Classification and mechanism of spring and summer floods in northern Xinjiang from 2006 to 2011

The significant socioeconomic impact of extreme flooding provides an incentive to improve our understanding of flood drivers. In this study, floods that occurred in northern Xinjiang from 2006 to 2011 were divided into three categories: rainstorm‐type, warming‐type, and mixed‐type. These three types of floods primarily occurred from April to July, with most occurring in May and June. Through analysis of the atmospheric circulation evolution process of the three types of floods, it can be concluded that when a rainstorm‐type flood occurs, northern Xinjiang is affected by an anomalous cyclone that forms in front of the strengthened trough over northern Europe. Anomalous cyclones provide favorable conditions for precipitation, which is conducive to rainstorm‐type floods. As for the warming‐type flood event, northern Xinjiang is affected by an anomalous anticyclone formed by the eastward movement of the blocking system in the middle of the Eurasian continent. Before the third type of mixed flood event occurred, northern Xinjiang was affected by an anomalous cyclone formed by energy propagation along the northwesterly wind belt. In addition, the energy propagating along the westerly wind belt along the southern road is conducive to the formation of a high‐pressure ridge in southern Xinjiang. In addition, the analysis of temperature conditions indicates that the daily maximum temperature showed a warming trend from 5 to 1 day before the warming‐type and mixed‐type flood event occurred. These results provide valuable insights for flood risk management by identifying atmospheric circulation patterns and temperature conditions associated with floods in northern Xinjiang.

and marine disasters (Liu et al., 2010).With global warming and human development, the risk of flooding is projected to increase and cause severe losses to the national economy, the lives of humans, and property in the future (Alfieri et al., 2017;Hirabayashi et al., 2013;Wang & Wang, 2021).These facts highlight the need to understand flood risk trends.
Several tall mountains in Xinjiang intercept more water vapor and are conducive to the formation of precipitation and snow in the mountainous areas, among which the rainstorm in the middle and low mountains in spring and summer is easy to cause flood disasters (Lyu et al., 2021).In addition, the glacial area of Xinjiang accounts for 44% of the total glacial area of China.In the context of global warming, the accelerated melting of glaciers caused by climate warming has significantly increased the frequency of floods in Xinjiang over the past 30 years (Ablikim et al., 2015;Dong et al., 2004;Shang et al., 2016;Shen et al., 2013;Wu & Zhang, 2003).According to the China Water Statistics Yearbook, Xinjiang has over 3484 rivers covering an area of ≥50 km2 , with a total length of 138,961 km, ranking fourth in number and third in total length among the Chinese provinces.Among them, the Irtysh River in northern Xinjiang is the only water system that injects into the Arctic Ocean, and the Irtysh River is located in the Altai Mountain which has a rich amount of snow and ice (Gao et al., 2022).The valley of the Irtysh River in the Altai Mountain region is conducive to the entry of westerlies, and the water vapor entering the region is intercepted and lifted by the Altai Mountain, forming the abundant precipitation zone on the south slope of the Altai Mountain (Pang et al., 2023).In addition, due to the high latitude, the Altai Mountain is often affected by the strong cold air moving east from Siberia and the Ural Mountains in winter and spring, resulting in heavy snowfall here (Ren et al., 2023).From November to March of the next year, in the Altai Mountains at an altitude of 1500-2400 m, there is a large amount of snow and a long duration of snow.In spring, if there is an obvious warming process or a strong precipitation process, it is easy to cause flood events, which poses a serious threat to farmland, water conservancy facilities, people's lives, and property in the downstream areas of the basin.Therefore, studying the temporal distribution, occurrence rules, and meteorological causes of floods, and exploring the predictability of such disasters is crucial for flood prevention and mitigation in northern Xinjiang (Ablikim et al., 2015;Mao et al., 2012).
Generally speaking, on the basis of analyzing the meteorological causes of flood events, applying meteorological elements forecast to hydrological forecast is an effective method to improve the forecast of the flood.Due to the different climate backgrounds of different types of flood, the forecasting methods and indicators of flood events are also different.It is necessary to study the different types of floods and their mechanisms to improve the understanding of flood disasters in arid areas.Some studies have focused on major tributary flood processes, flood classification, and individual cases of major flood disasters in Xinjiang (Ruan, 2006;Zhang, 2012;Zhang et al., 2021;Zhang, Luo, & Wang, 2021).In general, two types of snowmelt floods occur in the northwest arid areas of China: warming snowmelt floods and rain-snow mixed floods (Chen et al., 2021).The former primarily occurs in northern Xinjiang and is concentrated from mid-March to early April.The latter occurs primarily in the Qilian, Kunlun, and Tianshan Mountains and is concentrated in May and June (Chen et al., 2021).According to the analysis of AihaitiÁ (2015), four different types of floods occur in the Yeerqiang River, namely glacier and snow melting type, glacier dam break type, rainstorm type, and mixed type flood.Xie (2014) indicated three types of floods coexisting with different causes in the Tizhnaphu River: ice and snow melting flood; rainstorm flood; and mixed storm and ablation flood.These studies highlighted the importance of temperature changes and precipitation during flood formation (Liu et al., 2020;Qiu & Yan, 1994;Wei, 2006).Therefore, determining the typical large-scale weather characteristics, temperature, and precipitation changes related to local floods is crucial for the classification and forecasting of floods.For example, Zhang (2012) demonstrated that precipitation in the upper reaches of the Toshigan River Basin was higher than normal before the snowmelt flood and also determined the occurrence time and intensity of warm river basins using a path analysis method.The Iranian subtropical high eastward expansion type, Western Pacific Subtropical high westward extension type, and Xinjiang ridge development type are the three atmospheric circulation types of snowmelt (ice) floods in Xinjiang (Zhang et al., 2021;Zhang, Luo, & Wang, 2021).
These studies provide valuable insights into the types of spring and summer floods and their meteorological conditions.Nevertheless, deficiencies remain in analyzing and recognizing flood characteristics and quantitatively revealing the development and evolution of meteorological conditions.The remainder of this article is organized as follows: first, we describe the data and methods used in the study and then analyze the meteorological and hydrological characteristics of different types of spring and summer floods.Second, we present concluding remarks.

| Data
Similar to several other counties in Xinjiang, field data on the extent, timing, and depth of flooding in northern Xinjiang is generally limited because of the scarcity of detailed ground-based observations and is rarely available for large flood events (Liu et al., 2021).We collected data on flood disaster time at the Kuwei hydrological station from 2006 to 2011.However, there are no meteorological observation stations at the Kuwei hydrological station.Fuyun meteorological station is the closest to the Kuwei hydrological station (Figure 1).Therefore, the daily stationbased precipitation and temperature datasets for 2006-2011 that were collected at Fuyun the Chinese Meteorological Administration station were used to analyze the meteorological background of the floods.This dataset was provided by the National Climate Center of the Chinese Meteorological Administration.Daily mean atmospheric variables with a horizontal resolution of 1 Â 1 were extracted from ERA5, the latest European Centre for Medium-Range Weather Forecasts reanalysis product (Hersbach et al., 2020).

| Methods
The wave activity flux calculated based on the method proposed by Takaya and Nakamura (2001) was applied in this study to illustrate the stationary waves related to extreme flood meteorological conditions over northern Xinjiang.

| Classification of floods in northern Xinjiang
Based on daily streamflow data for northern Xinjiang from 2006 to 2011, 28 flood events were selected.
The occurrence times of the 28 flood events in northern Xinjiang from 2006 to 2011 and the average precipitation and temperature changes in the first 5 days are listed in Table 1.The average precipitation in the 5 days before the flood event varied from 0 to 6.86 mm/day, and the temperature change varied from positive to negative, with the minimum being À9.6 C and the maximum being 8.6 C. In general, flood events can be divided into three categories according to the two indices of average precipitation and temperature change: rainstorms floods, ice and snowmelt floods (warming type), rainstorm and warming mixed floods.For the first type (rainstorm-type floods), the average precipitation in the first 5 days was positive, but the temperature change was negative.For the second type (warming-type floods), there was no precipitation in the first 5 days of the flood; however, there was significant warming (positive temperature change).For the rainstorm and warming mixed floods, both the average precipitation and temperature change in the first 5 days of the flood were positive.As given in Table 2, of the 28 flood events from 2006 to 2011, 7 rainstorm flood events, 16 warming-type flood events, and 5 rainstorms and warming mixed flood events occurred.

| RESULTS
The seasonal characteristics of three types of flood events are first analyzed.In northern Xinjiang, floods regularly occur in May and June (11 times) but less frequently in April and July (4 and 2 times, respectively; Figure 2).The rainstorm-type floods occurred from April to July, with the highest frequency occurring in May.Both the warming-type and mixed-type floods occurred from April to June, with the highest frequency of warming-type floods occurring in June.The seasonal characteristics of different types of floods may be related to the seasonal characteristics of temperature change which is 6.4, 6.5,  Analyzing the evolution process of the atmospheric circulation of the three flood events is helpful for finding early signals of flood events.Figure 3 shows the circulation patterns associated with each flood type.Before the occurrence of the first type of flood event, there was a weak "+ À +" wave train structure from Europe to Xinjiang at 500 hPa (Figure 3a).From 2 days before the outbreak to the day of the outbreak, the trough in northwest Xinjiang continuously developed, deepened, and squeezed to the east (Figure 3b).Anomalous cyclone formed in front of the trough and occurred over northern Xinjiang on the day of the flood event (Figure 3c).The anomalous cyclone in the middle and high latitudes is conducive to the occurrence of heavy precipitation events, that can easily cause rainstorm flood events (Henn et al., 2020).Before the warming flood event, a high-pressure ridge developed over northern Xinjiang, and the positive geopotential height anomaly continuously strengthened and extended to the east (Figure 3d,e).On the day of the flood event, an anomalous anticyclone occurred in northern Xinjiang, and the airflow descended and heated, which then lead to snowmelt flooding (Gelfan, 2010; Figure 3f).Four to two days before the occurrence of the third type of mixed flood   event, Eurasia was affected by a low-pressure trough, and northern Xinjiang was located in the middle of the blocking anomaly, resulting in a temperature increase (Figure 3g,h).However, from 2 days before the flood to the same day, the low-pressure trough in northern Xinjiang deepened continuously, forming an anomalous cyclone, which was conducive to the occurrence of heavy precipitation events, and then led to mixed floods (Figure 3i).
To analyze the evolution mechanism of these three types of flood events in northern Xinjiang from the perspective of dynamics, the T-N wave flux diagnosis was used to analyze the wave activity process from 4 days before the outbreak of these three types of flood events on the same day (Figure 4).During the three types of flood events, the transient wave activity primarily fluctuated along the northward-westerly and southwardwesterly directions.The transient wave train regulated the long-wave activity of the northward westerly wind, which primarily manifested as the regulation of the evolution of the trough ridge.In particular, the meridional component of the direction of the transient wave flux, namely the direction of energy propagation, plays a crucial role in regulating the strength and evolution of trough and ridge systems.Before the rainstorm-type flood event occurred, the disturbance over northern Europe strengthened rapidly, and the energy continued to spread downstream, causing weak disturbance in central Europe to move eastward and develop into a shallow trough (Figure 4a,b).On the day of the flood event, northern Xinjiang was abnormally affected by cyclone circulation (Figure 4c).During the formation of the warming-type flood event, the blocking system was located in the middle of the Eurasian continent, and the center was located north of Balkhash Lake (Figure 4d,e).The blocking system was in a stable position, and the energy transport on its west side caused the blocking to continuously develop, strengthen, and spread to the southeast, forming a positive geopotential high disturbance over the Xinjiang region on the day of the flood event (Figure 4f).The wave activity process of the mixed-type flood events differed from that of the other two types.Four to two days before F I G U R E 3 Composites of 500 hPa geopotential height (contours; units: da gpm) and its anomaly (color shading; units: gpm) for (a,d,g) Day 4, (b,e,h) Day 2, and (c,f,i) Day 0 relative to flood events in northern Xinjiang over the period 2006-2011: (a-c) the rainstorm-type flood events; (d-f) the warming-type flood events; (g-i) the mixed-type flood events.Stippling denotes where the anomalies are significant at the 90% confidence level based on the Student's t-test.
the flood, the disturbance over the central Eurasian continent continued to receive energy transport from northern Europe, which caused a negative disturbance over the Siberian plain to continue to deepen, and a trough of low pressure in the central Eurasian developed (Figure 4g,h).In addition, some energy fluctuated along the westerly belt along the southern road, causing a weak positive disturbance in eastern Xinjiang 4 days before the flood (Figure 4g).This disturbance continued to strengthen eastward and moved to Xinjiang 2 days before the flood (Figure 4h).On the day of the flood event, a positive disturbance developed and moved to southeastern Xinjiang (Figure 4i).The final energy transfer resulted in a strong low-pressure trough over northern Xinjiang and a weak high-pressure ridge over southern Xinjiang.
By analyzing the evolution of the wave activity process of these three types of flood events, it was found that rainstorm-type and mixed-type flood events were primarily affected by the low-pressure northern trough.The difference between the two is that mixed flood events have energy that fluctuates along the westerly wind belt on the south road in the early stage of the outbreak, resulting in a weak positive disturbance in eastern Xinjiang which strengthen and move eastward.The positive disturbance located in Xinjiang 2 days before the outbreak and caused process warming.When a warming flood event occurs, it appears as a positive geopotential height disturbance over the Xinjiang region.
In addition to analyzing the meteorological dynamic conditions of the three types of floods, the water vapor conditions related to precipitation are also a crucial part of rainstorm-type and mixed-type floods.There was strong northerly water vapor transport over northern Xinjiang from 4 to 2 days before the occurrence of rainstorm-type flood event and mixed-type flood event, providing sufficient water vapor transport for precipitation (Figure 5a,c), whereas the water vapor transport was weak from 4 to 2 days before the occurrence of the warming-type flood event (Figure 5b).

| DISCUSSION
In addition to analyzing the dynamic process, we aimed to analyze in detail the thermal conditions of three types of flood.The evolution process of the daily maximum temperatures from 5 days before to 5 days after three types of flood was analyzed (Figure 6).In the three types of flood, the daily maximum temperature showed an increasing trend from 5 to 1 day before the occurrence of warming-type flood and mixed-type flood (Figure 6b,c), while the temperature of rainstorm-type flood event showed a decreasing trend (Figure 6a).Combined with the evolution process of circulation anomalies above (Figure 4), this was because the weak positive-potential height anomaly gradually developed into a strong positive-potential height anomaly over the northern F I G U R E 4 Composites of 500 hPa perturbing stream function (shading), 500 hPa T-N wave activity flux (arrows; units: m 2 Ás À2 ) for (a,d, g) Day 4, (b,e,h) Day 2, and (c,f,i) Day 0 relative to flood events in northern Xinjiang over the period 2006-2011: (a-c) the rainstorm-type flood events; (d-f) the warming-type flood events; (g-i) and the mixed-type flood events.Stippling denotes where the anomalies are significant at the 90% confidence level based on the Student's t-test.
region of Xinjiang due to the continuous transmission of upstream energy before the occurrence of the warming-type flood event.As for the mixed-type flood event, the warming may be related to the weak positive geopotential height anomaly in northern Xinjiang 2 days before the flood.Due to the positive correlation between surface air temperature and geopotential height (Zhu et al., 2008), there are significant warming before mixed-type flood and warming-type flood.In addition, the local temperature was also affected by local advection and radiation.Figures S1 and S2 show advection and radiation changes during the early stages of the warming-type flood and mixed-type flood.Strong warm advection was observed in northern Xinjiang from 4 to 2 days before the warming-type and mixed-type flood event, which provided conditions for temperature increase (Figures S1a,b and S2a,b).
Regarding net radiation, strong positive radiation anomalies occurred in northern Xinjiang from 4 to 2 days before the warming-type and mixed-type flood event, which was also conducive to local warming (Figures S1d,e and S2d,e).

| CONCLUSIONS
Floods are a crucial disaster type often caused by rainstorms and/or water from snowmelt.This study attempted to analyze the atmospheric circulation background and dynamic and thermal conditions of three types of floods in northern Xinjiang.Floods in northern Xinjiang primarily occur from April to July and can be divided into three types: rainstorm-type floods, warmingtype floods, and mixed-type floods (Table 3), among F I G U R E 6 Evolution of the daily maximum surface air temperature 5 days before and after the occurrences of rainstorm-type (a), warming-type (b), and mixed-type (c) flood events in northern Xinjiang.which warming-type floods account for the largest proportion.
Through analysis of the atmospheric circulation evolution process of the three types of floods, it can be concluded that when a rainstorm-type flood occurs, the disturbance over northern Europe strengthens rapidly, and the energy spreads downstream.The low-pressure trough in northwestern Xinjiang deepened and squeezed eastwards.An anomalous cyclone formed in front of the trough and occurred over northern Xinjiang, which is conducive to the occurrence of rainstorm-type flood.For the warming-type flood events, the blocking system in the middle of the Eurasian continent was in a stable position, and the energy transport on its west side made the blocking develop, strengthen, and spread to the southeast.On the day of the warming-type flood event, northern Xinjiang was affected by an anomalous anticyclone.Warming-type floods occur because of airflow subsidence and heating.Before the third type of mixed flood event occurred, the disturbance over the central Eurasian continent continuously received energy transport from northern Europe, which caused the low-pressure trough in northern Xinjiang to deepened continuously, and an anomalous cyclone was formed, which was conducive to the occurrence of heavy rainfall.In addition, the energy propagating along the westerly wind belt on the southern road was conducive to a weak positive disturbance in eastern Xinjiang, which strengthened and shifted eastward.Finally, southern Xinjiang is affected by the highpressure ridge, which is conducive to temperature rise and causes mixed-type floods.Also, the temperature conditions of warming-type flood were analyzed.From 5 to 1 day before the warming-type and mixed-type flood event, the daily maximum temperature showed a warming trend.Simultaneously, there were strong warm advection and positive radiation anomalies in northern Xinjiang from 4 to 2 days before the warming flood event, which was conducive to local warming.
Overall, this study highlights the mechanisms of different types of floods in northern Xinjiang.With more large-sample hydrometeorological datasets becoming readily accessible, the next step is to extend the research to a larger scale and longer period to better understand of variations in flooding mechanisms.In addition, many studies have focused on the influence of uppertropospheric circulation systems and various teleconnections on the climate of Xinjiang, such as South Asian high, Asian westerly jet, circumglobal teleconnection pattern, ENSO, and NAO (Chen et al., 2010;Huang et al., 2013;Huang et al., 2015;Taschetto et al., 2020;Yeh et al., 2018;Zhang, Huang, et al., 2021;Zhang, Luo, & Wang, 2021).Note that the teleconnection pattern, which affects the rainstorm-type flood in spring (Figure 4c), is reminiscent of the circumglobal teleconnection pattern found in previous studies (Branstator, 2002;Ding & Wang, 2005).However, the relationship between uppertropospheric circulation systems and various teleconnection patterns and floods in northern Xinjiang has not been well established, which deserves further study.Also, the circumglobal teleconnection pattern has undergone decadal changes in the past (Wu et al., 2016), which may have a certain impact on the decadal changes of flood in northern Xinjiang, and relevant studies can be conducted based on longer periods in the future.
Location of Fuyun meteorological station (blue) and Kuwei hydrological station (red), the shadings indicate the terrain height, the light blue lines indicate the river in northern Xinjiang.
and 2.0 C between April-May, May-June, and June-July during 2006-2011.
Average precipitation and temperature change in the first 5 days of 28 flood events in northern Xinjiang from Proportion of three types of flood events in different months (orange: rainstorm-type floods; blue: warmingtype floods; green: mixed-type floods).

F
I G U R E 5 Composites of integrated water vapor transport (200-700 hPa; arrow; kgÁm s À1 ) for Day 0 relative to three type flood events in northern Xinjiang over the period 2006-2011: (a) the rainstorm-type flood events; (b) the warming-type flood events; (c) the mixed-type flood events.Shading denotes where the anomalies are significant at the 90% confidence level based on the Student's t-test.
Three types of flood events (rainstorm floods, warming-type floods, and rainstorm and warming mixed floods).