Spatial and temporal evolution characteristics of meteorological drought in the Northwest of Yellow River Basin and its response to large-scale climatic factors

Drought is a major natural cause of agricultural, economic, and environmental devastation (Cook et al. 2014). According to the Intergovernmental Panel on Climate Change (IPCC) assessment report, global warming has resulted in a significant increase in global drought events over the last century (Masson-Delmotte et al. 2021). Furthermore, drought is one of the most destructive climatic phenomena, which can occur in almost all climatic regimes (Band et al. 2022).

Extreme drought events have had a significant impact on water resources, ecological water environment, and agricultural production in many regions of China (Mishra & Singh 2010; Esfahanian et al. 2017). Drought will undoubtedly occur after prolonged periods of no precipitation, but its occurrence, extent, and duration are difficult to predict. As a result, objectively quantifying their characteristics is difficult. The drought happening in most parts of the world depends mostly on lack of precipitation and increased temperature profoundly (Shamshirband et al. 2020). Arid and semi-arid regions are more sensitive to climate change (Huang et al. 2016a). As the global climate continues to warm, droughts in these places are anticipated to grow more frequent and severe (Dai 2013). This implies that these areas will experience more difficulties.

Hydrological drought results in a water deficit with river runoff insufficient to meet water supply needs in a given period. A drought index is an effective way to monitor hydrological drought (Mishra & Singh 2011). These dry and semi-arid areas have been the subject of much research and monitoring of drought conditions. Kousari et al. (2014) used the Reconnaissance Drought Index (RDI) to research the trend of drought severity in the areas mainly covered by arid and semi-arid climate conditions in Iran. The results show that the decreasing trend of RDI time series means the worsening of drought severity (Lashkari et al. 2021) investigated the effects of changes in precipitation and temperature on spatio-temporal drought and humidity variations throughout the diverse (principally arid- and semi-arid) climates during recent decades using the Pedj Drought Index (PDI). The results show that droughts were more frequent than humidity events in the study area in the past, and that temperature warming was the main driver of severe droughts in history. Like many other areas in arid or semi-arid climates, the Northwest of Yellow River Basin suffers significant drought-related issues. This area's environment has changed to become drier and warmer, and as a result, the risk of drought is predicted to increase.

In studies on drought analysis and monitoring systems in the Yellow River Basin, most scholars have employed the Palmer drought severity index (PDSI; Palmer 1965) or the standardized precipitation index (SPI; McKee et al. 1993). Some researchers examined the drought detection adaptability of various satellite precipitation products using the SPI index, as well as the hydrological drought in the Yellow River Basin (Liu et al. 2015; Wei et al. 2019). The results show that SPI has good application in drought monitoring. Although the SPI includes multi-scale properties, it solely examines precipitation and does not account for the impact of temperature and evapotranspiration on regional dry and wet conditions. PDSI takes rainfall and evapotranspiration into consideration, but PDSI has a fixed time scale and therefore cannot be used to assess types of droughts that occur at different time scales (Musei et al. 2021). As a result, the SPI and PDSI have several flaws in terms of monitoring the area's dry and wet conditions. The Standardized Precipitation Evapotranspiration Index (SPEI) is more flexible because it is based on the SPI calculation but combines precipitation and evapotranspiration, thus, making up for the shortcomings of PDSI's fixed time scale and numerous required parameters (Tan et al. 2015). Currently, SPEI has been widely used in regional dry and wet weather monitoring (Vicente-Serrano et al. 2010). Some researchers employed the SPEI in conjunction with the Vegetation Condition Index (VCI) to examine the transfer link between agricultural drought and hydrological drought time, as well as the frequency and intensity of drought based on grid division (Wang et al. 2018; Hao et al. 2021; Cao et al. 2022). The results showed that SPEI and VCI could effectively reveal the change law of meteorological drought and agricultural drought (Liu et al. 2016).

Changes in global temperature and atmospheric circulation, such as the El Niño-Southern Oscillation (ENSO; Rasmusson & Wallace 1983), Atlantic Oscillation (NAO; Hurrell & Loon 1997), Pacific Decadal Oscillation (PDO; Mantua & Hare 2002), and Arctic Oscillation (AO; Thompson & Wallace 1998), have been shown in previous studies to play a major role in drought formation (Mishra & Singh 2010; Dai 2011). Exploring the evolutionary link between these factors would aid current knowledge of drought changes in the research area. As a result, it is critical to investigate the impact of macroclimate factors on regional drought, which has a significant impact on drought warning and water resource management. However, it is still unclear what atmospheric circulation index most affects the Northwest of Yellow River Basin. In this research area, there are also not many investigations on the connections between atmospheric circulation parameters and drought. Therefore, cross-wavelet analysis and cross-wavelet coherence analysis are used in this study to reveal the correlation between SPEI and large-scale factors as well as their common energy region and phase relationship.

The majority of research in the Northwest of Yellow River Basin on drought-related features focuses on the effect of rainfall on wet and dry conditions. However, with rising temperatures and, consequently, increased reference evapotranspiration, declining precipitation and severe climate change are putting more pressure on water resources and agriculture (Byakatonda et al. 2018); therefore, decision makers must have accurate information about the amount of evapotranspiration and drought risk assessment for each region (Sharafi & Ghaleni 2022). Therefore, the selection of climate parameters that affect the accurate evaluation of evapotranspiration is of great significance for the accurate evaluation of drought indicators. Moreover, the accuracy of evapotranspiration is so important that it could even provide important evidence for identifying models that more accurately estimate drought under climate change conditions. To determine monthly potential evapotranspiration, most studies (Wang et al. 2020; Cao et al. 2022) employ the Thomthwaite (Thornthwaite 1948) and Hargreaves–Samani model (Hargreaves & Samani 1985). This method, on the other hand, ignores the impact of meteorological conditions. This study uses temperature, rainfall, solar radiation, wind speed, and other variables to discuss 52 meteorological stations in the Northwest of Yellow River Basin, based on previous research. Furthermore, the multi-scale SPEI analysis is based on the Penman–Monteith (P-M) model, and employs Morlet wavelet analysis (Fengying 1999), dry area coverage, drought frequency, Mann–Kendall sudden change examination (Gocic & Trajkovic 2013) to conduct a qualitative and quantitative analysis of the region's spatial and temporal distribution of drought. The association of SPEI with large-scale climate conditions is also discussed in this study by cross-wavelet analysis and wavelet coherence analysis. These researches can help with drought evaluation and forecasting in the study area.

The Northwest of Yellow River Basin has high terrain in the west and low terrain in the east. The source area in the west has an average elevation of more than 4,000 m. The topography is dominated by plateaus, covered by glaciers and snow. Most areas receive between 200 and 650 mm of precipitation per year, with an average annual temperature of 12–14 °C. Summer temperatures are high, and precipitation decreases gradually from east to west; winter is cold and dry, water resources are scarce, and the ecological environment is fragile. Extreme drought events are common in this region as a result of human activity and climate change. Affected by geographical location and East Asian monsoon climate, hydrometeorological conditions in the Yellow River Basin are complex and changeable. Due to water scarcity and an uneven distribution of precipitation, the Yellow River Basin is one of China's drought-affected regions (Huang et al. 2015). Consequently, it is crucial to monitor the drought in the Yellow River Basin. The production, life, and environment of this region have all been severely impacted by the trend of drought intensity and persistence in recent years (Wang et al. 2018). The Northwest of Yellow River Basin is located in the Northwest of China. The region is arid and semi-arid, which is more susceptible to drought and has the basic characteristics of drought.

The relationship between SPEI and large-scale oceanic atmospheric circulation in the research area, including ENSO, NAO, PDO, and AO, is covered in this article. Multivariate ENSO Index Version 2 (MEI.v2) is the multivariate ENSO index, which is a novel version of the conventional Multivariate ENSO Index (MEI) index. On top of the classic MEI index, it is derived using six variables as proxies for ENSO-related atmospheric and oceanic conditions, and it uses OLR data of monthly output longwave radiation (OLR) from the NOAA Climate Data Record (CDR). The data time period for the MEI.v2 index is from 1979 to 2020, hence, the MEI.v2 index in this article only looks at data from 1979 to now. The other three indices are following the SPEI time series, from 1961 to the present. The data of ENSO, NAO, and PDO data is downloaded from NCAR (National Center for Atmospheric Research) (ucar.edu). AO data is collected from NCEI (National Center for Environmental Information) (noaa.gov).

The SPEI is used in this study to evaluate the characteristics of interdecadal, interannual, and seasonal droughts in the study area, as well as drought area coverage and frequency. It employs spatial interpolation to investigate the spatial and temporal aspects of drought in the Northwest of Yellow River Basin during the last 60 years, as well as Morlet wavelet analysis and Mann–Kendall sudden change examination to investigate drought trends and cycle characteristics. Using crossed wavelet power spectrum (Rioul & Vetterli 1991) and coherence spectrum (Torrence & Compo 1998), the impact of significant climatic influences on regional drought is investigated.

In this formula, is canopy net radiation (MJ·m²·d⁻¹), and ⁠, respectively, are saturation pressure and actual vapor pressure (kPa), T is average temperature (°C), G is soil heat flux (MJ·m²·d⁻¹), γ and are hygrometer constant (kPa·°C⁻¹) and wind speed at 2 m height (m·s⁻¹), respectively, is the slope of saturated water vapor pressure vs temperature curve (kPa·°C⁻¹).

In this formula, is scale parameters, and ⁠, respectively, are morphological parameters and initial parameters.

The drought standard (Wang et al. 2011) is shown in Table 1.

The SPEI values of 3-month and 12-month time series are analyzed in this study. The four seasons are defined by SPEI_3 and are March–May, June–August, September–November, and December–February, respectively. SPEI_12 characterizes the year scale. SPEI_3 represents the short-term impact of water scarcity on drought, whereas SPEI_12 represents the interannual fluctuation features.

The Mann–Kendall examination method (Gocic & Trajkovic 2013) is a rank-based nonparametric trend test method. It does not require samples to conform to a certain distribution law, and the change trend is not need to follow the linear law. It is unaffected by a few outliers, and results are unaffected by partial missing data. The Mann–Kendall examination method can not only test the change trend of time series, but also test whether the time series has a sudden change.

Climate factors will show periodic changes, so exploring the periodic changes of climate factors plays an important role in climate prediction and prevention of climate disasters. Wavelet analysis (Fengying 1999) is used to explore the periodic change law of climate elements, it not only can count the frequency components of sequence signals effectively, but also the frequency distribution in time domain can be located. Cross-wavelet analysis (Rioul & Vetterli 1991) combines the characteristics of wavelet transform and cross-spectrum analysis. Compared with Fourier transform, cross-wavelet analysis can better reflect the time–frequency domain variation characteristics and coupling oscillations of two time series (Hudgins et al. 1993). The cross-wavelet transform cannot reveal the low energy region of the two time series in the time–frequency domain, but the cross-wavelet coherence (Torrence & Compo 1998) analysis offsets this limitation (Adamowski & Prokoph 2014). Therefore, cross-wavelet analysis and cross-wavelet coherence analysis are used in this study to reveal the correlation between SPEI and large-scale factors as well as their common energy region and phase relationship. The detailed calculation method is given in Grinsted et al. (2004). In this study, Morlet is used as a complex-valued wavelet, because complex-valued wavelet has imaginary part, which can express the phase well, and Morlet wavelet is not only non-orthogonal, but also an exponential complex-valued wavelet regulated by Gaussian, which can have a good balance between the localization of time and frequency. The periodic variation law of SPEI_3 and SPEI_12, as well as the multi-time scale characteristics, are analyzed using the wavelet real part contour map and wavelet variance map of complicated Morlet wavelet analysis to disclose the cyclical characteristics of drought.

Figure 3(b)–3(e) depicts the seasonal fluctuations in SPEI_3 and their M-K test curves. In spring, SPEI_3 has a linear trend rate of −0.05 every 10 years. The SPEI_3 value in 1967 was 1.67, which was the rainy spring in the preceding 60 years and reached a slightly humid level. The SPEI_3 reading in 2000 was −1.34, indicating considerable dryness and the driest spring in 60 years. SPEI has a linear trend rate of −0.02 every 10 years in the summer. In 1979, the SPEI_3 value was 0.98, the wettest summer in the previous 60 years, with a somewhat humid level. In 2018, the SPEI_3 rating was −1.07, indicating moderate dryness, and it was the driest summer in the past 60 years. In the autumn, the SPEI linear trend rate was −0.02 every 10 years. The SPEI_3 values in 1961 and 1972 were 1.14 and −1.62, respectively, signifying the two wettest and driest autumns, with moderately humid and severely dry conditions. In the winter, the linear trend rate of SPEI was −0.05 every 10 years. In 1999, the SPEI_3 value was −1.36, while in 2008, it was 1.45. They were the wettest and driest years on record, with moderately humid and moderately dry conditions.

To summarize, the SPEI_3 value variations in the four seasons have been trending downward for the past 60 years. The SPEI_12 drought index is on the decline, and the drought situation is becoming increasingly dire. On a long time scale, the SPEI is less affected by precipitation than by temperature. The seasonal SPEI varies a lot, and it is easily influenced by precipitation and temperature fluctuations. The variations in trend are largely consistent, and the changes in trend slope become increasingly apparent around the year 2000.

To summarize, most locations in the Northwest of Yellow River Basin have had an increase in aridity during the last 60 years, while a minor portion has experienced an increase in humidification. The aridification sites are mostly in the north-central part of the country, and a higher percentage of them passed the significance test (0.01 level). Several stations in Qinghai Province, particularly Golmud and Nuomuhong, have been humidified to significant levels on different scales. In comparison to the other three seasons, the autumn drought trend is more noticeable, followed by summer and winter, while spring has the least noticeable drought trend.

Warming and humidification trends have been observed in desert parts of Northwest China from 1961 to 2018 (Zhang et al. 2021), which agrees with our findings. However, the current humidification trend reflects merely a change in amount, which is insufficient to modify the region's underlying climate state. It is still in the arid and semi-arid climate zones, as well as the temperate arid zone.

Figure 8(c) shows that the performance of the SPEI and NAO indexes in the power spectrum is nearly identical to that of the AO index. The NAO index exhibits a significant positive association with the SPEI, implying that it has a comparable impact on drought in the research area as the AO index. The SPEI and the PDO index, shown in Figure 8(e), exhibit two distinct and substantial resonance periods, 1967–1973 and 1989–1995, with vibration period signals of 14–23.5 and 38.3–51.5 months, respectively. The phase difference shows that the SPEI and the PDO index both have positive phase shifts during the substantial resonance period, but a negative correlation during the non-significant resonance period from 2007 to 2018. The high-energy parts of the power spectrum and the coherence spectrum are nearly identical, implying that the PDO index has an important effect on SPEI in these periods. However, the correlation changed in different periods, so the impact of PDO index on drought conditions was different in different periods. This may be due to the influence of human activities and the change of the underlying surface, so that it has undergone a stage change (Begueria et al. 2014). The relationship between these indicators during the various eras indicated above may undergo periodic modifications as a result of this changing pattern. In earlier years, PDO reduced drought conditions in the study area. However, in later years, PDO increased drought conditions. The SPEI and the MEI.v2 index are substantially correlated at the 95% confidence level, as shown in Figure 8, and there are two unique significant resonance periods. It showed oscillation cycle signals of 33.5–76.8 months from 1983 to 1999 and 22.4 to 48 months from 2004 to 2019. Its phase difference implies that during the resonance period, the SPEI and MEI.v2 index are in the negative phase. Figure 8 illustrates that high energy was exhibited as modest periodic fluctuations with durations of 0–6 months from 1995 to 2012. All of them revealed a strong negative connection, showing that MEI.v2 exacerbated drought conditions in the study area. This suggests that the ENSO phenomenon has exacerbated the region's drought. ENSO, PDO, AO, and NAO are all intimately linked to drought conditions in the research area and have a significant impact on the climate. ENSO had a greater link than the other three indices, showing that ENSO had a significant impact on the drought in the study area. ENSO is the most significant interannual oscillation signal among climatic components, according to certain studies, and it is the influence of air-sea coupling on climate (Webster et al. 1998; Ye & Wu 2018). This is in line with the finding that the ENSO event had a significant impact on the drought in the study area.

The Northwest of Yellow River Basin is an arid and semi-arid region. Low precipitation and rising temperatures increase moisture deficits and thus increase meteorological and hydrological droughts. Here, we use multiple aspects of drought to specify the severity of the drought, and the annual SPEI in the region shows a clear downward trend, with over 75% of the sites declining. The amount of drought in the study area did not significantly alter, but there was an increase in the frequency and severity of drought occurrences. Although there was little change in the study area's drought coverage, but drought events increased in frequency and severity. The total yearly precipitation trend in the studied area was insignificantly increasing (Guo et al. 2020). The occurrence of a local increase in precipitation in the Northwest of Yellow River Basin is insufficient to affect the current state of water scarcity in the studied area. According to some research, the positive vorticity advection anomaly over the region may be produced by the subtropical westerly jet's southern displacement in Asia, which causes the cyclone to migrate upward, resulting in higher precipitation (Peng & Zhou 2017). But water is still scarce in the research region. The Northwest of Yellow River Basin is located in the interaction zone of southeast monsoon and southwest monsoon. Precipitation and temperature changes are impacted by energy shifts in the Pacific Ocean. Due to the research area's large temperature increase brought on by global warming, the region's evapotranspiration likewise exhibits an increasing tendency, which causes a significant drop in SPEI (Figure 2). However, evapotranspiration has not shown a significant upward trend in the past 60 years, which indicates that evapotranspiration is not only affected by temperature, but also related to all parameters of the P-M model. Therefore, a more accurate evapotranspiration calculation model can comprehensively consider the influence of meteorological factors.

From 1961 to 2020, drought conditions in the Northwest of Yellow River Basin became more severe, and worsened after the 2000s. In recent years, the rise rate of evapotranspiration was significantly accelerated (5.29 mm/10a). As a result, increased warming and reduced precipitation in most areas have increased the severity of droughts. Related studies (Zhu & Chang 2017) show that the climate in recent 55 years is warmer and the precipitation fluctuates significantly, so the decrease of precipitation and the rise of temperature are the main reasons for the worsening of drought. Abnormal atmospheric circulation makes warm and cold air masses weak and unable to accumulate, which is one of the main reasons for the drought in the study area. In addition to the causes of climate change, human activities have also influenced the development of drought by altering the underlying surface of the Northwest of Yellow River Basin. A series of ecological and environmental problems, such as the over-exploitation of water resources, the decrease of vegetation coverage, and the decrease of groundwater level, will reduce the ability to resist drought and accelerate the formation of drought (Miao et al. 2016). Some scholars have shown that irrigation index, water fraction, and groundwater availability are the most important parameters for assessing drought vulnerability (Sahana et al. 2021). In addition, the excess greenhouse gases generated by rapid economic and social development also contribute to the development of drought (Bista et al. 2017). Therefore, under the specific natural geographical and climatic conditions, coupled with the influence of human activities, the drought in the Northwest of Yellow River Basin is becoming more and more serious. The research area has experienced substantially higher temperatures due to global warming (Masson-Delmotte et al. 2021), as well as dramatic melting and shrinking of glaciers, which has increased soil moisture content and river runoff (Liu 2006). These occurrences could explain the research area's growing wetness patterns. Although the warming and humidification trend in local areas will help to mitigate the negative effects of drought to some extent, the local hydrological system and ecological environment must still be enhanced (Wang et al. 2019). However, changing the local basic climatic state is far from sufficient. The climate in the research area is still arid and semi-arid.

The characteristics of drought trend based on M-K trend test show that the drought has an aggravating trend during 1961–2020, which is consistent with the findings of some scholars (Liu et al. 2020; Wang et al. 2020). The SPEI in spring showed the most significant downward trend (−0.05/10a). Previous studies have found that droughts were more frequent in spring and summer than in autumn and winter (Qin et al. 2016). This is consistent with the results of this research, which found that the frequency of droughts was higher in spring and summer (38.04% and 38.48%) than in autumn and winter (37.72% and 36.47%). The distribution of precipitation in different seasons is unbalanced, with less rainfall in spring and great interannual variation, accompanied by rising temperature and decreasing rainfall, so the drought in spring and summer is the most severe (Qin et al. 2016). In addition, the frequency of drought is higher in the western part of the study area, so the western part of the study area (such as Qinghai Province) is more prone to drought, indicating that the drought resistance measures in Qinghai Province are relatively weak.

Furthermore, large-scale climatic conditions play an important role in the research area's drought fluctuation trend (Huang et al. 2016b). By affecting the westerlies, the NAO and AO have a substantial impact on temperature and precipitation (Sung et al. 2006), changing the aridity of the study area. In this study, it was found that the NAO, AO, and SPEI index all changed in positive phase during the resonance period, and the correlation coefficient reached more than 0.65. This showed that AO and PDO reduced drought conditions in the study area. Changes in wind speed and air temperature are linked to ENSO and PDO is directly tied to snowmelt timing, which has an impact on water vapor transfer and glacier melting (Casey & Adamec 2002). In our research, SPEI and MEI.v2 index changed in negative phase during the resonance period, while SPEI and PDO index changed in positive phase during the significant resonance period, so ENSO and PDO had opposite effects on the generation of drought. It is worth noting that the PDO index and SPEI showed a negative correlation during the non-significant resonance period from 2007 to 2018. Therefore, the regulation of PDO index on the occurrence of drought is not invariable and will change in different periods. In addition, the significant resonance periods (1983–1999 and 2004–2019) of SPEI and MEI.v2 index are significantly longer than those of the other three indexes, so ENSO time has the most significant influence on drought in this region. On a worldwide scale, ENSO is a significant indication of interannual and interdecadal climate change (Webster et al. 1998; Ye & Wu 2018), and research has demonstrated that it interacts with other atmospheric oscillations (PDO, AO, NAO, MJO) (Tang & Yu 2008; Santoso et al. 2012; Wang et al. 2017). As a result, ENSO is recognized as a significant influencing factor in the region's extreme weather (drought, flood), which is consistent with the research findings indicating ENSO has the greatest impact on drought in this study area.

This research uses precipitation and meteorological data to determine the SPEI in the research area. It analyzes the link between SPEI and large-scale climate variables, as well as interannual fluctuations in precipitation, air temperature, and SPEI. However, the drought factors described in this study are limited. Furthermore, the physical mechanisms of drought are not considered, such as topography, vegetation, and soil. According to studies, human activities will increase the occurrence of heat and drought, and their influence will grow in the future (Samaniego et al. 2018). As a result, greater research into the interplay of physical mechanisms and drought at different geographic scales is required in order to develop targeted drought resistance approaches. Possible measures include strengthen regional farmland water conservation infrastructure, promoting a variety of farmland water-saving technology, and regularly varying the agricultural planting structure.

The trend changes in the time series of precipitation, evapotranspiration, and SPEI in the research area from 1961 to 2020 are discussed in this work. Drought area coverage and drought frequency were used to examine the characteristics of interdecadal, interannual, and seasonal droughts in the research area. Cross-wavelet analysis is also used to investigate the possible link between large-scale climate causes and dryness in this region. When paired with the research on the drought features of the Northwest of Yellow River Basin in this paper, different measures for different locations will help to raise the total study area's drought resistance level. The following are the primary conclusions:

• 1.

  The annual-scale precipitation and evapotranspiration interannual fluctuation trends in the Northwest of Yellow River Basin were not substantial, and the seasonal and annual SPEI indicate different dry and wet phases. The drought situation quickly altered in 1968, and by 1974, it had reached a significant level of 0.05. Spring, summer, autumn, and winter SPEI values all showed a clear decrease trend, with the change becoming more apparent about 2000.

• 2.

  The majority of the research region indicated a drying tendency, with the upper and middle parts of the watershed becoming drier and the downstream areas more humid. The places with an aridification tendency are predominantly found in the center and northern sections of the region, in terms of geographical disparities. In comparison to the other three seasons, the autumn drought trend is more noticeable, followed by summer and winter, while spring has the least noticeable drought trend.

• 3.

  Drought coverage in the Northwest of Yellow River Basin revealed no notable rising trend, and annual and quarterly drought coverage in the study area did not change significantly. The dry area in winter exhibited an insignificant increase trend, while the spring, summer, and autumn showed an insignificant downward trend. Drought frequency displayed an erratic and fluctuating fluctuation pattern, with a high value localized primarily in the western region.

• 4.

  In the studied area, the major cycle of yearly drought variation is 37 and 5 years. The primary spring cycle lasts 45 years or 10 years. Summer's main cycles are 20 and 5 years in length. The main autumn cycles are 36, 10, and 5 years. Winter cycles last 45, 22, and 5 years, respectively. Furthermore, ENSO, NAO, PDO, and AO are all closely linked to the incidence and change of drought in the research area, with ENSO having a larger association than the other three indexes and hence having a greater impact on the drought.

This work was supported by the National Natural Science Foundation of China (Grant numbers 52169010 and 51869023), the Training Project for the Top Young Talents in Ningxia (Grant number 030103030008), the Natural Science Foundation of Ningxia (Grant number 2021AAC03043), and Ningxia Key Research and Development Program (Grant number 2019BEB04029).

Material preparation, data collection, and analysis were performed by YF and XS. The first draft of the manuscript was written by YF. WL and XW performed supervision, and reviewed paper. QZ helped in producing figures and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.