1. Introduction
Interactions between the land and the atmosphere have a broad and great impact on the changes of water, carbon, and extreme climates, especially in hot spots of land–atmosphere coupling (Dirmeyer et al. 2021; Humphrey et al. 2021; Seneviratne et al. 2006; Teuling 2018). Changes in land surface conditions can influence the climate through their effects on latent heat flux (LE) and sensible heat flux (H), and the subsequent effects of these energy fluxes on heat and moisture in the atmospheric boundary layer (Abdolghafoorian and Dirmeyer 2021; Dirmeyer 2011; Gentine et al. 2013). Earth’s climate has warmed unequivocally in the past century, and temperature hot extreme associated with global warming has also risen and may continue to increase in the future (Lewis et al. 2019). Both observations and climate model simulations show that the rate of extreme temperature rise can differ from the rate of mean temperature rise, and the relationship between the two rates varies spatially (Seneviratne et al. 2016; Donat et al. 2017; Vogel et al. 2017). In Europe and other midlatitude regions where dense cities are located, the local rate of increase in hot extreme exceeds the rate of increase in mean temperature (Meehl and Tebaldi 2004; Teuling 2018). Such regional warming of hot extreme can be associated with more frequent extreme weather/climate events and more destructive drought and heatwave events, and severe effects on terrestrial ecosystems and human society (Li et al. 2018; Vasseur et al. 2014).
Previous studies report that land–atmosphere feedback and the partitioning of LE and H have profound impacts on the changes of drought and hot extreme especially in the climate transitional zones (Teuling 2018; Vogel et al. 2017). When water availability in the soil is limited for evapotranspiration, the change in surface energy balance, i.e., H increases with a decrease of LE, largely determines the influence of soil moisture dryness on local temperature (Gallego-Elvira et al. 2016; Miralles et al. 2014). In certain areas, such as the Mediterranean region, western United States, southeastern Australia, southern Africa, and eastern Russia, increases in extreme high temperature were found to be significant and accompanied by more frequent compound dry and hot events (Zhou et al. 2019; Hao et al. 2020; Wang et al. 2021b). The increase in hot extremes in various regions may be driven by different physical mechanisms. There are many anthropogenic drivers for changes in regional thermodynamic conditions, such as land use/cover change and agricultural management practices (e.g., irrigation) (Lemordant et al. 2018; Liao et al. 2018; Yuan et al. 2020). Changes in those land surface conditions can cause dynamical variations in the partitioning of surface energy fluxes, which in turn feedback on regional air temperature (Forzieri et al. 2020; Gentine et al. 2016; Schwingshackl et al. 2017).
Annual mean temperature increase rate is generally higher than that of the ocean due to the small inertia of the continents on most lands (Byrne and O’Gorman 2013). Yet, the increase in temperature shows great spatial differences on land, with the largest increase in average temperature at the mid- to high-latitude region of the Northern Hemisphere (IPCC 2018, 2019). Several studies pointed out that the increase in hot extremes can be impacted by surface dryness feedback, leading to a sharp increase in hot extremes during the drought period (Seneviratne et al. 2013; Miralles et al. 2019; Teuling 2018). Moreover, existing studies suggest that the increasing trends in hot temperature and heat wave caused by the feedback of soil moisture limitations are independent of natural climate variability (Berg et al. 2016; Hirschi et al. 2011; Vogel et al. 2017). However, existing studies are usually based on climate model simulations or regional short-term observations (Koster et al. 2011; Sippel et al. 2017). As such, key impacts from local and remote human land/water management processes are usually assumed to be static and the models might not have the right land–atmosphere response to soil moisture (Lansu et al. 2020; Lemordant et al. 2016; Lemordant and Gentine 2019). Therefore, the model-simulated accelerated warming of hot extremes may be unreliable and scale dependent. Owing to the lack of long-term, high-quality observational data of surface energy fluxes on a global scale, it is difficult to assess whether model projections show realistic behavior under recent climate change conditions (Abdolghafoorian and Dirmeyer 2021; Seneviratne et al. 2010).
This study aims to provide observational evidence for the increase in hot extremes accelerated by changes in surface energy partitioning. Our research is inspired by the idea that partitioning of LE and H is dynamically modified by various changing environmental factors such as human irrigation and vegetation stomata/leaf area in response to global warming and rising atmospheric CO2 concentrations; however, these changing environmental factors and their impacts tend to be missing from models, which likely leads to inaccurate projections of climate warming and hot extremes (Thiery et al. 2020; Lansu et al. 2020; Lemordant and Gentine 2019). Therefore, it is necessary to identify hot spots of temperature extreme increases and to investigate the cause of hot extremes increasing faster in specific regions than in other regions. We do so using ground meteorological data and ground observation-derived energy fluxes that capture the integrated effects of vegetation dynamic changes and human activities. In a previous study (Wang et al. 2021a), we retrieved long-term LE and H during the 1950–2017 period, using globally distributed weather stations and eddy-covariance flux towers for model training and validation. This approach is informed by multilayer neural network models and constrained by a boundary layer energy budget. Previous studies mainly use soil moisture to explore climate feedback effects from climate models, and thus they might not reflect all changes in the surface energy partitioning due to vegetation physiological change and human land/water management. In this study, for the first time, we observed maximum and average temperatures as well as employed observation-derived LE and H, to map where hot spots of increasing hot extremes have existed over the past few decades, and what physical processes might explain them.
2. Data and methods
a. Dataset
This study uses weather station observational data of 2-m air temperature including daily maximum temperature and daily mean temperature over continents during the period 1975–2017. The meteorological data used in this study were collected from the Global Summary of the Day (GSOD), which was provided by the National Oceanic and Atmospheric Administration (NOAA) (https://www.ncdc.noaa.gov/data-access). We used globally distributed weather stations with the observation records from 1975 to 2017 (Fig. 1b). It is well known that the climate warming trend is more pronounced after the 1970s because of reduced aerosol cooling impact (Nicholls et al. 2020), and thus we focused our analysis after 1975. Meanwhile, the global coverage of weather stations is extensive, and the quality of meteorological observation data is well controlled in this study period.
The spatial distribution of the flux towers and weather stations used in this study.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
The spatial distribution of the flux towers and weather stations used in this study.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
The spatial distribution of the flux towers and weather stations used in this study.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
There are various surface flux products such as the FLUXCOM and the surface flux data driven by model tree ensemble (FLUXNET-MTE), which are based on the satellite remote sensing, reanalysis, and instantaneous meteorological observational data (Jung et al. 2011, 2019; Miralles et al. 2013). These products may not represent the long-term effects of confounding factors, such as changes in atmospheric CO2 or species composition. In this study, for the first time, we use the daily LE and daily H from Wang et al. (2021a), which are estimated by a machine learning approach built from the daily observational data of globally distributed flux towers and weather stations. The machine-learned surface energy fluxes are purely relying on ground-based observations. As such, a major advantage of these retrievals is that they do not rely on any assumption on the link between environmental factors and fluxes. For instance, if the openings of plant stoma were to close, they would increase H and reduce LE, which in turn lead to an increase in 2-m air temperature and a decrease in 2-m humidity (Gentine et al. 2016; Salvucci and Gentine 2013). The artificial neural network used to predict surface fluxes was developed based on the integrated daily data of the FLUXNET2015 (Fig. 1a), and then the weather station daily observations during the 1950–2017 period were used to drive the well-trained model. Top-of-atmosphere shortwave radiation, 2-m mean temperature, 2-m maximum and minimum temperature, 2-m relative humidity, and surface wind speed are the input variables used to train and drive the neural network model. The artificial neural networks algorithm has been shown to be a powerful nonlinear regression tool, and it has good performance in simulating the nonlinear relationship between different environmental variables and energy fluxes (Chen et al. 2020; Haughton et al. 2018). For detailed information on the weather station observational data and the retrieval of the surface energy fluxes, we refer the reader to Wang et al. (2021a).
b. Metrics
For each weather station, we use the 2-m daily mean temperature during 1975–2017 to calculate the long-term trend in local annual average temperature (Tmean) in this study. The daily observations of maximum temperature were also employed to analyze the changes in annual 99th percentile maximum temperature (Tq99th). In this study, the linear trend slopes of Tmean and Tq99th are used to assess the rates of changes in Tmean (ΔTmean) and Tq99th (ΔTq99th) over the 1975–2017 period. Land–atmosphere coupling mainly refers to the feedback of variations in surface processes to the boundary layer conditions and thus soil moisture is usually used to represent the state of the land surface. However, soil moisture is indirectly modifying the atmosphere through changes in evapotranspiration and the surface energy partitioning between latent heat and sensible heat (Donat et al. 2017; Gallego-Elvira et al. 2016; Schwingshackl et al. 2017). Moreover, long-term direct observational data of soil moisture is not available at present, especially on the global and daily scales, and therefore we do not use a soil moisture–based index to perform our analysis. We use the evaporation fraction (EF), the ratio of latent heat to the sum of latent heat and sensible heat fluxes, as an indicator of surface energy partitioning and a proxy of soil moisture variability (Gentine et al. 2007, 2011). Relative to the soil moisture data, the variability of EF contains the signals of vegetation physiological effects, land use/cover change, and the influence of human activities on surface energy partitioning.
c. Statistical methods
Long-term trends of Tq99th, Tmean, LE, H, EF, and Tq99th relative to Tmean are quantified using the linear estimation method. The significance of the trend is tested using the Mann–Kendall (MK) method, a nonparametric diagnosis method (Wang et al. 2017). The point-computed trends in various variables are spatially interpolated to 0.5° × 0.5° latitude/longitude resolution using the kriging spatial interpolation method. Meanwhile, the Pearson’s correlation coefficient (R) is used to evaluate the relationship between EF and extreme temperature.
3. Results analysis
a. Long-term trends in Tq99th, Tmean, LE, and H
Mean temperature shows an upward trend on a global scale, which is consistent with the tendency of climate warming (Fig. 2a). The mean temperature rise is greater in the Northern Hemisphere than in the Southern Hemisphere. Long-term trends in hot extreme (Tq99th) show large spatial difference over the global land surfaces (Fig. 2b). Hot temperature extremes with a more pronounced increase are located in southwestern North America, Europe, southern Africa, northeast Asia, and Australia. In these hot spots, the warm tail of daily hot extreme increases disproportionately with the climate warming trend.
Spatial patterns of long-term trends in the 99th percentile maximum temperature (ΔTq99th) and annual mean temperature (ΔTmean) during the 1975–2017 period. The small gray squares indicate the stations with a significant upward trend tested by the Mann–Kendall (MK) method. Greenland is excluded from the analysis due to its perennial snow and ice coverage.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
Spatial patterns of long-term trends in the 99th percentile maximum temperature (ΔTq99th) and annual mean temperature (ΔTmean) during the 1975–2017 period. The small gray squares indicate the stations with a significant upward trend tested by the Mann–Kendall (MK) method. Greenland is excluded from the analysis due to its perennial snow and ice coverage.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
Spatial patterns of long-term trends in the 99th percentile maximum temperature (ΔTq99th) and annual mean temperature (ΔTmean) during the 1975–2017 period. The small gray squares indicate the stations with a significant upward trend tested by the Mann–Kendall (MK) method. Greenland is excluded from the analysis due to its perennial snow and ice coverage.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
After the solar radiation reaches the surface, a fraction is used in photosynthesis, but most returns to the atmosphere in the form of LE and H. LE mainly controls the moisture exchange in the atmospheric boundary layer, while H controls the daily temperature cycle. Typical areas where H decreases over 1975–2017 include the Amazon basin, North Africa, Southeast Asia, and the eastern coast of Australia, while other regions mainly show a slight increasing trend (Fig. 3a). Note that the density of weather stations in the Amazon region and North Africa is relatively low, which is likely to contribute uncertainty in our spatial interpolation. The LE shows a downward trend in southwestern North America, eastern Amazon, North Africa, and west Asia (Fig. 3b). The areas with significant increasing trends in LE are mainly located near the equator. On the one hand, this is mainly attributed to the increase in atmospheric temperature and the increase in vapor pressure deficit caused by global warming, which provides an energy source for the increasing trend in LE. On the other hand, the tropical and subtropical regions have sufficient water supply, and precipitation over the period of analysis is also increasing.
Long-term trends in the sensible heat flux (ΔH) and the latent heat flux (ΔLE) during the 1975–2017 period. The small gray squares indicate the stations with a significant trend (p < 0.05).
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
Long-term trends in the sensible heat flux (ΔH) and the latent heat flux (ΔLE) during the 1975–2017 period. The small gray squares indicate the stations with a significant trend (p < 0.05).
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
Long-term trends in the sensible heat flux (ΔH) and the latent heat flux (ΔLE) during the 1975–2017 period. The small gray squares indicate the stations with a significant trend (p < 0.05).
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
b. Trends in Tq99th relative to Tmean
To identify the hot spots where the warming rate of temperature hot extreme is larger than the trend in local annual mean temperature, we derive the ratio ΔTq99th/ΔTmean at all weather stations. To avoid the inflation of the quotient, we masked the areas with ΔTmean ranging from −0.01 to 0.01 in the analysis of ΔTq99th/ΔTmean (Fig. 4a). These regions largely correspond to regions with reduced EF (Fig. 4b). The spatial patterns of trend in annual Tq99th relative to the trend in local annual Tmean (ΔTq99th/ΔTmean) reveal hot spot regions where temperature hot extreme increases rapidly (Fig. 4a). Southwestern North America, Europe, southern Africa, northeast Asia, and southeastern Australia show an obvious upward trend in temperature hot extreme compared to the mean. Note the long-term trends in Tmean are close to zero in areas of Australia, Southeast Asia, and India, although the trends in Tq99th are also less than 1 (Fig. 2). In the case of declined EF, Europe, southwest North America, southern Africa, and northeast Asia are hot spots with a faster upward rate in temperature extreme than local annual average. This decrease in EF accompanied by increased ΔTq99th/ΔTmean were typically observed in northern Africa, southern Australia, and the Amazon region. Yet, it needs to be pointed out that the pattern might be also influenced by the observational quality and density of weather stations, such as in the Amazon. A significant trend in ΔTmax/ΔTmean is not observed here, while the modeled results indicate that the Amazon basin is a hot spot area with increasing hot extreme (Donat et al. 2017). These differences indicate that there is a need to reconcile local weather observational results and model simulations in the future, especially in the areas with strong convection. In addition, some other regions including the tropical areas around the equator, Southeast Asia, the east and west coasts of Africa, and the northern part of South America present a downward trend in ΔTq99th/ΔTmean and exhibit a change rate of hot extreme that are less than the rate of mean temperature increases (Fig. 4).
(a) Long-term trends in the 99th percentile temperature (ΔTq99th) relative to the trends in annual mean surface temperatures (ΔTmean) and (b) long-term trends in annual evaporative fraction (ΔEF) during the 1975–2017 period. Areas with ΔTmean from −0.01 to 0.01 that are close to 0 are masked using green polygon in (a). The small gray squares in (b) indicate the stations with a significant trend at the 0.05 level. Dashed boxes indicate the focus regions for further analysis.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
(a) Long-term trends in the 99th percentile temperature (ΔTq99th) relative to the trends in annual mean surface temperatures (ΔTmean) and (b) long-term trends in annual evaporative fraction (ΔEF) during the 1975–2017 period. Areas with ΔTmean from −0.01 to 0.01 that are close to 0 are masked using green polygon in (a). The small gray squares in (b) indicate the stations with a significant trend at the 0.05 level. Dashed boxes indicate the focus regions for further analysis.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
(a) Long-term trends in the 99th percentile temperature (ΔTq99th) relative to the trends in annual mean surface temperatures (ΔTmean) and (b) long-term trends in annual evaporative fraction (ΔEF) during the 1975–2017 period. Areas with ΔTmean from −0.01 to 0.01 that are close to 0 are masked using green polygon in (a). The small gray squares in (b) indicate the stations with a significant trend at the 0.05 level. Dashed boxes indicate the focus regions for further analysis.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
Hot extreme trend relative to mean temperature trend may differ across seasons, as latent heat is a more important cooling mechanism in the summer so that increased dryness may intensify hot extremes more obvious and frequent in the summer. Indeed, temperature hot extremes on land usually occurs in local summer (or dry season in the tropics), and thus the changes in hot extreme can be closely related to the change in summer conditions (Argüeso et al. 2016). Therefore, we compared the relation between the trend in hot extreme and local summer average temperature (Fig. 5). Hot extremes usually occur in June–August (JJA) in the Northern Hemisphere and hot extremes in the Southern Hemisphere usually occur in December–February (DJF). When focusing on trends in Tq99th and Tmean on local summer (Figs. 5a,b), temperature hot extreme shows accelerated warming trend at hot spots, and the corresponding EF pattern shows a significant decline trend. Consistent with the signals in the annual trend, the summer temperature hot extreme shows an accelerated warming trend relative to the summer average temperature, especially in Europe, western North America, and southern Africa. This is not the case in other regions such as northeast China and west Asia, where the summer hot extreme rise does not significantly exceed the annual average temperature. Although southern Australia also exhibits a faster rate of increase in hot extreme than the average, the corresponding change in EF is relatively weak. Therefore, recently reported hot weather or heatwave events in Australia may be associated with atmospheric circulation (Wang and Zhang 2017), rather than induced or dominated by drought event and local land–atmosphere interactions.
(a),(b) The spatial patterns of trends in hot extreme (ΔTq99th) relative to the trends in mean summer temperature (ΔTmean) and (c),(d) the spatial patterns of trends in mean summer EF during the period of 1975–2017 [average on June–August (JJA) in the Northern Hemisphere and average on December–February (DJF) in the Southern Hemisphere]. Areas with ΔTmean from −0.01 to 0.01, which are close to 0, are masked using a green polygon in (a) and (b). The small gray squares in (c) and (d) indicate the stations with a significant trend at the 0.05 level.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
(a),(b) The spatial patterns of trends in hot extreme (ΔTq99th) relative to the trends in mean summer temperature (ΔTmean) and (c),(d) the spatial patterns of trends in mean summer EF during the period of 1975–2017 [average on June–August (JJA) in the Northern Hemisphere and average on December–February (DJF) in the Southern Hemisphere]. Areas with ΔTmean from −0.01 to 0.01, which are close to 0, are masked using a green polygon in (a) and (b). The small gray squares in (c) and (d) indicate the stations with a significant trend at the 0.05 level.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
(a),(b) The spatial patterns of trends in hot extreme (ΔTq99th) relative to the trends in mean summer temperature (ΔTmean) and (c),(d) the spatial patterns of trends in mean summer EF during the period of 1975–2017 [average on June–August (JJA) in the Northern Hemisphere and average on December–February (DJF) in the Southern Hemisphere]. Areas with ΔTmean from −0.01 to 0.01, which are close to 0, are masked using a green polygon in (a) and (b). The small gray squares in (c) and (d) indicate the stations with a significant trend at the 0.05 level.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
The ensemble simulations from Coupled Model Intercomparison Project phase 5 models (CMIP5) are different with some of the observation-driven ΔTq99th/ΔTmean patterns in this study. Donat et al. (2017) showed that the pattern of increase in hot extreme relative to the local average is widespread across all land surfaces, between the mid twentieth century (1951–80) and the late twenty-first century (2070–99). Beside inherent climate variability, this decrease in extremes is present in regions of large agricultural expansion, such as the U.S. Midwest (Mueller et al. 2016), which is a well-known warming hole region due to the increase in evapotranspiration induced by agricultural expansion (Thiery et al. 2017, 2020). Indeed, land use/cover change can strongly affect local warming or mitigate some of the impact of climate change (Mahmood et al. 2010). The lower increase in Tq99th compared to Tmean may be associated to the increasing frequency of extreme precipitation event driven by climate warming, especially in the humid regions of the tropics and subtropics where the LE increase and the H decrease (Fig. 3). As the climate warms, the water vapor contained in the atmosphere is generally considered to follow the Clausius–Clapeyron relation that the atmosphere can hold 7% more moisture for 1°C temperature increase, leading to an increase in the frequency of extreme precipitation event on most of the global land regions (Ban et al. 2015; Morrison et al. 2019). Over the past decades, the frequency of extreme precipitation event has shown an overall increasing trend, potentially mitigating the magnitude of increase in temperature hot extreme (Papalexiou and Montanari 2019; Wang et al. 2017; Yin et al. 2021).
c. Relationship between Tq99th and EF
All regional mean trends of Tmean and Tq99th of the four focus regions present an increase trend during 1975–2017, and the magnitude of increase in the Tq99th is greater than the magnitude of increase in Tmean (Table 1). Most stations in Europe show a significant upward trend in temperature and a significant downward trend in EF, and thus Europe may have been affected by a wide range of temperature extreme recently. Among the four regions, the trend of Tmean shows the largest upward trend in southwest North America, i.e., 0.0092°C yr−1. If the accuracy of the measuring thermometer is ±0.2°C, this trend can be detected in 22-yr record length data at a site. The trend of Tq99th shows the largest upward trend in southern Africa, i.e., 0.0278°C yr−1. On the contrary, EF presents a downward trend in the four regions, and southern Africa has the largest decline trend, followed by south west North America. If the reason for the warming of Tmax in the hot spots in Figs. 4 and 5 is related to local land–atmosphere feedback processes, there should be a significant negative correlation between ΔTq99th/ΔTmean and EF (Seneviratne et al. 2013). On most land regions, EF shows a negative correlation with Tq99th and Tmean, especially in the “hot spot” regions (Fig. 6). In southwest North America, Europe, southern Africa, and northeast Asia, the correlation between EF and Tq99th (correlation EF–Tq99th) exhibit significant negative correlations, which suggest that strong negative correlation between the EF and the Tq99th are closely associated with the increasing temperature hot extreme. During the dry season, dependence between temperature extremes and surface energy fluxes potentially poses more intense compound dry and hot events in the hot spots (Zscheischler and Seneviratne 2017).
(a) Spatial pattern of the Pearson correlation coefficients between EF and Tq99th, and (b) spatial pattern of the Pearson correlation coefficients between EF and Tmean. The gray squares indicate the stations with a significant correlation at the p < 0.001 level.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
(a) Spatial pattern of the Pearson correlation coefficients between EF and Tq99th, and (b) spatial pattern of the Pearson correlation coefficients between EF and Tmean. The gray squares indicate the stations with a significant correlation at the p < 0.001 level.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
(a) Spatial pattern of the Pearson correlation coefficients between EF and Tq99th, and (b) spatial pattern of the Pearson correlation coefficients between EF and Tmean. The gray squares indicate the stations with a significant correlation at the p < 0.001 level.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
Summary of the trends in Tmean, Tq99th, and EF for the four focus regions. NSUT represents the number of stations with a significant upward trend at the 0.05 level, and NSDT represents the number of stations with a significant downward trend at the 0.05 level. TN is the total number of weather stations. The trends are tested using the Mann–Kendall (MK) method.
A negative correlation between ΔTq99th/ΔTmean and ΔEF is detected in the four hot spots and both (negative) correlations are significant at the p < 0.05, although the p value of the negative correlation is relatively weak over southern Africa (Fig. 7). The negative correlations are significant (p < 0.05) in southwest North America (R = −0.27, p = 0.003), southern Africa (R = −0.29, p = 0.049), and northeast Asia (R = −0.17, p = 0.032), while in Europe this negative correlation (R = −0.26) is significant at the p < 0.001 level (Fig. 7). Moreover, a significant negative correlation (R = −0.18, p < 0.001) is also detected in the trends in ΔTq99th/ΔTmean and the trends in the correlation between EF and Tq99th, across all weather stations with a significant (p < 0.001) negative correlation between EF and Tq99th (Fig. 8). Therefore, the observation-driven findings emphasize that the large-scale impact of land–atmosphere interactions on hot extreme, in which a decline in EF that associated with a decrease in soil moisture typically can accelerate the regional warming of hot extreme. Those climate feedback effects of changes in the surface energy partitioning are significant over North America and Europe. However, the negative relationship between ΔEF and ΔTq99th/Tmean does not always exist. Over central North America, the observed scaling rates are relatively small, indicating that hot extreme is increasing at a slower rate than the temperature average. Consistent with previous studies (Donat et al. 2013; Mueller et al. 2016; Portmann et al. 2009; Tavakol et al. 2020), this is the “cooling” region of the midwest area of United States partly due to human land management or agricultural expansion. In the Amazon, significant downward trend in EF and significant negative correlation between EF and the Tq99th have not been found in our observation-driven results. However, these trends and significant negative correlations have been predicted in climate models (Donat et al. 2017). Only a few partial areas with significant negative correlations between EF and Tq99th (Tmean) are detected in the eastern part of the Amazon (Fig. 6). In this region, the increasing evapotranspiration caused by the warming climate has likely offset the increase rate in extreme high temperature driven by land–atmosphere interactions, mainly due to the limited current heat stress on canopy flux (Gentine et al. 2019; Green et al. 2020).
Scatterplot for the trend in annual mean evaporative fraction (ΔEF; decade−1) and the trend in maximum temperature relative to the trend in local mean temperature (ΔTq99th/ΔTmean; °C/°C) in the four hot spot regions. All trends are calculated for the period 1975–2017. Each dot represents one weather station and N is the number of stations. Solid red lines indicate the linear regression between ΔEF and ΔTq99th/ΔTmean. Here, R is the Pearson correlation; the p value is the significance level for the correlation analysis.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
Scatterplot for the trend in annual mean evaporative fraction (ΔEF; decade−1) and the trend in maximum temperature relative to the trend in local mean temperature (ΔTq99th/ΔTmean; °C/°C) in the four hot spot regions. All trends are calculated for the period 1975–2017. Each dot represents one weather station and N is the number of stations. Solid red lines indicate the linear regression between ΔEF and ΔTq99th/ΔTmean. Here, R is the Pearson correlation; the p value is the significance level for the correlation analysis.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
Scatterplot for the trend in annual mean evaporative fraction (ΔEF; decade−1) and the trend in maximum temperature relative to the trend in local mean temperature (ΔTq99th/ΔTmean; °C/°C) in the four hot spot regions. All trends are calculated for the period 1975–2017. Each dot represents one weather station and N is the number of stations. Solid red lines indicate the linear regression between ΔEF and ΔTq99th/ΔTmean. Here, R is the Pearson correlation; the p value is the significance level for the correlation analysis.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
Scatterplot for the correlation EF–Tq99th vs ΔTq99th/ΔTmean during the 1975–2017 period. The black dots represent all global weather stations with a significant negative correlation between EF and Tq99th (p < 0.001). Each dot represents one weather station, and N is the number of stations. Solid red lines show the linear regression.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
Scatterplot for the correlation EF–Tq99th vs ΔTq99th/ΔTmean during the 1975–2017 period. The black dots represent all global weather stations with a significant negative correlation between EF and Tq99th (p < 0.001). Each dot represents one weather station, and N is the number of stations. Solid red lines show the linear regression.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
Scatterplot for the correlation EF–Tq99th vs ΔTq99th/ΔTmean during the 1975–2017 period. The black dots represent all global weather stations with a significant negative correlation between EF and Tq99th (p < 0.001). Each dot represents one weather station, and N is the number of stations. Solid red lines show the linear regression.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
4. Discussion
It needs to be emphasized that the statistical analyses also have inherent limitations as a strong correlation does not represent causality, because there are other mechanisms that affect extreme climate change. Indeed, there are internal connections between surface energy flux and various meteorological elements, including solar radiation, temperatures, humidity, and wind speed, and these connections are extremely nonlinear. LE and H are two independent variables predicted by their respective artificial neural network models. The correlation analysis in this study was performed for trends. As such, the analysis was not performed based on the original data series of surface flux and meteorological data. In our previous study (Wang et al. 2021a), we found that the retrieval of LE is sensitive to top-of-atmosphere solar radiation (SW_IN_POT), relative humidity (RH), and surface wind speed (WS), and the retrieval of H is sensitive to SW_IN_POT, Tmax, and minimum temperature (Tmin). Therefore, air temperature is not the determined factor in the model, and this is consistent with our theoretical basis that the responses of surface fluxes to changing environments are extremely nonlinear. If temperatures are the only determined factors in the model, the LE and H would not present a large spatial difference across the global land. Our observation-driven results show that regional warming of hot extremes coincides with inferred changes in the surface energy partitioning. The findings are important and provide new insight into the potential role of land–atmosphere feedbacks in impacting hot extreme.
To further provide evidence and to avoid using the EF that contain the same data source with Tmax in the correlation analysis, we designed an experiment that the ANN model for estimating LE and H was trained using SW_IN_POT, RH, Tmean, and WS as inputs (Table S1 in the online supplemental material). As such, it avoids using the Tmax data to retrieve surface energy fluxes and perform subsequent correlation analysis between ΔEF and ΔTq99th. It is shown that the model performance can be used to estimate daily LE and H (Fig. S1). Overall, similar results are also found in case Tmax and Tmin are not used as input variables to the machine learning model. Significant negative correlations between ΔEF and ΔTq99th at both p < 0.001 and p < 0.05 levels are detected in the four hot spots, especially over Europe, southwestern North America, and northeast Asia (Fig. 9). Therefore, the observation-driven results in this study have effectively identified the regions where land–atmosphere coupling has significantly accelerated the increasing hot extreme.
Spatial distribution of the significant negative correlations between ΔEF and ΔTq99th. The LE and H used to calculate EF are retrieved by the ANN model using SW_IN_POT, RH, Tmean, and WS as input. This ANN model does not use Tmax and Tmin as input variables as compared with the previous one.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
Spatial distribution of the significant negative correlations between ΔEF and ΔTq99th. The LE and H used to calculate EF are retrieved by the ANN model using SW_IN_POT, RH, Tmean, and WS as input. This ANN model does not use Tmax and Tmin as input variables as compared with the previous one.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
Spatial distribution of the significant negative correlations between ΔEF and ΔTq99th. The LE and H used to calculate EF are retrieved by the ANN model using SW_IN_POT, RH, Tmean, and WS as input. This ANN model does not use Tmax and Tmin as input variables as compared with the previous one.
Citation: Journal of Hydrometeorology 23, 3; 10.1175/JHM-D-21-0114.1
This study is remarkable because it offers an initial ground observation-based assessment of the relationship between hot temperature extremes and surface energy flux partitioning, globally. The results are partly different from CMIP5-based findings and underscore the importance of accounting for the influences of vegetation physiologic changes and human water/land management on surface water and heat fluxes. There is a need to reconcile local observations and model simulations in the future, especially in tropical areas with intense convection such as the Amazon. Furthermore, the regional representativeness of ground observations used need to be better understood. Besides, the analysis remains to be limited by the spatial density of available ground stations. Extending model validation beyond the limited distribution of in situ weather stations and flux towers will help reduce the uncertainty of Earth system model predictions.
5. Conclusions
In this study, we provided observational evidence of regional increasing hot extreme accelerated by the partitioning of surface energy fluxes, using ground-based weather observations and machine-learned surface energy fluxes. For the first time, the observation-driven LE and H were used to quantify the relationship between changes in surface energy partitioning and temperature hot extreme during the 1975–2017 period. On a global scale, we identified the hot spots where the change rate of hot extreme exceeded the change rate of local average temperature. Main concluding remarks are summarized as follows.
-
The hot spot regions that annual/summer temperature hot extreme increased faster than local temperature average were identified to be southwest North America, Europe, northeast Asia, and southern Africa.
-
Our observation-driven findings show regional warming of hot extremes coincides with inferred changes in the surface energy partitioning. All hot spots exhibit significant negative correlations between ΔEF and ΔTq99th/ΔTmean, implying that the surface energy partitioning potentially played an important role in the regional warming of hot extreme.
-
The detected hot spot regions are likely to face more frequent and intense compound dry and hot extremes under local land–atmosphere feedback. Our observation-driven findings have great implications in providing observational evidence for regional warming of hot extreme accompanied by a change in surface energy fluxes partitioning.
Acknowledgments.
This work was financially supported by the National Key Research and Development Program of China (2017YFA0603603) and the China Postdoctoral Science Foundation (2020M681656). We sincerely thank the three anonymous reviewers for their valuable and enlightening comments on our manuscript.
Data availability statement.
The data that support the findings of this study are available upon request from the authors.
REFERENCES
Abdolghafoorian, A., and P. A. Dirmeyer, 2021: Validating the land–atmosphere coupling behavior in weather and climate models using observationally-based global products. J. Hydrometeor., 22, 1507–1523, https://doi.org/10.1175/JHM-D-20-0183.1.
Argüeso, D., A. Di Luca, S. E. Perkins-Kirkpatrick, and J. P. Evans, 2016: Seasonal mean temperature changes control future heat waves. Geophys. Res. Lett., 43, 7653–7660, https://doi.org/10.1002/2016GL069408.
Ban, N., J. Schmidli, and C. Schär, 2015: Heavy precipitation in a changing climate: Does short-term summer precipitation increase faster? Geophys. Res. Lett., 42, 1165–1172, https://doi.org/10.1002/2014GL062588.
Berg, A., and Coauthors, 2016: Land-atmosphere feedbacks amplify aridity increase over land under global warming. Nat. Climate Change, 6, 869–874, https://doi.org/10.1038/nclimate3029.
Byrne, M. P., and P. A. O’Gorman, 2013: Link between land-ocean warming contrast and surface relative humidities in simulations with coupled climate models. Geophys. Res. Lett., 40, 5223–5227, https://doi.org/10.1002/grl.50971.
Chen, Z. J., Z. C. Zhu, H. Jiang, and S. J. Sun, 2020: Estimating daily reference evapotranspiration based on limited meteorological data using deep learning and classical machine learning methods. J. Hydrol., 591, 125286, https://doi.org/10.1016/j.jhydrol.2020.125286.
Dirmeyer, P. A., 2011: The terrestrial segment of soil moisture climate coupling. Geophys. Res. Lett., 38, L16702, https://doi.org/10.1029/2011GL048268.
Dirmeyer, P. A., G. Balsamo, E. M. Blyth, R. Morrison, and H. M. Cooper, 2021: Land-atmosphere interactions exacerbated the drought and heatwave over northern Europe during summer 2018. AGU Adv., 2, e2020AV000283, https://doi.org/10.1029/2020AV000283.
Donat, M. G., and Coauthors, 2013: Updated analyses of temperature and precipitation extreme indices since the beginning of the twentieth century: The HadEX2 dataset. J. Geophys. Res. Atmos., 118, 2098–2118, https://doi.org/10.1002/jgrd.50150.
Donat, M. G., A. J. Pitman, and S. I. Seneviratne, 2017: Regional warming of hot extremes accelerated by surface energy fluxes. Geophys. Res. Lett., 44, 7011–7019, https://doi.org/10.1002/2017GL073733.
Forzieri, G., and Coauthors, 2020: Increased control of vegetation on global terrestrial energy fluxes. Nat. Climate Change, 10, 356–362, https://doi.org/10.1038/s41558-020-0717-0.
Gallego-Elvira, B., C. M. Taylor, P. P. Harris, D. Ghent, K. L. Veal, and S. S. Folwell, 2016: Global observational diagnosis of soil moisture control on the land surface energy balance. Geophys. Res. Lett., 43, 2623–2631, https://doi.org/10.1002/2016GL068178.
Gentine, P., D. Entekhabi, A. Chehbouni, G. Boulet, and B. Duchemin, 2007: Analysis of evaporative fraction diurnal behaviour. Agric. For. Meteor., 143, 13–29, https://doi.org/10.1016/j.agrformet.2006.11.002.
Gentine, P., J. Polcher, and D. Entekhabi, 2011: Harmonic propagation of variability in surface energy balance within a coupled soil-vegetation-atmosphere system. Water Resour. Res., 47, 143–158, https://doi.org/10.1029/2010WR009268.
Gentine, P., A. A. M. Holtslag, F. D’Andrea, and M. Ek, 2013: Surface and atmospheric controls on the onset of moist convection over land. J. Hydrometeor., 14, 1443–1462, https://doi.org/10.1175/JHM-D-12-0137.1.
Gentine, P., A. Chhang, A. Rigden, and G. Salvucci, 2016: Evaporation estimates using weather station data and boundary layer theory. Geophys. Res. Lett., 43, 11 661–11 670, https://doi.org/10.1002/2016GL070819.
Gentine, P., A. Massmann, B. R. Lintner, S. H. Alemohammad, R. Fu, J. K. Green, D. Kennedy, and J. V.-G. de Arellano, 2019: Land–atmosphere interactions in the tropics–a review. Hydrol. Earth Syst. Sci., 23, 4171–4197, https://doi.org/10.5194/hess-23-4171-2019.
Green, J. K., J. Berry, P. Ciais, Y. Zhang, and P. Gentine, 2020: Amazon rainforest photosynthesis increases in response to atmospheric dryness. Sci. Adv., 6, eabb7232, https://doi.org/10.1126/sciadv.abb7232.
Hao, Z., W. Li, V. P. Singh, Y. Xia, X. Zhang, and F. Hao, 2020: Impact of dependence changes on the likelihood of hot extremes under drought conditions in the United States. J. Hydrol., 581, 124410, https://doi.org/10.1016/j.jhydrol.2019.124410.
Haughton, N., G. Abramowitz, and A. J. Pitman, 2018: On the predictability of land surface fluxes from meteorological variables. Geosci. Model Dev., 11, 195–212, https://doi.org/10.5194/gmd-11-195-2018.
Hirschi, M., and Coauthors, 2011: Observational evidence for soil-moisture impact on hot extremes in southeastern Europe. Nat. Geosci., 4, 17–21, https://doi.org/10.1038/ngeo1032.
Humphrey, V., A. Berg, P. Ciais, P. Gentine, M. Jung, M. Reichstein, S. I. Seneviratne, and C. Frankenberg, 2021: Soil moisture-atmosphere feedback dominates land carbon uptake variability. Nature, 592, 65–69, https://doi.org/10.1038/s41586-021-03325-5.
IPCC, 2018: Global warming of 1.5°C: An IPCC special report on the impacts of global warming of 1.5°C above pre-industrial levels and related global greenhouse gas emission pathways, in the context of strengthening the global response to the threat of climate change, sustainable development, and efforts to eradicate poverty. V. Masson-Delmotte et al., Eds, IPCC, 616 pp., https://www.ipcc.ch/sr15/.
IPCC, 2019: Climate change and land: An IPCC special report on climate change, desertification, land degradation, sustainable land management, food security, and greenhouse gas fluxes in terrestrial ecosystems. P. R. Shukla et al., Eds., IPCC, 896 pp., https://www.ipcc.ch/srccl/.
Jung, M., and Coauthors, 2011: Global patterns of land-atmosphere fluxes of carbon dioxide, latent heat, and sensible heat derived from eddy covariance, satellite, and meteorological observations. J. Geophys. Res. Biogeosci., 116, G00J07, https://doi.org/10.1029/2010JG001566.
Jung, M., and Coauthors, 2019: The FLUXCOM ensemble of global land-atmosphere energy fluxes. Sci. Data, 6, 74, https://doi.org/10.1038/s41597-019-0076-8.
Koster, R. D., and Coauthors, 2011: The second phase of the Global Land–Atmosphere Coupling Experiment: Soil moisture contributions to subseasonal forecast skill. J. Hydrometeor., 12, 805–822, https://doi.org/10.1175/2011JHM1365.1.
Lansu, E. M., C. C. van Heerwaarden, A. I. Stegehuis, and A. J. Teuling, 2020: Atmospheric aridity and apparent soil moisture drought in European forest during heat waves. Geophys. Res. Lett., 47, e2020GL087091, https://doi.org/10.1029/2020GL087091.
Lemordant, L., and P. Gentine, 2019: Vegetation response to rising CO2 impacts extreme temperatures. Geophys. Res. Lett., 46, 1383–1392, https://doi.org/10.1029/2018GL080238.
Lemordant, L., P. Gentine, M. Stéfanon, P. Drobinski, and S. Fatichi, 2016: Modification of land‐atmosphere interactions by CO2 effects: Implications for summer dryness and heat wave amplitude. Geophys. Res. Lett., 43, 10 240–10 248, https://doi.org/10.1002/2016GL069896.
Lemordant, L., P. Gentine, A. S. Swann, B. I. Cook, and J. Scheff, 2018: Critical impact of vegetation physiology on the continental hydrologic cycle in response to increasing CO2. Proc. Natl. Acad. Sci. USA, 115, 4093–4098, https://doi.org/10.1073/pnas.1720712115.
Lewis, S. C., A. D. King, S. E. Perkins-Kirkpatrick, and D. M. Mitchell, 2019: Regional hotspots of temperature extremes under 1.5°C and 2°C of global mean warming. Wea. Climate Extreme, 26, 100233, https://doi.org/10.1016/j.wace.2019.100233.
Li, X., Q. You, G. Ren, S. Y. Wang, Y. Q. Zhang, J. L. Yang, and G. F. Zheng, 2018: Concurrent droughts and hot extremes in Northwest China from 1961 to 2017. Int. J. Climatol., 39, 2186–2196, https://doi.org/10.1002/joc.5944.
Liao, W. L., and Coauthors, 2018: Stronger contributions of urbanization to heat wave trends in wet climates. Geophys. Res. Lett., 45, 11 310–11 317, https://doi.org/10.1029/2018GL079679.
Mahmood, R., Sr., and Coauthors, 2010: Impacts of land use/land cover change on climate and future research priorities. Bull. Amer. Meteor. Soc., 91, 37–46, https://doi.org/10.1175/2009BAMS2769.1.
Meehl, G. A., and C. Tebaldi, 2004: More intense, more frequent, and longer lasting heat waves in the 21st century. Science, 305, 994–997, https://doi.org/10.1126/science.1098704.
Miralles, D. G., and Coauthors, 2013: El Niño–La Niña cycle and recent trends in continental evaporation. Nat. Climate Change, 4, 122–126, https://doi.org/10.1038/nclimate2068.
Miralles, D. G., A. J. Teuling, C. C. van Heerwaarden, and J. V.-G. de Arellano, 2014: Mega-heatwave temperatures due to combined soil desiccation and atmospheric heat accumulation. Nat. Geosci., 7, 345–349, https://doi.org/10.1038/ngeo2141.
Miralles, D. G., P. Gentine, S. I. Seneviratne, and A. J. Teuling, 2019: Land-atmospheric feedbacks during droughts and heatwaves: State of the science and current challenges. Ann. N. Y. Acad. Sci., 1436, 19–35, https://doi.org/10.1111/nyas.13912.
Morrison, A., G. Villarini, W. Zhang, and E. Scoccimarro, 2019: Projected changes in extreme precipitation at sub-daily and daily time scales. Global Planet. Change, 182, 103004, https://doi.org/10.1016/j.gloplacha.2019.103004.
Mueller, N. D., E. E. Butler, K. A. McKinnon, A. Rhines, M. Tingley, N. M. Hobrook, and P. Huybers, 2016: Cooling of US Midwest summer temperature extremes from cropland intensification. Nat. Climate Change, 6, 317–322, https://doi.org/10.1038/nclimate2825.
Nicholls, Z. R. J., and Coauthors, 2020: Reduced complexity model intercomparison project phase 1: Introduction and evaluation of global-mean temperature response. Geosci. Model Dev., 13, 5175–5190, https://doi.org/10.5194/gmd-13-5175-2020.
Papalexiou, S. M., and A. Montanari, 2019: Global and regional increase of precipitation extremes under global warming. Water Resour. Res., 55, 4901–4914, https://doi.org/10.1029/2018WR024067.
Portmann, R. W., S. Solomon, and G. C. Hegerl, 2009: Spatial and seasonal patterns in climate change, temperatures, and precipitation across the United States. Proc. Natl. Acad. Sci. USA, 106, 7324–7329, https://doi.org/10.1073/pnas.0808533106.
Salvucci, G. D., and P. Gentine, 2013: Emergent relation between surface vapor conductance and relative humidity profiles yields evaporation rates from weather data. Proc. Natl. Acad. Sci. USA, 110, 6287–6291, https://doi.org/10.1073/pnas.1215844110.
Schwingshackl, C., M. Hirschi, and S. I. Seneviratne, 2017: Quantifying spatiotemporal variations of soil moisture control on surface energy balance and near-surface air temperature. J. Climate, 30, 7105–7124, https://doi.org/10.1175/JCLI-D-16-0727.1.
Seneviratne, S. I., D. Lüthi, M. Litschi, and C. Schär, 2006: Land–atmosphere coupling and climate change in Europe. Nature, 443, 205–209, https://doi.org/10.1038/nature05095.
Seneviratne, S. I., T. Corti, E. L. Davin, M. Hirschi, E. B. Jaeger, I. Lehner, B. Orlowsky, and A. J. Teuling, 2010: Investigating soil moisture–climate interactions in a changing climate: A review. Earth-Sci. Rev., 99, 125–161, https://doi.org/10.1016/j.earscirev.2010.02.004.
Seneviratne, S. I., and Coauthors, 2013: Impact of soil moisture-climate feedbacks on CMIP5 projections: First results from the GLACE-CMIP5 experiment. Geophys. Res. Lett., 40, 5212–5217, https://doi.org/10.1002/grl.50956.
Seneviratne, S. I., M. G. Donat, A. J. Pitman, R. Knutti, and R. L. Wilby, 2016: Allowable CO2 emissions based on regional and impact-related climate targets. Nature, 529, 477–483, https://doi.org/10.1038/nature16542.
Sippel, S., J. Zscheischler, M. D. Mahecha, R. Orth, M. Reichstein, M. Vogel, and S. I. Seneviratne, 2017: Refining multi-model projections of temperature extremes by evaluation against land–atmosphere coupling diagnostics. Earth Syst. Dyn., 8, 387–403, https://doi.org/10.5194/esd-8-387-2017.
Tavakol, A., V. Rahmani, and J. Harrington Jr., 2020: Evaluation of hot temperature extremes and heat waves in the Mississippi River basin. Atmos. Res., 239, 104907, https://doi.org/10.1016/j.atmosres.2020.104907.
Teuling, A. J., 2018: A hot future for European droughts. Nat. Climate Change, 8, 364–365, https://doi.org/10.1038/s41558-018-0154-5.
Thiery, W., E. L. Davin, D. M. Lawrence, A. L. Hirsch, M. Hauser, and S. I. Seneviratne, 2017: Present-day irrigation mitigates heat extremes. J. Geophys. Res. Atmos., 122, 1403–1422, https://doi.org/10.1002/2016JD025740.
Thiery, W., and Coauthors, 2020: Warming of hot extremes alleviated by expanding irrigation. Nat. Commun., 11, 290, https://doi.org/10.1038/s41467-019-14075-4.
Vasseur, D. A., and Coauthors, 2014: Increased temperature variation poses a greater risk to species than climate warming. Proc. Biol. Sci., 281, 20132612, https://doi.org/10.1098/rspb.2013.2612.
Vogel, M. M., R. Orth, F. Cheruy, S. Hagemann, R. Lorenz, B. J. J. M. van den Hurk, and S. I. Seneviratne, 2017: Regional amplification of projected changes in extreme temperatures strongly controlled by soil moisture-temperature feedbacks. Geophys. Res. Lett., 44, 1511–1519, https://doi.org/10.1002/2016GL071235.
Wang, R., J. Chen, X. Chen, and Y. Wang, 2017: Variability of precipitation extremes and dryness/wetness over the southeast coastal region of China, 1960–2014. Int. J. Climatol., 37, 4656–4669, https://doi.org/10.1002/joc.5113.
Wang, R., P. Gentine, J. B. Yin, L. J. Chen, J. Y. Chen, and L. H. Li, 2021a: Long-term relative decline in evapotranspiration with increasing runoff on fractional land surfaces. Hydrol. Earth Syst. Sci., 25, 3805–3818, https://doi.org/10.5194/hess-25-3805-2021.
Wang, R., L. Ü. Guonian, L. Ning, L. Yuan, and L. Li, 2021b: Likelihood of compound dry and hot extremes increased with stronger dependence during warm seasons. Atmos. Res., 260, 105692, https://doi.org/10.1016/j.atmosres.2021.105692.
Wang, X. H., and H. Zhang, 2017: Effects of Australian summer monsoon on sea surface temperature diurnal variation over the Australian north-western shelf. Geophys. Res. Lett., 44, 9856–9864, https://doi.org/10.1002/2017GL075008.
Williams, I. N., and M. S. Torn, 2015: Vegetation controls on surface heat flux partitioning, and land-atmosphere coupling. Geophys. Res. Lett., 42, 9416–9424, https://doi.org/10.1002/2015GL066305.
Yin, J., S. Guo, P. Gentine, S. C. Sullivan, L. Gu, S. He, J. Chen, and P. Liu, 2021: Does the hook structure constrain future flood intensification under anthropogenic climate warming? Water Resour. Res., 57, e2020WR028491, https://doi.org/10.1029/2020WR028491.
Yuan, K., and Coauthors, 2020: Deforestation reshapes land-surface energy-flux partitioning. Environ. Res. Lett., 16, 024014, https://doi.org/10.1088/1748-9326/abd8f9.
Zhou, S., and Coauthors, 2019: Land-atmosphere feedbacks exacerbate concurrent soil drought and atmospheric aridity. Proc. Natl. Acad. Sci. USA, 116, 18 848–18 853, https://doi.org/10.1073/pnas.1904955116.
Zscheischler, J., and S. I. Seneviratne, 2017: Dependence of drivers affects risks associated with compound events. Sci. Adv., 3, e1700263, https://doi.org/10.1126/sciadv.1700263.