The anthropogenic acceleration and intensification of flash drought over the southeastern coastal region of China will continue into the future

Yumiao Wang ,Xing Yuan,*

a Key Laboratory of Hydrometeorological Disaster Mechanism and Warning of Ministry of Water Resources/Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters, Nanjing University of Information Science and Technology, Nanjing, China

b School of Hydrology and Water Resources, Nanjing University of Information Science and Technology, Nanjing, China

Keywords:Flash drought Climate change Drought onset speed Detection and attribution

ABSTRACT The frequent and rapid onset of flash drought poses a serious threat to agriculture and ecosystems.Detecting human influences on flash droughts and estimating their future risks under climate change have attracted great attention.Focusing on a record-breaking flash drought event in the southeastern coastal region of China in summer 2020,the authors found that the suppression of convective precipitation and high temperature caused by the persistent high geopotential height anomalies and land–atmosphere dry coupling were important reasons for the rapid onset and strong intensity of the flash drought.Event attribution analysis with the latest CMIP6 data showed that anthropogenic climate change has not only increased the likelihood of an onset speed and intensity like those of the 2020 flash drought event,by about 93% ± 20% and 18% ± 15%,respectively,but also increased the chance of their simultaneous occurrence,by about 86% ± 38%,according to their joint probability distribution.Under a business-as-usual future scenario (SSP2-4.5),the likelihood of such an onset speed,intensity,and their simultaneous occurrence will further increase,by 85% ± 33%,49% ± 8%,and 81% ± 48%,respectively,as compared with current climate conditions.This study highlights the importance of anthropogenic climate change for accelerating and intensifying flash drought in the southeastern coastal region of China.

1.Introduction

During the summer of 2020,the southeastern coastal region of China ((20°N,111°E)–(25°N,121°E)–(29°N,119°E)–(24°N,109°E)) experienced unexpected hot weather.Meanwhile,widespread drought occurred rapidly due to the influence of an extreme deficit of precipitation.With 758 000 km2of land having suffered from severe drought,the event has been identified as one of the top 10 extreme weather and climate events of 2020 in China (Liu,2021).In addition to the severe damage it caused,the uniqueness of this drought event was its rapid onset speed.It developed into a serious drought condition within 10 days,posing a considerable challenge for the early warning system,and making it identifiable as a typical “flash drought ”event (Svoboda et al.,2002 ;Otkin et al.,2018 ;Yuan et al.,2020).Previous studies have proved that summer drought in the southeastern coastal region of China has enhanced because of the increase in temperature and decrease in light precipitation under the influence of climate change (Yao et al.,2010 ;Li et al.,2016).Also,frequent flash drought events accompanied by heat waves have a devastating impact on crop yields in this region,as well as on human health and the security of the eco-environment(Zscheischler et al.,2018 ;Zhang and Yuan,2020 ;Zhang et al.,2020),especially in a warming world.Therefore,starting with this case study of the 2020 extreme flash drought,it is of great significance to understand the causes of flash droughts under climate change in the southeastern coastal region of China and estimate the risk of such events in the future to help with disaster prevention and mitigation.

It has become an indisputable fact that human activities are the main factors causing global warming (IPCC,2021),and the contribution of anthropogenic climate change to increasing the likelihood of extreme drought events has been detected in different parts of the world,such as China (Chen and Sun,2017),western Canada (Szeto et al.,2016),East Africa (Lott et al.,2013),and California (Diffenbaugh et al.,2015).Chen and Yuan (2021) also concluded that the increase in drought frequency over China will be amplified by a higher global warming level in the future,especially over South China.However,the above research mainly focused on the severity and frequency of drought events at the seasonal time scale;the impact of climate change on the evolution of drought events at subseasonal scales has yet to be studied.

Flash drought is identified when a rapid decline in soil moisture (SM)occurs (Christian et al.,2019 ;Yuan et al.,2019 ;Qing et al.,2022),which extends the drought analysis to subseasonal time scales (Yuan et al.,2020).With the frequent and rapid onset of flash drought around the world (Yuan et al.,2015,2019 ;Otkin et al.,2016 ;Wang et al.,2016 ;Wang and Yuan,2018 ;Gerken et al.,2018),concern has arisen as to whether climate change is having a significant impact on this phenomenon (Yuan et al.,2020).On the basis of flash drought events during 2015/16 in southern Africa,Yuan et al.(2018) found that anthropogenic climate change has tripled the frequency of flash drought.Subsequently,anthropogenic climate change was also implicated in increasing the frequency and severity of flash drought in China (Yuan et al.,2019).Focusing on the onset process of a 2019 flash drought in South China,Wang and Yuan (2021) concluded that anthropogenic climate change has increased the likelihood of such onset speeds in the transitional season of summer to autumn.However,whether the acceleration of flash drought will continue into the future requires further investigation.In addition,the land–atmosphere coupling characteristics during extreme flash drought events in regions with strong oceanic influences (e.g.,the southeastern coastal region of China) remain unclear,which hinders our understanding of the mechanisms behind the acceleration and intensification of flash drought.

In this study,we analyzed,at the subseasonal time scale,the whole process of evolution of the flash drought that occurred in the southeastern coastal region of China in 2020,and further quantified the impact of anthropogenic climate change and estimated future risks based on CMIP6 climate models.The specific questions that we sought to answer were: (1) What were the evolutionary characteristics and physical causes of the 2020 flash drought in the southeastern coastal region of China?(2) What is the contribution of anthropogenic climate change to such events? (3) How will flash drought events change in the future?

2.Data and methods

2.1. Data

Hourly SM data (top 1 m,weighted interpolation from the SM within 0–7 cm,7–28 cm,and 28–100 cm according to their layer thickness) and evapotranspiration data were collected from the fifth major global reanalysis produced by ECMWF (ERA5;Hersbach et al.,2020 ;Bell et al.,2021).Daily precipitation and temperature data were collected from CN05.1 (Wu and Gao,2013 ;Xu et al.,2009).The hourly 2-m temperature,2-m dewpoint temperature,and surface air pressure,as well as the atmospheric specific humidity and temperature at 500–1000 hPa (16 layers in total) from ERA5 were used to calculate land–atmosphere coupling indexes,and the hourly 850-hPa wind and 500-hPa geopotential height from ERA5 were used for investigating the circulation characteristics.Both datasets cover the period 1961 to 2020 at a spatial resolution of 0.25°×0.25°.Besides,the daily mean full-column SM data from eight CMIP6 (Eyring et al.,2016) models that include natural-only forcings(NAT,1961–2014;25 realizations),all forcings (ALL,1961–2014;25 realizations),and a future projection under the business-as-usual scenario (SSP2-4.5,2015–2100;18 realizations) were used to carry out attribution analysis and future estimation (Table S1).

2.2. Methods

2.2.1.Flashdroughtevolutionanalysis

According to the evolution of the pentad-mean SM percentile,we identified the flash drought period during 2020 in the southeastern coastal region of China and divided it into three stages: onset,persistence,and recovery.Here,we focus on the onset speed and intensity of the flash drought.The flash drought onset speed is defined as the decline in regionally averaged SM percentiles during the onset and before the onset stages (the mean SM percentile during 4–13 June minus the mean percentile during 14–23 June),and the intensity of flash drought is defined as the differences between 40% and the mean SM percentile during the whole flash drought period (40% minus the mean percentile during 14 June to 11 September),according to the definition proposed by Wang and Yuan (2021).

At different stages of flash drought,we investigated the physical causes through analyzing the conditions of the atmospheric boundary layer (ABL) and the spatial patterns of atmospheric circulation.Convective triggering potential (CTP;Findell and Eltahir,2003) is the vertical integral of the area between the ambient temperature profile and the moist adiabat from 100-hPa to 300-hPa above the surface,and a greater CTP means that the ABL is more unstable:

whereg(m s-2) is the gravitational acceleration,ZPSurf-100andZPSurf-300(m) are the heights of 100 hPa and 300 hPa above the surface,Tparcel andTenv(K) are the temperatures of a saturated air parcel and the environment,and dz(m) is the ABL thickness.Low-level humidity index(HI) was calculated by the sum of the differences between temperature and dewpoint temperature at 50-hPa and 150-hPa above ground level,and a higher HI expresses a drier ABL (Lytinska et al.,1976):

whereTPSurf-50andTPSurf-150(K) are the temperature at 50-hPa and 150-hPa above ground level,andTd,PSurf-50 andTd,PSurf-150 (K) are the dewpoint temperature at 50-hPa and 150-hPa above ground level.The CTP and HI were calculated at the daily time scale and then averaged and standardized at different stages of the flash drought.

2.2.2.Multivariatecopulamethod

In this study,we used the copula function to evaluate the return period of this flash drought event and analyzed the joint distribution of precipitation and temperature.The copula method links random variables with the joint distribution of their respective marginal distributions (Salvadori and De Michele,2010).The joint cumulative distribution functions ofX1andX2can be defined as

whereX1andX2express the variables,andx1andx2represent the thresholds.According to Kolmogorov–Smirnov (K-S) test results,Gaussian and generalized extreme value distributions were selected as the marginal distributions.Furthermore,the Plackett and Clayton copula functions were selected as the optimal functions for the joint distributions through log likelihood estimation (Zhang and Singh,2012 ;see Table S2 for detail) by using the MATLAB software.

2.2.3.Attributionanalysisandfutureestimation

The datasets of CMIP6 under NAT and ALL scenarios for historical simulations and SSP2-4.5 for future scenarios were used to analyze the influence of anthropogenic climate change on 2020-like flash drought events and the associated future risk.The fraction of attributable risk(FAR;Stott et al.,2004) can be calculated as

where FAR anthropogenic and FAR future are the fractions of attributable risk for anthropogenic climate change and the future,Pnatis the probability of an event in the CMIP6/NAT simulation,andPall andPssp2 -4.5 are the probabilities of the same event in the CMIP6/ALL and CMIP6/SSP2-4.5 simulations,respectively.The choice of fitting functions was consistent with the choice in observation,and the uncertainties of FARanthropogenicand FAR future were estimated via bootstrapping 1000 times.

3.Results

3.1. Flash drought during 2020 in the southeastern coastal region of China

Fig.1 depicts the overall process of evolution of the flash drought case in 2020 in the southeastern coastal region of China employed in this study.Before its onset (9–13 June),although near-normal conditions of SM and precipitation prevailed across the study region,the strong evapotranspiration caused by the unusually high temperature (greater than 80%) provided favorable conditions for flash drought onset.During the drought onset stage (14–23 June),the combination of an extreme deficit of precipitation (less than 20%),intensified evapotranspiration,and persistently high temperature (greater than 90%),led to a rapid decline in the SM percentile from above 50% to below 30% within 10 days across the region.Here,the strong evapotranspiration was mainly caused by the unusually high temperature.Subsequently,the drought condition was persistent (24 June to 2 August),and the SM percentile was even lower than 5%.Meanwhile,the precipitation deficit and high temperature still maintained,while the evapotranspiration gradually decreased and changed into water control due to the low SM.From 3 August to 11 September (recovery stage),with the occurrence of precipitation and decrease in temperature,the flash drought gradually recovered and the SM percentile finally increased to above 50% during 12–16 September.

Fig.1.Evolution of the 2020 flash drought in the southeastern coastal region of China.Selected pentad-mean percentiles of (a) soil moisture (SM),(b) precipitation(P),(c) evapotranspiration (ET),and (d) temperature (T) during June–September 2020 are shown for illustrating the drought onset,persistence,and recovery processes.The percentiles were calculated based on the 1961–2020 climatology.The black rectangles represent the study area.

Land–atmosphere coupling and large-scale atmospheric circulation anomalies are of great significance for the evolution of hydrometeorological extremes (Seneviratne et al.,2006 ;Teuling et al.,2010).According to Fig.2,HI showed a positive anomaly and the study region was controlled by anomalously high pressure before flash drought onset,which means that the atmosphere was drier and hotter than normal.During the onset and persistence stages,the intensified anomalously high pressure increased the downward solar radiation at the surface,which stimulated the enhancement of sensible heat transport between the land surface and atmosphere.Under this condition,the combination of positive CTP and HI anomalies caused by warm–dry air entrainment was able to inhibit convective precipitation through land–atmosphere dry coupling,and the corresponding increase in atmospheric evapotranspiration demand due to the dry atmosphere was also conducive for the rapid decline in SM.During the recovery stage,the center of the anomalously high pressure moved southward and the hot weather was relieved.Furthermore,accompanying the water vapor transport,the ABL became moist and convective precipitation occurred.

Fig.2.Standardized anomalies (z_score) of mean (a) convective triggering potential (CTP) and (b) low-level humidity index (HI) during flash drought in 2020.From top to bottom represents the anomalies before onset,during onset,and in the persistence and recovery stages,respectively.(c) Mean 500-hPa geopotential height(shading;units: gpm) and 850-hPa winds (vectors;units: m s-1) anomalies at different stages,and the arrows represent the wind speed and direction.The anomalies are relative to the 1961–2020 climatology,and the black and blue rectangles represent the study area.

According to the SM percentile,precipitation and temperature from 1961 to 2020,we calculated the return periods for the joint distribution of drought onset speed and intensity (Fig.3 (a)) of the 2020 flash drought,along with the joint distribution of precipitation and temperature during both the drought onset (14–23 June) and whole drought(14 June to 11 September) periods (Fig.3 (b,c)) by using the copula method.The joint return period for the drought onset speed and intensity like those in the 2020 flash drought is about once in 2454 years.Correspondingly,the joint return period for the mean precipitation and temperature during the drought onset and whole drought periods are once in 724 years and once in 377 years,respectively.On the whole,the precipitation deficit,strong evapotranspiration,and high temperature caused by land–atmosphere dry coupling and anomalously high pressure stimulated the record-breaking rapid onset and intensity of flash drought in 2020.

Fig.3.Concurrent return periods of soil moisture (SM) and meteorological conditions in the southeastern coastal region of China during the identified 2020 flash drought period (i.e.,14 June to 11 September) from 1961 to 2020.(a) Concurrent return periods of the regional-mean SM rate of decline and deficit intensity(negative values represent wet conditions),and (b) regional mean temperature (T) and precipitation (P) during the 2020 flash drought onset period (14–23 June).(c) Concurrent return periods of regional mean T and P during the whole drought period (14 June to 11 September).The blue dots represent the historical results during 1961–2019,and the red dots indicate the results for the year 2020.The black isolines are the return period levels calculated with (a) Clayton and (b,c)Plackett copula functions.

3.2. Attribution of the flash drought and assessment of its future risk

For the models used in this study,considering the inconsistency in soil depth and the limitation in regional simulation (IPCC,2021),all the full-column soil moisture datasets were processed into percentiles,making the comparison among models and across regions as fair as possible,and the results of the K-S test between CMIP6/ALL simulations and ERA5 means that they have a similar distribution at the 95%confidence level (p＞0.05;Table S1) and can be used in the attribution analysis and future risk estimation.Comparing the probability density functions (PDFs) of the decline in the SM percentile,there is a significant increase for the tail values from NAT to ALL and then to SSP2-4.5 (Fig.4 (a)).The results of FAR anthropogenic and FAR future show that the probability of a drought onset speed like that in 2020 has increased by about 93% ± 20% (95% confidence interval (CI)) under the influence of anthropogenic climate change,and such a probability will further increase by about 85% ± 33% (95% CI) in the future.From the PDFs of the difference between 40% and the mean SM percentile,there is a significant movement to drier conditions (Fig.4 (b)).Also,the probability of a 2020-like drought intensity has increased by about 18% ± 15% under the influence of anthropogenic climate change,and such a probability will further increase by about 49% ± 8% in the future.

Correspondingly,the joint probability for the onset speed and intensity of a flash drought like that during 2020 has also increased,by 86%± 38%,due to anthropogenic climate change,and the risk will increase by 81% ± 48% in the future.Furthermore,the joint cumulative distribution functions (CDFs,Fig.4 (c)) show that such severe and rapid-onset drought events will become more frequent in the future,which increases the risk of drought adaptation.

4.Conclusions and discussions

The southeastern coastal region of China experienced a recordbreaking flash drought from mid-June to mid-September in 2020.Through analyzing the land–atmosphere coupling and large-scale atmospheric circulation at different stages of the flash drought,we found that the suppression of convective precipitation and the high temperature caused by the anomalously high pressure and land–atmosphere dry coupling were important reasons for the rapid onset and strong intensity of this flash drought event.Moreover,attribution analysis was conducted based on CMIP6 data under NAT and ALL scenarios and the results showed that anthropogenic climate change has not only increased the likelihood of an onset speed and intensity like those in the 2020 flash drought event,by about 93% ± 20% and 18% ± 15%,respectively,but also increased their simultaneous occurrence,by about 86% ± 38%.Through comparing the onset speed and intensity of flash drought in historical ALL simulation and future SSP2-4.5 projection,we found that the likelihood of such an onset speed,intensity,and their simultaneous occurrence will further increase in the future,making drought prediction and early warning more challenging.

Fig.4.Historical attribution and future projection of flash drought onset speed and intensity in the southeastern coastal region of China.(a) Gaussian distributions of flash drought onset speed.(b) Gaussian distributions of the flash drought intensity.(c) Plackett copula distributions of the flash drought onset speed and intensity.The probability density functions (PDFs) and cumulative distribution functions (CDFs) were estimated from the CMIP6 ensemble simulations under natural-only forcings (NAT),all forcings (ALL),and business-as-usual (SSP2-4.5) future scenarios.The black lines show the values of the corresponding variables in 2020.

During the 2020 extreme flash drought event,the strong evapotranspiration driven by the high temperature was very important for the rapid onset of flash drought,while the weak evapotranspiration under the control of low SM may be the main reason for the slowdown of SM decline in the persistence stage.Therefore,the point at which evapotranspiration changes from energy control into water control is vital for the evolution of flash drought and needs further research (Yuan et al.,2020).Furthermore,considering that the summer precipitation of the southeastern coastal region of China is strongly influenced by the East Asian summer monsoon (EASM;Ding et al.2008,2009),exploring the variation characteristics of the EASM against the background of climate change is also of great significance,to understand the physical mechanism of drought events in the region.

Funding

This work was supported by the National Natural Science Foundation of China [grant number 41875105 ],the National Key R&D Program of China [grant number 2018YFA0606002 ],and the Natural Science Foundation of Jiangsu Province for Distinguished Young Scholars[grant number BK20211540 ].