Leading Pattern of Spring Drought Variability over East Asia and Associated Drivers

Drought events, characterized by prolonged periods of abnormally low precipitation (Chen et al.^[1]; Li et al.^[2]), can exert significant impacts on the hydrological cycles, ecosystems, and human society (Huang et al.^[3]; Chen et al.^[4]; Kostopoulou and Giannakopoulos^[5]). The occurrence of severe droughts not only affects crop growth and agricultural production in the spring (Han et al.^[6]; Santini et al.^[7]; Higgins et al.^[8]), but also leads to an increased frequency of natural disasters such as heatwaves and wildfires in summer due to dry surface conditions (Purich et al.^[9]; Wang et al.^[10]; Heino et al.^[11]). In recent decades, drought events have become more frequent across the world, resulting in substantial socioeconomic impacts (Suska and Piotr^[12]; Li et al.^[13]; Dong et al.^[14]). Notably, East Asia is a region particularly susceptible to the drought disasters (Zhou and Wang^[15]; Deng et al.^[16]). For instance, Southwestern China experienced unprecedented droughts during the spring of 2010, resulting in water scarcity, crop failures, and significant economic losses (Zhang et al.^[17]; Liu et al.^[18]). Between July and August 2022, the Yangtze River basin of China faced its most severe drought on record, causing water levels in China's largest freshwater lake, Poyang Lake, to plummet from 19 meters in June to 9 meters by the end of August (Mallapaty^[19]). Moreover, the frequency and severity of extreme dry events are projected to increase in a warming climate (Qiao et al.^[20]; Tang et al.^[21]; Ullah et al.^[22]). Hence, it is important to enhance our understanding of the changes and variations in drought events over East Asia, as well as their underlying physical causes.

The drought severity in East Asia is significantly influenced by the Asian monsoon system (Yu et al.^[23]; Cen et al.^[24]; Huang et al.^[25]). Previous studies have indicated that the East Asian summer monsoon circulation has weakened in recent decades (Wang^[26]; Wang et al.^[27]; Zhou and Yu^[28]), causing reduced precipitation in North China and consequently exacerbating the dry conditions there (Li et al.^[29]; Zhou et al.^[30]; Zhang et al.^[31]). In contrast, the changes in the South Asian summer monsoon can affect the transport of water vapor from the tropical Indian Ocean through the Bay of Bengal to Southwestern China, affecting the dryness in the region (Simmonds et al.^[32]; Zhou and Yu^[28]). Moreover, the global SST anomalies can significantly regulate the interannual to interdecadal variations of the drought events in East Asia. Particularly, the El Niño-Southern Oscillation (ENSO) plays a crucial role in driving the interannual variations of extreme dry conditions in China (Cao et al.^[33]; Wang and Li^[34]). The El Niño-like SST can lead to the intensification and westward shift of the Northwestern Pacific subtropical high, resulting in reduced precipitation and increased temperatures in South and Southwestern China, which in turn favor the occurrence of drought events (Wang et al.^[35]). In addition, Asian large-scale orography forcing, particularly the Tibetan Plateau (TP), significantly increases ascent, low-middle cloud formation, and resultant strong spring cloud radiative cooling over Southeastern China (SEC) and downstream ocean, thereby affecting climate change (Kiehl and Trenberth^[36]; Li et al.^[37]). Studies suggest that the western Pacific subtropical high anomaly was stronger, westward and southward in the spring of 2019, which directly led to significantly more precipitation in Northeastern China, eastern Northwestern China, and South China, while the Huang-Huai, Jiang-Huai and most Yunnan provinces had less precipitation (Liu and Chen^[38]).

Although previous studies have extensively explored the dry and wet extremes in East Asia (Du et al.^[39]; Huang et al.^[3]; Qiao et al.^[20]), most are focused on the trends, on individual cases, or for specific regions. The leading mode of drought variability across East Asia during the boreal spring and its underlying causes remain unclear. Since spring marks the period of crop development in East Asia, alterations in springtime drought conditions can significantly affect crop growth, posing a potential threat to regional food prices and security. Therefore, this study aims to assess the severity of drought in East Asia and investigate its variations and physical causes during the boreal spring. Section 2 presents the data and methods used in this research. Sections 3 and 4 present the research findings, followed by a summary and discussion in Section 5.

The Standardized Precipitation Evapotranspiration Index (SPEI) is a multiscalar drought index derived from climatic data. It serves as a tool for identifying the onset, duration, and severity of drought conditions in comparison to typical circumstances (Vicente-Serrano et al.^[40]). Mathematically, SPEI is simply computed by total precipitation minus potential evapotranspiration. However, the estimation of potential evapotranspiration is difficult in regions without sufficient data, which follows the Penman-Monteith method (Allen et al.^[41]):

In Formula 1, ET₀ is the potential evapotranspiration (mm d^–1), and R_n is the net radiation (MJ m^–2 d^–1). G is soil heat flux (MJ m^–2 d^–1); T is the daily average temperature (℃); u₂ is the wind speed at 2 m (m s^–1), obtained by multiplying the wind speed at 10 m by 0.75. e_s is saturated vapor pressure (kPa); e_a is the actual water vapor pressure (kPa); Δ is the slope of saturated water vapor pressure-temperature curve (kPa ℃^–1); γ is a hygroscopic constant (kPa ℃^–1) (Valiantzas^[42]; McColl^[43]). SPEI offers a comprehensive assessment of drought by taking into account both evapotranspiration and precipitation, enabling the detection of drought across various temporal scales (Vicente-Serrano et al.^[40]; Homdee et al.^[44]; Wang et al.^[45]). In the context of studies in China, the Penman-Monteith method has been found to be particularly suitable (Chen et al.^[46]).

ERA5 is the state-of-the-art reanalysis dataset, which is generated through the 4D-Var data assimilation of the ECMWF Integrated Forecast System (IFS) and the CY41R2 model (Hersbach et al.^[47]). In this study, the variables that we analyze in this study include 2-m temperature, precipitation, soil moisture (at a depth of 1 m underground), three-dimensional winds, geopotential height, surface pressure, sea surface temperature, specific humidity, total cloud cover, and solar downward short-wave radiation. The horizontal resolution of these variables is 1°×1° (longitude × latitude) for the period from 1979 to 2020. The study area of this study, i.e., East Asia, spans from 15°N to 55°N and from 70°E to 150°E.

In this study, various statistical methods have been employed to investigate the characteristics and the associated drivers of spring drought severity across East Asia. These methods include the Empirical Orthogonal Function (EOF) analysis, correlation/regression analysis, composite analysis, trend analysis, and lagged correlation/regression analysis. The correlation coefficient (r) is calculated as follows:

where x₁, x₂, ..., x_n and y₁, y₂, ..., y_n is a random variable for two groups of data (Ahlgren et al.^[48]). The range of correlation coefficient r is between [–1, 1]. The greater the absolute value of r, the greater the correlation between the two sets of data.

Figure 1 shows the climatology, trend, and standardized deviation of SPEI in the spring for the period 1979–2020. The climatology of SPEI shows that Northwestern China, North China, and Mongolia typically featured an arid climate (Fig. 1a). The trend in SPEI reveals that Northwestern China, North China, and Mongolia experienced a drying trend over the past several decades, while Southwestern China displayed a wetting trend (Fig. 1b). The standardized deviation of SPEI is relatively large in Northwestern China, North China, and Southwestern China, implying that there exists a larger variability of drought in these regions (Fig. 1c).

Figure 1.  (a) Climatology, (b) linear trend, and (c) standardized deviation of the spring SPEI over East Asia.

Figure 2 further presents the first EOF mode (EOF1) of SPEI variability in East Asia during the boreal spring. EOF analysis is also known as principal component analysis (PCA) (Pearson^[49]). The first EOF mode (EOF1) has the most obvious evolution of spatial distribution characteristics and coefficients over time and has the largest interpretation variance. Only EOF1 is analyzed here. The EOF1 shows a north-south dipolar pattern, with opposite signs of SPEI anomalies between northeastern and southwestern parts of East Asia (Fig. 2a). Moreover, the primary component of EOF1 (PC1) shows an overall upward trend, generally changing from negative phases to positive phases (Fig. 2b). Combined with the spatial characteristics of EOF1, it indicates that Northwestern China, North China, central China, and Mongolia have experienced a drying trend, while Southwestern China has experienced a wetting trend.

Figure 2.  (a) Spatial pattern and (b) primary component of EOF1 of variability in the spring SPEI over East Asia. In (b), the black curve and dashed line indicate the original PC1 and its regression trends, respectively, while the bars represent the PC1 with the trends removed.

Figure 3 shows the SPEI composite results of extremely dry and wet years based on a 1.5 standardized deviation of the de-trended PC1. The years with standard deviations higher than 1.5 are characterized by dry surface conditions in Northwestern China, North China, and Mongolia and wet surface conditions in Southwestern China (Fig. 3a). The opposite results were obtained for the years with standard deviations below 1.5 (Fig. 3b). The differences in SPEI between the positive and negative phases of PC1 further highlight the dry-wet pattern, where the absolute value of the SPEI differences can be higher than 4.5 in North China and Southwestern China (Fig. 3c).

Figure 3.  Composite patterns of spring SPEI are based on the years when the standardized deviations of PC1 are (a) more and (b) less than 1.5. (c) Difference between (a) and (b).

Figure 4 displays the lagged correlations between PC1 and the SPEI. The signal of the East Asian dry-wet patterns appeared four months before spring. In periods with two- and zero-month advance, the correlation coefficients between PC1 and SPEI become more significant, with the main dry areas occurring in Northwestern China, North China, and Mongolia, while wet areas in Southwestern China.

Figure 4.  Lag-lead correlations between SPEIs in different lags and the detrend PC1 during 1979–2020.

The SPEI variations are closely related to the temperature and precipitation. As illustrated by Fig. 5, the anomalous dryness in Northwestern China, parts of North China, and Southern Mongolia corresponds to increased air temperature (left panels) and decreased precipitation and soil moisture (middle and right panels, respectively) four and two months preceding the spring drought (left panels). By comparison, the anomalous wetness in Southwestern China is correlated with abnormal temperature and precipitation patterns, opposite to those in the dry areas. These significant lagged correlation patterns also imply that the dry-wet variations of East Asian drought, as revealed by the EOF1, possess a potential for subseasonal and seasonal predictions.

Figure 5.  Lag-lead correlations of 2-m temperature (left), precipitation (middle), and soil moisture (right) in different leads with the detrend PC1.

Figure 6 shows the spatial distributions of regressions of geopotential height and horizontal winds onto the PC1 over East Asia. The results show that preceding the spring dry and wet variations, an abnormal high pressure accompanied by anticyclonic circulation occurred over Northwestern China and Mongolia, and an abnormal low pressure accompanied by cyclonic circulation existed over Japan. North China is located between the anomalous high and low pressure systems and is affected by northerly wind anomalies.

Figure 6.  Lagged regressions of geopotential height (units: dagpm) and horizontal winds (units: m s^–1) in different leads at 850 hPa (left), 500 hPa (middle), and 300 hPa (right) onto the detrend PC1. Areas with significant anomalies exceeding the 95% confidence level are stippled.

Figure 7 further presents the regression patterns of water vapor flux, total cloud cover, and solar shortwave radiation onto PC1. The results show that the water vapor flux anomalies are closely linked to the wind anomalies. The drying areas such as Northwestern China, North China, and Mongolia are primarily featured by divergent anomalies of water vapor fluxes. The variability of precipitation is directly controlled by local vertical velocity anomaly, and anomalous ascent (descent) leads to excessive (deficient) precipitation (Ni and Hsu^[50]; He and Li^[51]). According to isentropic gliding mechanism, the climatological isentropic surfaces tilt northward with altitude in the extratropics, and anomalous northerly (southerly) wind generates descent (ascent) anomaly by gliding along the sloping isentropic surfaces (Hoskins et al.^[52]; Wu et al.^[53]; He^[54]), associated with anomalous divergence (convergence) in lower troposphere. Therefore, the dry areas are accompanied by water vapor divergence, reduced total cloud cover, and increased short-wave solar radiation due to the anomalous anticyclone and stronger descending motions. By comparison, the anomalous wetness in Southwestern China has obtained increased water vapor transport from the north Pacific, increased total cloud cover, and decreased solar radiation. As a result, the increased surface air temperature and reduced precipitation favor anomalous dry conditions in Northwestern China and North China and vice versa in Southwestern China.

Figure 7.  Regressions of the vertical integrals of (a) water vapor flux (vectors, units: kg m^–1 s^–1) and divergence (units: kg m^–2 s^–1), (b) total cloud cover, and (c) solar download radiation in spring onto the detrend PC1. Areas with significant anomalies exceeding the 95% confidence level are stippled.

The tropical SST anomalies can exert significant impacts on weather and climate extremes in East Asia. Fig. 8a displays the correlation coefficients between PC1 and the SST, which shows that the spatial patterns of dry North China and wet Southwestern China are correlated with the La Niña - like SST anomalies, coinciding with warm SST anomalies in the western Pacific and cold SST anomalies in the central and eastern Pacific and the Indian Ocean. Consequently, the SST gradients between the western Pacific and the central and eastern Pacific increase. Fig. 8b shows the regression patterns of PC1 with geopotential height and horizontal winds, which show anomalous high and low pressures over East Asia, which resemble a Rossby wave-train linked to the SST anomalies. When the western Pacific Sea temperature rises, convective activities are motivated, abnormal high pressure will be generated over the northwest of Mongolia, Alaska, and the western coastal areas of the United States, while obvious abnormal low pressure will appear over Japan, forming a "+–+" teleconnection wave train propagating from west to east.

Figure 8.  (a) Regression between SST in spring and the detrend PC1. (b) Regression of 300-hPa geopotential height and horizontal winds onto the detrend PC1.

Previous studies have indicated that SST changes in the Northern Indian Ocean and the tropical Northern Atlantic affect circulation systems in the Northwestern Pacific region through air-sea interactions, which has a significant impact on the weather and climate in East Asia (Kim and Hong^[55]; Chen et al.^[56]). The western Pacific (10°N–20°N, 120°E–180°), the middle Eastern Pacific (–5°S–15°N, 120°W–160°W), and the Northern Indian Ocean (–10°S–10°N, 70°E–120°E) are thus selected as three key ocean areas to define a new SST index:

where subscripts 1, 2, and 3, respectively, represent the mean SST over the three selected sea areas. The correlation coefficient r between S and PC1 is about 0.514, which has passed the significance test (Fig. 9a), indicating that the tropical zonal SST gradient in the Indo-Pacific region is significantly correlated with the spring dry and wet variations in East Asia. When the Indo-Pacific SST gradient increases, dry (wet) surface conditions tend to occur in Northwestern China, North China, and Mongolia (Southwestern China). In addition, the results of Figs. 9c and 9d are consistent with those of Fig. 7a and Fig. 4e, indicating that the tropical Pacific SST anomalies could significantly influence the spring dry and wet variations over East Asia by triggering atmospheric circulations.

Figure 9.  (a) SST index averaged over the area in the box shown in Fig. 8a (bars) and the detrend PC1 (curve). (b), (c), and (d) respectively represent the regressions of 300-hPa geopotential height and horizontal winds, the water vapor flux and divergence, and the SPEI onto the SST index.

Based on the global SPEI data calculated by the Penman-Monteith method and the ERA5 reanalysis product from 1979 to 2020, this study investigates the characteristics and possible drivers of the spring dry and wet variations in East Asia. The main conclusions are as follows: (1) Over the past several decades, the leading variability in spring SPEI over East Asia has been characterized by a dipole pattern, with increased dryness (wetness) in Northwestern China, North China, and Mongolia (Southwestern China). (2) The leading mode of SPEI variability in East Asia is significantly correlated with the anomalous anticyclone (cyclone) over North China and Mongolia (Southwestern China), leading to increased (decreased) temperature and decreased (increased) precipitation and soil moisture in North China and Mongolia (Southwestern China). (3) The tropical Pacific SST anomalies could influence the leading mode of SPEI variability in East Asia during boreal spring by triggering anomalous atmospheric circulation.

It should be noted that our research shows that the variability of spring dryness in East Asia signals 2–4 months in advance, which provides some help in predicting sub-seasonal and seasonal changes in the regional spring drought. Although our study suggests that the anomalous SSTs in the tropical Pacific may influence the spring drought in East Asia, these findings are primarily based on statistical analyses. Therefore, validation through numerical modeling experiments is critical for further substantiating the possible drivers. Furthermore, in addition to the tropical Pacific Ocean, the Northern Atlantic SST, and atmospheric variability, such as the Atlantic Multidecadal Oscillation and the North Atlantic Oscillation, could also influence the spring drought in East Asia. However, how these factors interact synergistically with the Pacific SSTs remains unclear and requires further in-depth analyses in the future.