volume (Figure 5a), precipitation in the Poyang Lake Basin begins to increase significantly after the 13th pentad, which indicates the beginning of the spring rainy season. The total volume of a pentad is more than 20 mm. The first peak appears between the 19th and 27th pentads. The pentad precipitation volume is >34.7 mm. The maximum precipitation occurs in the 27th pentad with a pentad volume of 40.6 mm. The reduction in precipitation in the 28th pentad indicates the end of the spring rainy season and the beginning of the Meiyu season. Precipitation increases rapidly after the 29th pentad. The second peak appears between the 31st and 35th pentads, and the mean pentad precipitation is 44.6 mm. After the 36th pentad, precipitation decreases rapidly, indicating the end of the Meiyu season. After the 37th pentad, the summer rainy season starts. Unlike the spring rainy season and the Meiyu season, most precipitation in the summer rainy season is associated with convective and typhoon-related processes. The third peak appears between the 42nd and 49th pentads, and the mean pentad precipitation is 22.6 mm. After the 49th pentad, pentad precipitation is <10 mm, indicating the end of the rainy season in the Poyang Lake Basin.
In terms of precipitation duration (Figure 5b), the accumulated pentad precipitation hours show a bimodal distribution. The maximum peak appears in the spring rainy season, and the mean pentad precipitation hours are 22.5 h from the 13th to 21st pentads. During the Meiyu season, the pentad precipitation hours start increasing again. The second peak appears between the 31st and 34th pentads, with mean pentad precipitation hours of 19.7 h. In the summer rainy season, the precipitation hours decrease significantly, i.e., the pentad precipitation hours are around 10 h.
In terms of intensity (Figure 5c), the precipitation also shows a bimodal distribution. In the spring rainy season, the intensity of hourly precipitation increases to reach a peak of 2.2 mm/h in the 27th pentad. After the spring rainy season, precipitation intensity first decreases and then increases again. It reaches the annual maximum of 2.5 mm/h at the 34th pentad and it remains largely unchanged between the 34th and 48th pentads. After the summer rain season, precipitation intensity decreases to no more than 1.0 mm/h.
In the Poyang Lake Basin, precipitation in the different stages of the rainy season has distinctive features. For example, there are more precipitation hours in the spring rainy season, greater pentad precipitation in the Meiyu season, and higher precipitation intensity in the summer rainy season. In the Poyang Lake Basin, days of continuous precipitation are common in March (13th to 19th pentads). Consequently, the number of precipitation hours in March is higher in comparison with other months.
In March, a belt of water vapor transport from the Western Pacific to southern China is formed, and convergence at the front of this strong belt of water vapor transport leads to substantial increase in precipitation in the region south of the Yangtze River (including the Poyang Lake Basin) in late March (He et al. 2008). The Poyang Lake Basin is located downstream of the center of strong southwesterly winds on the southeastern side of the Qinghai-Tibet Plateau. This strong wind and moisture convergence are the direct causes of the spring rainy season. After the onset of the South China Sea monsoon, there is a period of peak precipitation in the basin as the Meiyu season arrives (Wan and Wu 2008). The precipitation in the summer rainy season is most often related to severe convective weather or typhoons; consequently, precipitation intensity is relatively high (Wang et al. 2015).

Annual variation of precipitation
The annual variation characteristics of rainy season precipitation in the Poyang Lake Basin during 1978-2019 are shown in Figure 6. Annually, precipitation volume during the entire rainy season is approximately 1240 mm, with a maximum of 1677 mm in 2010 and a minimum of 911 mm in 1978 (Figure 6a). It can be seen that precipitation volume increases at the rate of 25.7 mm/10a during 1978-2019, but the trend is not significant (M-K value: 1.08). Annually, precipitation hours in the rainy season are 683 h, with a maximum of 864 h in 1999 and a minimum of 519 h in 2018 (Figure 6b). Unlike precipitation volume, precipitation hours during the study period decrease slightly at the rate of À4.0 h/10a, but the trend does not pass the significance test (M-K value: À0.32). Annually, precipitation intensity in the rainy season is approximately 1.82 mm/h, with a maximum of 2.26 mm/h in 1983 and a minimum of 1.52 mm/h in 1989 (Figure 6c). During the study period, precipitation intensity shows a trend of increase of 0.07 mm/h/10a, and it passes the significance test (M-K value: 3.84). Generally, during 1978-2019, precipitation in the rainy season shows a trend of increase; however, precipitation intensity is obviously enhanced owing to the reduction in precipitation hours.
Annually, precipitation in the spring rainy season is around 504 mm, with a maximum of 721 mm in 2010 and a minimum of 271 mm in 2011 (Figure 7-1a). During 1978-2019, it decreases at the rate of 10.4 mm/10a, but this trend fails to pass the 0.01 significance level test (M-K value: À0.65). Annually, precipitation hours in the spring rainy season are around 340 h, with a maximum of 440 h in 1980 and a minimum of 239 h in 2018 (Figure 7-1b). Similar to precipitation volume, precipitation hours also show a decreasing trend, with a rate of À18.6 h/10a that passes the significance test (M-K value: À3.00). Annually, precipitation intensity in the spring rainy season is around 1.48 mm/h, with a maximum of 1.91 mm/h in 2016 and a minimum of 1.07 mm/h in 2011 (Figure 7-1c). Unlike the other parameters, precipitation intensity shows an increasing trend of 0.06 mm/h/10a that passes the significance test (M-K value: 2.17). Overall, precipitation in the spring rainy season shows a trend of decrease but with a significant increase in precipitation intensity owing to the significant reduction in precipitation hours. Annually, precipitation in the Meiyu season is around 368 mm, with a maximum of 645 mm in 1995 and a minimum of 197 mm in 1991 (Figure 7-2a). Different from the spring rainy season, it can be seen that precipitation increases at the rate of 30.6 mm/10a after 1978, which is a trend that passes the 0.01 significance level test (M-K value: 2.80). The precipitation hours in the Meiyu season also increase at the rate of 11.2 h/10a, which is a trend that also passes the significance test (M-K value: 2.19) (Figure 7-2b). Precipitation shows a trend of slight increase of 0.05 mm/h/10a (Figure 7-2c). Overall, the volume, hours, and intensity of precipitation during the Meiyu season show trends of increase, and the increase of the total precipitation is mainly attributable to the increase in precipitation hours. Annually, precipitation in the summer rainy season is around 368 mm, with a maximum of 633 mm in 1999 and a minimum of 176 mm in 1978 (Figure 7-3a). Similar to the Meiyu season, it can be seen that precipitation volume increases at the rate of 14.0 mm/10a after 1978, but the trend fails to pass the 0.01 significance level test (M-K value: 1.34). During the study period, precipitation hours decrease at the rate of 0.5 h/ 10a, but this trend fails the significance test (M-K value: À0.13) (Figure 7-3b). Conversely, precipitation intensity shows a nonsignificant trend of increase of 0.06 mm/ h/10a (M-K value: 1.41) (Figure 7-3c). Overall, the volume and intensity of precipitation during the Meiyu season show trends of increase, whereas precipitation hours are decreasing. The increasing precipitation intensity is the main reason for the increasing precipitation volume.

Spatial distribution and variation of precipitation
It can be seen from Figure 8 that the spatial distribution pattern of precipitation in the spring rainy season is similar to that of the Meiyu season. The overall distribution indicates greater precipitation in the east of the Poyang Lake Basin than in the west. The centers of high values are concentrated in eastern central parts of the basin, while the centers of low value are mainly in northern and southwestern areas. The spatial distribution of precipitation in the summer rainy season has no obvious regularity, which might be related to topography, i.e., there is more precipitation in the mountainous areas than over the area of the central plain. Figure 9 illustrates

Conclusions and discussion
On the basis of the characteristics of precipitation, the rainy season in the Poyang Lake Basin was divided into three stages: the spring rainy season (13th to 27th pentads), Meiyu season (28th to 36th pentads), and summer rainy season (37th to 49th pentads). To elucidate further details regarding changes in the rainy season in the Poyang Lake Basin, we analyzed the spatiotemporal characteristics of precipitation in the three stages. On the basis of data obtained in this study and by others, the following conclusions were drawn.
1. The diurnal variation characteristics of precipitation in the three stages are obviously different. In the spring rainy season, precipitation is concentrated mainly during 06:00-09:00. As time passes, the occurrence of afternoon precipitation gradually increases. There is a bimodal distribution of precipitation in the Meiyu season: one peak appears during 06:00-10:00 and the other appears during 14:00-16:00. After the Meiyu season, precipitation decreases and there is only one peak during 16:00-18:00 in the summer rainy season. 2. Generally, precipitation in the rainy season is increasing. Although precipitation in the spring rainy season is decreasing, precipitation in both the Meiyu season and the summer rainy season is increasing but for different reasons. In the Meiyu season (summer rainy season), the increase of precipitation is mainly due to increased precipitation hours (precipitation intensity). It is worth noting that precipitation intensity in all three stages shows an increasing trend, which will bring a greater risk of flooding in the Poyang Lake Basin.
3. The spatial distributions of precipitation in the spring rainy season and the Meiyu season are similar, with most (least) precipitation in eastern (western) parts of the Poyang Lake Basin. However, the precipitation distribution in the summer rainy season is different, i.e., most precipitation occurs in the surrounding mountainous areas and less falls over the area of the central plain. Spatially, in the spring rainy season, precipitation at most stations in the Poyang Lake Basin is decreasing, especially in the south. Conversely, in the Meiyu season, most stations show precipitation is increasing, especially in the south. In the summer rainy season, the trends of change in precipitation at most stations are not significant; only two mountain stations show a significant trend of increase.
In East China, annual precipitation shows no obvious change, but precipitation concentration shows an increasing trend. The number of days with light rain (daily rain: 10 mm) shows significant decrease, while the number of days with heavy rain (daily rain: !50 mm) shows significant increase (Zhai et al. 2005;Qian et al. 2007;Yu et al. 2011;Huang and Wen 2013;Wen et al. 2015;Li et al. 2016). Similar to East China, precipitation intensity in the Poyang Lake Basin is also increasing significantly (Zhan et al. 2013;Gao and Ren 2016). However, this study established that although precipitation in the Poyang Lake Basin might follow the general variation of precipitation in East China, it has its own characteristics. The increase in precipitation volume in the Meiyu season is greater than the decrease in volume in the spring rainy season, which means that the annual volume of precipitation in the entire rainy season is increasing. The increase in Meiyu precipitation is the primary reason for the overall increase in the volume of rainy season precipitation in the Poyang Lake Basin. Wan and Wu (2008) suggested that the spring rainy season in the Poyang Lake Basin is the result of dynamic and thermal forcing associated with the high terrain on the Qinghai-Tibet Plateau, and its rain belt coincides with southern parts of the Wuyi Mountains, highlighting that the spring rainy season in the Poyang Lake Basin is closely related to terrain. The occurrence of heavy rainfall in summer in the Poyang Lake Basin is a positive response to regional warming (Gao et al. 2001;Tang et al. 2018). Changes in the Meiyu season might be related to variation in uppertropospheric temperature (Yu et al. 2004;Xin et al. 2006;Yu and Zhou 2007). For example, cooling of the upper troposphere leads to weakening of the East Asian summer monsoon. Consequently, precipitation in northern China is reduced, while precipitation in the Yangtze River Basin is increased further. Thus, precipitation in the Meiyu season in the Poyang Lake Basin is increasing. The factors that influence precipitation in the Poyang Lake Basin are very complex and inconsistent throughout the rainy season, and further study is required to determine their relative importance within the context of climate change.

Funding
This study was supported by the Jiangxi Meteorological Bureau, Jiangxi Eco-meteorological Center, and the Natural Science Foundation of Jiangxi Province (20202BABL203036).