The patterns, magnitude, and drivers of unprecedented 2022 mega-drought in the Yangtze River Basin, China

The Yangtze River Basin in China was hit by an unprecedented extreme drought in 2022. Such a record-breaking event is jointly driven by a few outlier factors and shows abnormal phenomena. The exceptionality of this drought event cannot be fully described by any individual indicator. Therefore, we performed this comprehensive study to highlight the exceptionality of 2022 Yangtze River Basin drought. We evaluated three drought characteristics: onset period (the time interval of moisture condition from normal to extreme), intensity, and affected area and their compound features. Additionally, historical climatology (1950–2022) and four past severe events are used as references. Lastly, we investigated the driving mechanisms of this event from synoptic perspective. Our results indicate the 2022 drought ranked as the most severe event in history. The recurrence interval of the 2022 Yangtze River drought is estimated to be 120–400 years by considering individual and combined drought characteristics. Behind the scenes, the synoptic environment triggered the abnormally high temperatures and abrupt alteration of precipitation, which is the main driver of the event. The 2022 Yangtze River drought is an archetype of an event characterized by both rapid intensification over time and space, which is indicative for drought monitoring and early warning in a warming climate.


Introduction
Climate change and anthropogenic greenhouse gas emissions have led to growing occurrences of dry and hot climate extremes (e.g.droughts, heatwaves, or compound hot and dry anomalies) around the globe (PaiMazumder and Done 2016, Schumacher et al 2019, Zscheischler and Fischer 2020, Christian et al 2020, AghaKouchak et al 2021, Christian et al 2021, Hao et al 2022).Meanwhile, the intensity of such extremes is also projected to increase as a result of the rising temperature, which threatens the welfare of human society and ecosystem sustainability (Chiang et al 2021, Zhou et al 2023).The year 2022 is one of the warmest years where high temperatures swept across majority of the Northern Hemisphere.Many regions such as the southwestern U.S., southern Europe, and India suffered serious drought during this year, resulting in devastating fatalities on social, agricultural, and ecological sectors (Annual 2022Drought Report, 2023;C3S 2022C3S , Rodrigues, 2023)), which alerts the community again that the detrimental impacts of such climate extremes in the context of global warming.
The Yangtze River Basin (24 • 30 ′ -35 • 45 ′ N and 90 • 33 ′ -122 • 25 ′ E) in China was also hit by drought during 2022.The river is 6300 km long, and flows through multiple terrains, including the Tibet Plateau, mountainous areas and plains, with elevation decreasing from northwest to southeast.The drainage area of the basin is approximately 1.8 million km 2 , accounts for one-fifth of the land area of China.
Affected by both East and South Asian monsoon, precipitation in the Yangtze River Basin also presents significant seasonality, with more than half of annual precipitation falls in summer and autumn.In contrast to previous historical events (Wang et al 2011, Hoerling et al 2014, Otkin et al 2016, Yuan et al 2019, Liu et al 2020a), the drought in this year started in July which is typically referred to as the flood season in the Yangtze River basin, and breaks multiple longstanding records during the observational periods 1950-present (2023 spring).With rapid intensification, the 2022 drought expanded quickly, resulted in majority of the basin area getting affected.According to the local hydrological monitoring center, the water levels of the two largest lakes in the Yangtze River Basin, i.e. the Poyang Lake and the Dongting Lake, reached to the lowest in early August, approximately three to four months earlier than usual.The sudden outbreak and rapid development of drought conditions in 2022 are unprecedented in China, which is more like a flash drought evolves (Otkin et al 2018, Yuan et al 2019, Liu et al 2020b, 2020c, Mohammadia et al 2022, Lesinger and Tian 2022).The annual statistics suggested that the 2022 drought in the Yangtze River Basin has affected 0.6 million ha of cropland, and the direct economic losses were estimated to be 7.5 billion USD.The drought condition has not recovered until present (2023 April), which is deem to induce far-reaching impacts on the ecosystem sustainability (Forzieri et al 2022, Li et al 2023, Yao et al 2023).
The 2022 drought in China is generally a climate outlier which is challenging to anticipate and manage, however, the unique characteristics of the megadrought compared with previous events, as well as the driving mechanism behind are not fully understood.The goal of this study therefore, is to make comprehensive analysis of the 2022 drought to illustrate how unusual event is.The findings are promising to improve our understanding of such 'Black Swan' so as to promote the prediction and early warning skills of subseasonal-to-seasonal drought.We specify the scientific questions as: (1) Which characteristics manifest the unusualness of the 2022 Yangtze River basin drought?(2) The extent to which the 2022 drought can be recognized a record-breaking extreme compared with historical events.And (3) what is the driving mechanism behind the event?The rest of the paper is structured as follows.Section 2 describes the data and method used in this study.Section 3 provides results and discussion, and a conclusion is drawn in section 4.

Data and methods
The root-zone soil moisture was obtained from the hourly modeled soil moisture data for the top three soil layers (0-100 cm) from the enhanced version of the fifth generation of the European Centre for Medium-Range Weather Forecasts reanalysis (ERA5land) during 1950-2022(Muñoz-Sabater et al 2021, Zhang et al 2022).The spatial resolution of ERA5land is 9 km, meanwhile, it also provides integrated products at varied spatial resolutions.In this study, we used the modeled reanalysis dataset at a spatial resolution of 0.25 • .In addition, the 500 hPa geopotential height and the water vapor flux data were also downloaded from ERA5 to analyze the synoptic conditions during the 2022 drought event.The temporal resolution of the two synoptic data is hourly, at a spatial resolution of 0.25 • .The hourly data were averaged into daily values for analysis.Daily precipitation, and maximum temperature observations were from 735 national meteorological sites (see figure S1 for the spatial distribution) in the Yangtze River Basin, during the period of 1950-2022.The normalized difference vegetation index (NDVI) was derived from Moderate Resolution Imaging Spectroradiometer (MODIS) produced on 16 d intervals and at a spatial resolution of 0.05 • , with more than 20 years of continuous data available (from 2001 to February 2023).NDVI was converted into anomaly value (for each grid cell, the images were classified into 24 groups (i.e.sub sample) according to the ordinal number in each year, and the anomaly value were derived as subtracting the mean from each original value and divided by the standard deviation based on each sub sample) to investigate the vegetation condition during the drought episode.
The root-zone soil moisture data were generated as the averages of three soil-layers (i.e.layer 1: 0-7 cm, layer 2: 7-28 cm, and layer 3: 28-100 cm), then converted into soil moisture percentile (SMP) for drought analysis.The hourly data were averaged into daily values, and classified into 12 groups according to their calendar month.Thirteen candidate theoretical probability distributions (e.g. the Beta, Gamma, Loglogistic, Generalized extreme value, Loglog, Weibull, Exponential, Generalized Pareto) were employed to fit the samples in each month, respectively, and the best fit was chosen as the one that passed the Kolmogorov-Smirnov test at the 95% significance level, along with the minimum root-mean-square error (RMSE).The derived cumulative probabilities were assigned to each calendar day to constitute an SMP series (Liu et al 2015, 2020b, Madadgar et al 2017, Jiang et al 2023).Drought is classified by the U.S. Drought Monitor (USDM, Svoboda et al 2002) in five categories, i.e. mild (0.2 < SMP ⩽ 0.4), moderate (0.1 < SMP ⩽ 0.2), severe (0.05 < SMP ⩽ 0.1), extreme (0.02 < SMP ⩽ 0.05), and exceptional (SMP ⩽ 0.02).Considering the rapid evolution of the 2022 drought, we calculate the onset period to show how fast the drought developed.Different from drought duration which indicates the overall persistence of drought events, the onset period is defined as the temporal length of drought development from abnormally dry (i.e.SMP is around 0.4) to a certain drought category, which also provides valuable information for drought monitoring and early warning.An example of the Wuhan city is shown in figure S2 in the supporting information, and the onset period varies corresponding to the drought category of interest.Given the high intensity of the 2022 drought, in this study, we define the onset period as the days taken for drought condition from abnormally dry to extreme.The univariate and bivariate return periods were estimated to measure the magnitude of historical drought events (Mishra and Singh 2011).We also use 13 candidate theoretical probability distributions to fit the data samples.The univariate return period of drought onset, and mean drought intensity were estimated respectively.It is worth mentioning that since drought do not occur every year, the average inter-arrival time of drought events in the sequence need to be considered for return period calculation.Detailed equations were referred to as Shiau and Shen (2001) and Ma et al (2013), and also provided in the supplementary material.For the bivariate case, the copula function was employed to model the dependence structure between random variables X and Y (Nelsen 2006).The joint return period can be estimated through the marginal distributions of X and Y (namely drought onset and average drought severity in this study), and the copula function (Salvadori et al 2011).We use five copulas, including two meta-elliptical (Student t and Gaussian) types and three Archimedean (Clayton, Frank and Gumbel-Hougaard), to construct the joint probability distribution function, and their fitting results were assessed with the goodness-of-fit statistics, i.e. the Kolmogorov-Smirnov test (K-S) at the 5% significance level, Akaike and Bayesian Information Criteria test (AIC and BIC), correlation coefficient (CC) and RMSE.We select the Gumbel-Hougaard copula that passes the K-S test and obtains the lowest value of AIC, BIC, and RMSE (table S1).

Spatio-temporal characteristics of the 2022 drought
The 2022 drought in the Yangtze River Basin can be recognized as the worst drought during the period of 1950-2022 that the peak of weekly drought area percentage was approximately 82%, with 46% of the basin area suffering exceptional drought (figure S3).According to the daily series of drought area percentage in 2022 (figure 1(a)), parts of the region experienced several droughts before July, but in a short persistence (less than 1 month).The most damaging hit of the 2022 drought in this basin started from 1 July, where the drought area percentage presented an upward pattern.Rapid areal expansion was found from 4 August to 23 August, where the area percentage of exceptional drought increased from 10% to 70% within 11 d (from 4 August to 15 August), and reached peak on 23 August with approximately 80% of the basin area under exceptional drought.Such percentage values exceed the 99% of historically annual maximum drought area in this region.
The 2022 drought was also characterized by rapid intensification.For majority of the basin, it generally took less than 15 d for drought to develop from non-drought to extreme (figure 1(b)).Spatially, the middle and lower Yangtze River Basin suffered more severe drought than the upper region given the longer drought duration (figure 1(c)) and larger accumulated drought severity (figure 1(d)).Figures 1(e)-(j) show the spatial distributions of SMP and corresponding NDVI anomalies during the 2022 drought.The results suggest more quicker response of vegetation to the enhanced drought condition for the year 2022 than in other drought years, where the area of negative NDVI anomalies enlarged by 300% relative to the beginning of the year (see figures S4 and S5 in the supporting information).For example, there was virtually no drought (i.e.SMP > 0.4) over the whole basin and the anomalies of NDVI were also positive in late June (figures 1(e) and 1(h)).In late July after the drought initiated, negative NDVI anomalies were found in areas with SMP below 0.2 (figures 1(f) and (i)), and areas of negative NDVI anomalies also enlarged and covered most of the Yangtze River Basin within two weeks (figures 1(g) and (j)).

Comparison with historical events
Before 2022, the Yangtze River Basin also suffered several severe droughts.For example, the 1959-1961 drought, and the 1978 drought were two earlier drought events with more than half of the basin area affected (figure S3).The 2006The , 2011The , 2013The , and 2019 droughts were four typical events after the year 2000, and in this study, they were selected as reference drought years given their similar warming conditions as the year 2022.To explore how unusual the 2022 drought is, we compared the onset periods and drought average intensity during these drought years.Given the wide longitude ranges of the basin (i.e. from 90 • 33 ′ E to 122 • 25 ′ E), the longitudinal variations of drought characteristics were presented, shown in figures 2(a) and (b).For drought onset period, the 2022 drought presents minor longitudinal variation and also ranks as the fastest developed event (the value is generally lower than 10th quantile (Q 10% ) of the onset period of droughts during 1950-2022).Such rapid spread of drought over the entire basin makes it difficult for water resources regulation so as to alleviate the drought condition, since majority of reservoirs in the basin empty their storage before flood season for flood prevention as did   in previous years.This highlights the necessity for improving subseasonal-to-seasonal drought prediction skills, which is also crucial for effective drought management and adaption (Pendergrass et al 2020).In addition, the 2022 drought is also characterized by large intensity.Expect the source region, the drought intensity in the other regions of the basin generally approaches to historical extremes (above the 90th quantile (Q 90% ) of the average drought intensity during 1950-2022).
Figures 2(c)-(e) show the drought return periods in the Yangtze River Basin based on univariate information and bivariate information, respectively.The April-October data from 1950 to 2022 were employed considering most of drought in the Yangtze River Basin occurred in these seasons (figure S6), and the characteristics of 300 driest pixels (approximately 10% of basin area) in each drought year were extracted for analysis.Based on univariate information, the 2022 drought is estimated to be a 120 year event for onset period, and a 250 year event for average drought intensity.If we consider both onset period and average intensity, the recurrence interval of the 2022 Yangtze River drought is approximately 400 years, and the magnitude is far more severe than the 2006, 2011, 2013, and 2019 drought in the Yangtze River Basin.It is worth mentioning that the values of return period can be different by using different data samples (Robeson 2015).Nonetheless, it is evident that the 2022 drought in the Yangtze River Basin can be recognized as a record-breaking event given the both rapid intensification and large magnitude of drought intensity.The 2022 drought resulted in far-reaching impacts on ecological, agricultural, hydrological, and socioeconomic sectors, which can be comparable to those of the 2012 California drought (AghaKouchak et al 2014, Robeson et al 2015, PaiMazumder et al 2016).In-depth evaluation on these extreme events is necessary in future researches so as to better cope with climate change.

Driving factors of the 2022 drought
As described above, the 2022 Yangtze River basin drought has been identified as the most severe drought since 1960.Here we investigate the driving mechanism behind the record-breaking drought.Figure 3(a) shows the time series of 735-station averaged daily maximum temperature over the year 2022 and the other 4 drought years (2006, 2011, 2013 and 2019).It illustrates that relative to the multi-year mean daily maximum temperature (black solid line), 2022 experienced an abnormally hot spring, summer and fall, especially during the period between the 22 June and the 19 August.Such abnormally hot period covers the days before and during drought outbreak.As shown in the inserted figure, compared with the other four years with severe drought events, 2022 has the most stations (51.97%) that recorded extreme high temperature (above 40 • C), indicating the most widespread heat events.Additionally, the recorded extreme hot days in 2022 last longer than the other four years.On average, the extreme hot event in 2022 last 7.Along with temperature, precipitation is another factor that significantly influence the 2022 drought, and it shows a unique temporal pattern, as shown in figure 3  is substantially higher than the multi-year averaged value and even close to the historical high record (the upper bound of the gray area).The inserted figure provides more detailed evidence.Before June, precipitation is more than historical mean values.After that, the rise of accumulated precipitation slows down dramatically.During the period of June to October, the 2022 precipitation is far lower than the multiyear averaged value.Especially in the period of 19 July to 7 September, the precipitation in 2022 is 118.16 mm lower than the multi-year averaged precipitation, which leads to the quick intensification of drought in the summer of 2022.The combination of wet and dry half year may be one other aspect that can worsen the drought condition.Based on the precipitation amount before July, all the reservoirs in the upper stream were emptied to prepare for the upcoming flooding season.However, such regulation caused a worse consequence when the unexpected drought occurred.The sudden decreased precipitation since July is also associated with the synoptic conditions.In May and June, the moist air from the Indian Ocean and the South Pacific Ocean increases the precipitation in the Yangtze River Basin.However, in early July, with the extension of the west Pacific subtropical high, the northern region of China is mainly influenced by the westerly wind, which prohibits the high latitude water vapor from moving southward.At the same time, the west Pacific subtropical high also leads to anti-cyclone system in the Yangtze River Basin, which may further accelerate the divergence of water vapor flux so that to inhibit the occurrence of precipitation.
Besides the impacts of large-scale atmospheric circulation, land-atmosphere interactions may also play a role in driving the rapid increase in temperature.As shown in figure 5(a), the declining soil moisture and the increasing temperature show substantial opposite changes from day 150 to day 300, and the correlation coefficient between soil moisture against lag-1 d T max during this exceptional drought period is −0.4848.We further compared this value with those in other drought years during the period from 1950 to 2021.As shown in figure 5(b), (a) stronger negative correlation between soil moisture and lag-1 d T max is found during the 2022 drought episode than in other drought years, which is suggestive of the enhanced coupling strength during this drought event.It is worth mentioning that the correlation measures alone are insufficient to demonstrate links of causality, while they may provide indirect estimates of soil moisture-temperature coupling strength (Seneviratne et al 2010).For quantifying more exact causality between soil moisture and temperature, analysis based on multiple approaches, such as the numerical experiments with the land-atmosphere coupling models are recommended (Hall and Perdigão 2021).

Conclusion
In this study, a comprehensive analysis of the 2022 drought in the Yangtze River Basin is performed to illustrate how unusual the drought is during the period 1950-2022.We investigated the onset period for drought development from nondrought to extreme, the average intensity, accumulated severity, drought affected area, and the vegetation response, and also estimated the magnitude of drought based on univariate and bivariate return periods analysis.We also focused on the two main driving factors: temperature and precipitation to further reveal the possible reasons that cause the 2022 drought.The results suggest that the 2022 drought ranked as the worst event based on all above drought characteristics.Unlike previous historical events, the 2022 drought initiated in July, which is a flood season in usual years.Under the joint effects of the extreme high temperature and the sharp decreased precipitation after June, approximately 80% of the basin area experienced drought development from normal (SMP > 0.4) to exceptional (SMP ⩽ 0.02) within two weeks.Such rapid intensification of moisture condition both over time and space is unprecedented Y Liu et al in China, which is more challengeable for dealing with but often overlooked in previous researches.Widespread inhibiting effects on vegetation growth is apparent over the basin, where the area of negative NDVI anomalies enlarged by 300% relative to the beginning of the year, far exceeding the cases in previous drought years.
The return period results show that the 2022 drought is estimated to be a 120 year event for onset period, a 250 year event for average drought intensity, and a 400 year event if both onset period and average intensity were considered.The magnitude is far more severe than the 2006, 2011, 2013, and 2019 drought in the Yangtze River Basin.The 2022 drought in the Yangtze River Basin confirms the statement that drought may be set in quicker pace with strong intensity in the context of global warming (Trenberth 2014, Wouters et al 2022).As a unique feature of flash drought, the onset period measures how fast drought condition develops, which provides valuable information for drought early warning and subseasonal-to-seasonal drought prediction, and is also crucial for effective drought management and adaption (Pendergrass et al 2020).The results in this study highlight the importance of incorporating the drought onset period as an important metric in the systematic framework of drought monitoring, identification, early warning, and adaption in the future both for global or regional assessments.

Figure 1 .
Figure 1.(a) Daily series of area percentage for moderate drought (orange), severe drought (red), extreme drought (dark red), and exceptional drought (brown) averaged over the Yangtze River Basin during the 2022 drought.The four horizontal dashed lines show the 99%, 95%, 75%, and 50% of annual maximum drought area during 1950-2021.Spatial distributions of (b) the days consumed for drought from normal to extreme, (c) drought duration, and (d) accumulated drought severity during the 2022 drought.(e)-(g) are the spatial distributions of soil moisture percentile, and (h)-(j) are the spatial distributions of NDVI anomalies during the 2022 drought, respectively.Positive anomaly of NDVI indicate good health condition for vegetation, while negative values indicate an inhibitive effect on vegetation growth.

Figure 2 .
Figure 2. (a), (b) Line plots showing the longitudinal variation of (a) average onset period (in units of days) and (b) average drought intensity (in units of percentile) during the 2006 drought (blue dots), 2011 drought (orange dots), 2013 drought (bluish violet dots), 2019 drought (dark green dots), and 2022 drought (red line & dot) over the Yangtze River Basin.The black solid lines and color shades in (a)-(b) are the mean and quantiles of onset period and drought intensity during 1950-2022.(c), (d) Return periods of (c) drought onset period, and (d) mean intensity based on April-October data from 1950 to 2022.(e) Return periods based on the joint distribution of onset period and mean intensity.The solid curves show the marginal probability distribution of onset period (in blue) and mean intensity (in red), respectively.

Figure 3 .
Figure 3.Time series of maximum temperature and accumulated precipitation.(a) Time series of maximum temperature.The middle panel shows the percentage of stations exceeding 40 degrees over different years.(b) Time series of accumulated precipitation.The upper left corner panel shows the comparison between the total precipitation in 2022 and the multi-year average precipitation every 50 d.
3 d, followed by 2006, 2011, 2013 and 2019.The extreme high temperature is mainly affected by the abnormally strong west Pacific subtropical high during the summer in 2022.As shown in the left panel of figure4, the west Pacific subtropical high keeps growing since June (red color becomes darker), and controls the most region of Yangtze River Basin.Especially in July, the downdraft caused atmospheric nonadiabatic heating processes increases the temperature in the basin dramatically.
(b).Before July, the accumulated precipitationY Liu et al

Figure 4 .
Figure 4. Evolution of 500 geopotential height (left panel) and water vapor flux (right panel) from early May to early August.

Figure 5 .
Figure 5. (a) Timeseries of soil moisture percentile and maximum daily temperature (Tmax) in the grid cell where the cumulated drought severity is the strongest during the year 2022; (b) Boxplot of correlation coefficients between day 0 soil moisture percentile and day +1 Tmax based on non-2022 data (samples are extracted from the drought spells in drought years during 1950-2021), and the horizontal line represents the correlation coefficient based on the 2022 data.