Characteristics of diurnal variations of warm-season precipitation over Xinjiang Province in China

Ji Co , , , Shuping M , , Wihu Yun , , Zhiyn Wu

a Key Laboratory of Meteorological Disaster (KLME), Ministry of Education and Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters(CIC-FEMD), Nanjing University of Information Science and Technology, Nanjing, China

b Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China

c Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, Oklahoma, USA

d University of Chinese Academy of Sciences, Beijing, China

e Key Laboratory of South China Sea Meteorological Disaster Prevention, Haikou, China

Keywords:Diurnal variation Precipitation Rainfall duration Regional variation Arid and semi-arid regions

ABSTRACT Aimed at improving knowledge regarding the diurnal cycle of warm-season rainfall in northwestern China,this study investigated the diurnal variations of warm-season precipitation with different durations in Xinjiang Province, China, using an hourly gauge—satellite merged precipitation product during 2008 to 2019. Results show noticeable diurnal variations with distinctive regional features. The primary peak is in the early evening. Rainfall events with duration less than 3 h occur more often across the whole of Xinjiang and contribute more than half of the precipitation amount over its northern and southern peripheries, while rainfall events with duration more than 7 h over the Tianshan Mountains are responsible for the primary peak in the diurnal variations of warm-season precipitation.

1. Introduction

Studying the rainfall diurnal cycle is an important way to improve both the physical understanding of rainfall processes and the capability of numerical models in precipitation forecasting and simulation( Dai et al., 1999 ; Lin et al., 2000 ). The diurnal variation of precipitation has always been a hot topic and an atmospheric phenomenon of broad concern. Previous studies show large diurnal variations of warmseason precipitation and a maximum in late afternoon over most land areas worldwide ( Bleeker and Andre, 1951 ; Wallace, 1975 ; Schwartz and Bosart, 1979 ; Trenberth, 1998 ; Dai et al., 1999 ; Trenberth et al., 2003 ).With the rapid development of satellite-retrieved precipitation data in recent years, the diurnal variations of precipitation over the East Asian monsoon region have been extensively studied and distinct regional features revealed, especially over North China and central-eastern China( Yu et al., 2007a , 2014 ; He and Zhang, 2010 ; Chen et al., 2013 ; Wu et al.,2018 ). Most rainfall over land areas peaks in the late afternoon and the nocturnal diurnal peaks are often found in the downstream of elevated areas, coastal areas, or valleys ( Nesbitt and Zipser, 2003 ; Zhao et al.,2005 ; Zhou et al., 2008 ; Yuan et al., 2012 ).

Central Asia is located in the hinterland of the Eurasian continent and is one of the largest arid and semi-arid regions in the world. It is sensitive to global warming, which in turn threatens future ecosystem safety as well as regional stability ( IPCC, 2007 ; Li et al., 2015 ). Because of limited data availability, not enough in-depth studies on precipitation have been conducted in this important area. Recently, many scientists have started to pay more attention in this area. For example,Zhang et al. (2017) studied the annual changes of precipitation extremes in arid Central Asia (excluding Xinjiang Province). Hu et al. (2017 ,2018 ) explored the spatiotemporal variations of the annual precipitation in the last century, and then evaluated the validity and accuracy of three sets of global gridded data over Central Asia with a focus on seasonal and annual variations. Lai et al. (2020) investigated the fidelity of the APHRODITE dataset ( Yatagai et al., 2012 ) over Central Asia in terms of the daily extreme precipitation. Jiang and Zhou (2021) recently studied the trend and possible reasons for decreasing precipitation over northern Central Asia. However, the diurnal variation of precipitation over Central Asia has rarely been studied. Many of the land areas in Central Asia are in the shadows of high mountain ranges with arid and semi-arid climates ( Lioubimtseva and Cole, 2006 ). Yatagai and Yasunari(1995 , 1998 ) studied the interannual variations of summer precipitation and water vapor transport in the arid and semi-arid regions in eastern Central Asia and revealed their differences to characteristics in monsoon regions. Xinjiang Province, which is located in the center of the Eurasian continent and northwestern China, is a typical arid and semi-arid region and contains the Taklimakan Desert. Meanwhile, the topography of Xinjiang Province is complex. There are three mountain ranges located in the northern, central, and southern parts of Xinjiang Province, and two basins situated between the mountains. The complicity of the rainfall characteristics shown in Xinjiang Province could be a typical example in the arid and semi-arid regions in Central Asia, as well as the area with complex topography. Therefore, the diurnal variations of the arid and semi-arid regions in Xinjiang deserve further analysis.

With the increased availability of extensive and quality-controlled observational data, it has become possible to analyze the hourly precipitation features over northwestern China in detail. Aside from the diurnal features of total rainfall, different-duration rainfall events in eastern China have also been found to possess different features ( Yu et al.,2007b ). In this paper, the diurnal variations of total rainfall and rainfall events with different duration are analyzed over Xinjiang Province to improve our understanding of rainfall features over arid and semi-arid regions.

The rest of this paper is organized as follows: The data and methods are introduced in Section 2 . The basic characteristics of the diurnal variation of warm-season precipitation over Xinjiang are investigated together with regional features in Section 3 . The relationship between rainfall duration and diurnal variation is studied in Section 4 . And finally, Section 5 provides a summary and some further discussion.

2. Data and methods

The precipitation dataset used in this study is an hourly gauge—satellite merged precipitation product with a horizontal resolution of 0.1° latitude/longitude covering the period 2008—2019. It can be obtained from the National Meteorological Information Center of China ( Shen et al., 2014 ; available at http://data.cma.cn/data/detail/dataCode/ ). This dataset was developed by combining quality-controlled rain-gauge precipitation records at automatic weather stations with NOAA’s Climate Precipitation Center Morphing technique ( Joyce et al., 2004 ), and thus provides detailed distributions of rainfall variations over China. Many studies have proven the good performance of this dataset in analyzing heavy weather events over China ( Zhu, 2015 ; Sun et al., 2016 ). The seasonal distribution of precipitation in Xinjiang Province is similar to that in other regions of China, which means most precipitation starts in April and ends in September ( Joyce et al., 2004 ). Therefore, the following analysis focuses on the warm-season precipitation. Hourly rain-gauge records from 88 national stations ( Fig. 1 ) over Xinjiang during the same period of the merged precipitation product are also used, for validation.The rain-gauge data were collected and quality-controlled by the National Meteorological Information Center of the China Meteorological Administration.

One distinct characteristic of Xinjiang is its special topography.There are two basins within three mountain ranges from north to south( Fig. 1 ). With consideration of the terrain height and the traditions adopted by local observatories ( Yuan et al., 2004 ; Yang et al., 2020 ), the target domain of Xinjiang Province is separated into three sub-regions,i.e., the northern, central, and southern part of Xinjiang Province (hereafter represented by N, C, and S, respectively). The remaining area, including the ridges of the Altai and Kunlun Mountains with altitudes over 3000 m, were subtracted from N and S. The sub-regions N, C, and S are outlined in Fig. 1 .

Fig. 1. (a) Spatial distributions of the phase clock of the maximum precipitation in LST (unit vectors) and normalized PA (shaded colors) of the warm-season precipitation averaged between 2008 and 2019. The three sub-regions N, C,and S are labeled and marked by yellow lines. (b) Diurnal peaks revealed by the national stations of Xinjiang Province (colored dots) overlaid by topography in grey (units: m).

The hourly measurable precipitation more than 0.1 mm is considered as the standard for a rainfall event. Following Yu et al. (2007b) ,the number of hours between the start and the end of an event without any or at most one-hour intermittence is defined as the duration.Based on different durations, all rainfall events are classified into three categories, as follows: short-duration rainfall events (duration ≤ 3 h);medium-duration events (3 h6 h). Because the medium-duration rainfall events occupy less than 20% of the total rainfall and the diurnal amplitude of them is weak, only the short- and long-duration ones are analyzed below. The precipitation amount (PA, with units of mm) and frequency(PF, with units of %), normalized by their daily means, were calculated for investigating the diurnal cycles of warm-season precipitation over different sub-regions in Xinjiang.

3. Diurnal variations of warm-season precipitation and their regional differences

Fig. 1 (a) shows that most of the large PAs (normalized, shaded) in the warm season during 2008 to 2019 in Xinjiang are located in the mountains, with the maximum over the western part of the Tianshan Mountains. This proves the need for separating the entire province into the three sub-regions before analyzing the characteristics of their diurnal variations in the following analysis. As seen from the phase clock of the maximum precipitation in local solar time (LST) in the form of unit vectors in Fig. 1 (a), late-afternoon precipitation peaks prevail over most part of N. Over the southern part of N, which is the Jungar Basin, rainfall mainly reaches its hourly maximum around 2100 LST. Compared with the diurnal peaks revealed by the national stations of Xinjiang Province( Fig. 1 (b)), the contrast between the northern and southern parts of subregion N is verified by the station rain-gauge data. Maximum precipitation constantly occurs in the evening over C, which shows a clockwise diurnal phase evolution that links to the southern part of N. Afternoon precipitation prevails in the majority of S, except for its western part.In western S, a nocturnal precipitation maximum is found, which is also verified by the station rain-gauge data. The above results seen from highresolution data confirm the findings of previous studies with sparse unit vectors over Northwest China ( Fig. 1 in Yu et al., 2014 ). Fig. 2 presents occurrences of the peak time in the form of the percentage to the whole day averaged over different regions. It can be seen that maximum precipitation over N (blue line in Fig. 2 ) and C (green line in Fig. 2 ) mostly occurs in the early evening and then early at night. Conversely, percentages of nocturnal and midday maximum precipitation are dominant over S (red line in Fig. 2 ). Averaged over the whole of Xinjiang (black line in Fig. 2 ), the rainfall peaks mainly occur during early evening to midnight.

Fig. 2. Percentages of the LST of the maximum precipitation to the daily 24 h precipitation over N (blue, left-hand y -axis), C (green, left-hand y -axis), S (red,left-hand y -axis), and the whole of Xinjiang (black, right-hand y -axis).

Fig. 3. Diurnal variations of warm-season PAs averaged over N (blue), C(green), S (red), and the whole of Xinjiang (black) averaged between 2008 and 2019.

In order to obtain the detailed regional features of the diurnal variations, the diurnal cycles of warm-season precipitation ( Fig. 3 ) were calculated and averaged over Xinjiang along with its three sub-regions, as outlined by the yellow lines in Fig. 1 (a). Both N (blue line in Fig. 3 ) and C(green line in Fig. 3 ) have a primary peak in the early evening at around 1900 LST, plus weak secondary peaks at around 0000 LST. The mean precipitation diurnal cycle over S (red line in Fig. 3 ) during 0300—0900 LST is almost opposite to that of N and C. It presents peaks at around 0000 LST, 1300 LST, and 1900 LST, confirming previous studies over Northwest China ( Fig. 3 (a) in Zhao et al., 2005 ). These diurnal cycles over each sub-region are consistent with their corresponding maximum precipitation variations in Fig. 2 . The diurnal cycle averaged over the whole of Xinjiang (black line in Fig. 3 ) seems similar to that over N and C before 1400 LST, and to that over S after then.

The diurnal variations of rainfall with different intensity are shown in Fig. 4 . For all three regions, heavy rainfall tends to peak in the early evening. However, the diurnal amplitude of light rainfall over the three regions is relatively weak. Although most rainfall weaker than 5 mm h?1of N and C still occurs in the early evening, there is a secondary diurnal peak in the early morning. For S, rainfall weaker than 5 mm h?1is relatively low in the early evening, which is different from in N and C.

Fig. 4. Normalized (by the daily mean) diurnal cycle of rainfall with different intensity ( y -axis, units: mm h ? 1 ) averaged over (a) N, (b) C, and (c) S.

Fig. 5. Percentage of (a) short-duration and (c) long-duration rainfall events to total ones. The PA in shaded colors is overlaid by the PF in black isograms. Their mean diurnal phases are represented by unit vectors in (b) and (d) correspondingly with brown lines separating the N, C, and S sub-regions.

4. Relationship between rainfall duration and diurnal variation

Rainfall duration is also known as a key factor to separate out rainfall diurnal features over central-eastern China ( Yu et al., 2007b ). In this section, the relationship between rainfall duration and diurnal variation over Xinjiang is analyzed.

The spatial distributions of the percentage of rainfall events with different durations to all rainfall events in the warm season between 2008 and 2019 in terms of PA and PF can be seen in the left-hand panels of Fig. 5 . More than 60% of rainfall events over the whole of Xinjiang last less than 3 h. These short-duration rainfall events contribute more than 50% of the PA to the precipitation over N and S, while it is less than 30%over C. Fig. 5 (c) shows that large values of both PA and PF for longduration events occupy C rather than N or S. This regional difference may explain why there is only one primary peak in the diurnal cycle of precipitation over the whole of Xinjiang and C while there are more than one peak in N and S. To verify whether the evening peak in the three sub-regions mainly results from long-duration rainfall events, Fig. 5 (d)shows the spatial distribution of the mean diurnal phase of precipitation for long-duration events. It is found that the long-duration rainfall events exhibit a near uniform morning peak over the three sub-regions.Fig. 5 (b) reveals that there are 3 h differences between the mean diurnal phase of precipitation for short-duration events over N and S, and this leads to the opposite variations of N and S before dawn. According to previous studies ( Yu et al., 2007b ; Chen et al., 2010 ), short-duration rainfall over eastern China usually peaks in late afternoon, but shortduration rainfall over arid and semi-arid regions mainly peaks around local noon. For long-duration rainfall, diurnal peaks between 0000 and 0600 LST were found over eastern China ( Fig. 2 in Yu et al., 2007b ).Over most areas of C and S, the peaks of the long-duration rainfall are similar to those in eastern China. Over the northern and southern margins of Xinjiang, the peaks of long-duration rainfall are in the morning,which are about 4—6 h later.

To see the quantitative contributions of different sub-regions to the whole of Xinjiang with respect to rainfall events of different duration,Fig. 6 shows the percentage of both PA and PF in solid and dotted lines respectively. PF shows similar results to PA. The curves divide at around 6 h. Rainfall events lasting more than 6 h happening over C take the upper fork, which means long-duration events in the whole of Xinjiang mostly happen over C. Therefore, the characteristics of diurnal variations over the whole of Xinjiang result from two separable kinds of rainfall events that dominate over different sub-regions: the long-duration events over C are responsible for the primary peak in the diurnal cycle of precipitation, while the short-duration events over N and S are responsible for other characteristics in the diurnal variations of warm-season precipitation.

Fig. 6. Percentage of rainfall events with duration ranging from 1 to 24 h that happened over N (blue), C (green), and S (red) to the whole of Xinjiang. Solid lines represent PA and dotted lines represent PF.

5. Summary and discussion

Because of the ecological and social importance of Central Asia,which consists of many arid and semi-arid land areas, the diurnal variation of precipitation and its differences to that in monsoon regions deserves further analysis. Xinjiang Province, with complex topography but relatively abundant data availability, was chosen as an example to investigate the characteristics of diurnal variations in warm-season precipitation over Central Asia. With the increase in availability of extensive and quality-controlled observational data, it has become possible to analyze the hourly precipitation features over northwestern China in detail. The present study serves as a supplement to previous extensive studies over other parts of contiguous China. With consideration of the complex topography, this study investigates the diurnal cycle of precipitation in three sub-regions in Xinjiang based on an hourly gauge—satellite merged precipitation product with a spatial resolution of 0.1°× 0.1° during 2008 to 2019. Compared with the diurnal peaks revealed by the national stations of Xinjiang Province, the contrast between the northern and southern parts of sub-region N is verified by the station rain-gauge data. The precipitation amount and frequency are found to correspond well with terrain height over Xinjiang. Meanwhile, large diurnal variations with distinctive regional features can be seen. Different from the late-afternoon maximum over monsoon regions, rainfall events over N and C have a dominant peak around early evening, while there is a secondary maximum over N before noon. The diurnal cycle of precipitation over S is noticeably different, having three peaks with a nocturnal maximum.

Similar to the results in eastern China, the rainfall duration is also a key to separate the diurnal features of rainfall events of Xinjiang. It is found that: (1) short-duration rainfall events occur more frequently over the whole of Xinjiang and contribute more than half of the precipitation amount over N and S; (2) long-duration rainfall events over C are responsible for the primary peak in the diurnal variations of warm-season precipitation in Xinjiang; (3) rainfall events with duration less than 6 h over N, C, and S contribute almost equally, while long-duration rainfall events over C dominate compared to those over N and S.

Additional work is needed to investigate the applicability of the characteristics in Xinjiang to other arid and semi-arid regions in central Asia.The diurnal variation of rainfall is known to be influenced by not only weather systems scaled from monsoon systems, the western Pacific subtropical high to nocturnal low-level jets, but also mesoscale convective systems and multi-scale mountain—valley breezes. Studies of the relationships between the diurnal variations and large-scale atmospheric circulations are ongoing. A detailed analysis of the mechanism of either short- or long-duration rainfall events in Xinjiang is beyond the scope of this study.

Funding

This work was supported by the National Key Research and Development Project of China [Grant Nos. 2018YFC1507104 and 2018YFC1507603 ], the National Natural Science Foundation of China[Grants Nos. 91937301 , 41875074, and 41675060 ], and the National Key Scientific and Technological Infrastructure Project “EarthLab ”.

Acknowledgments

The computing for this project was performed at the OU Supercomputing Center for Education & Research (OSCER) at the University of Oklahoma (OU), USA.