1. Introduction
Floods and droughts have major consequences for human and ecosystem health, agriculture, and the economy. The number of billion-dollar droughts has increased substantially in recent years, raising concerns on changing hydrological extremes (https://www.ncdc.noaa.gov/billions/). Globally, floods and droughts affect more than 100 million people each year, with more devastating impacts in developing countries where more people live in rural areas with limited resources and fewer options to cope with extreme heat and food and water shortages (Miyan 2015). Detailed understanding of how the frequency and intensity of hydrological extremes may be responding to climate change across different regions and climate zones would be valuable for adaptation and minimizing economic damage and human hardship.
Previous studies have reported apparent modulation of the water cycle and its extremes by climate change. Blöschl et al. (2019) attributed increases of floods in northwestern Europe to increases in fall–winter precipitation and decreases of floods in eastern and southern Europe to decreases in snowfall and annual precipitation, respectively, during 1960–2010. In Russia, increasing flooding has been attributed to increases in daily precipitation, fast spring snowmelt, and thawing of permafrost (Frolova et al. 2017; Zolotokrylin and Cherenkova 2017). Russia has also seen increases in severe droughts and wildfires in recent years, including in its Arctic regions (Shvidenko and Schepaschenko 2013; Frolova et al. 2017; Ciavarella et al. 2021). In the southwestern United States, recent studies based on tree-ring data and climate reconstruction suggest that both drought occurrence and the declines in streamflow in major river basins were linked to anthropogenic warming (Woodhouse et al. 2016; Lehner et al. 2017; Martin et al. 2020; Milly and Dunne 2020; Williams et al. 2022). In addition, incidences of weather whiplash (a rapid shift from one extreme weather condition to its opposite), as California has recently experienced, have been on the rise and are projected to increase with global warming (Huang and Swain 2022; Zamora-Reyes et al. 2022).
However, systematic changes in hydrological extremes across continental to global scales have been difficult to confirm, as noted in global and U.S. climate reports (Douville et al. 2021; USGCRP 2018). In fact, using streamflow data from more than 1200 gauges in minimally altered catchments in North America and Europe over 60–80-yr periods, Hodgkins et al. (2017) showed that natural climate variability significantly affected the occurrence of floods, while linear trends associated with climate change were insignificant. On the other hand, the number of streamflow gauges is limited, especially in developing countries, and the gauges are often clustered around certain areas, making it difficult to rely on streamflow data alone for examining large-scale changes in hydrological extremes.
Soil moisture simulated by land surface models has been the basis for several studies of droughts and pluvials at continental to global scales (e.g., Andreadis et al. 2005; Sheffield et al. 2009; He et al. 2020; He and Sheffield 2020). A shortcoming of this approach is that output from large-scale models is subject to significant uncertainty due to deficiencies in model inputs and physics (e.g., Kato et al. 2007; Xia et al. 2012). In particular, many such models do not simulate permafrost, surface water, or groundwater, which are undergoing considerable changes (Rodell et al. 2018). Perhaps more relevant to the analysis of extreme events, the atmospheric reanalysis products used to drive these models may underestimate extreme precipitation (Hu and Franzke 2020) and are subject to large uncertainties at high elevations and during snow events (Cui et al. 2017) and therefore are likely to underestimate the occurrence and severity of floods. In general, a major challenge to detecting changes in hydrological extremes is the lack of global-scale spatially and temporally consistent observations.
The Gravity Recovery and Climate Experiment (GRACE) and GRACE Follow-On (GRACE-FO) satellite missions (hereafter GRACE/FO) have been mapping Earth’s gravity field nearly continuously since 2002 (Tapley et al. 2004; Landerer et al. 2020). Temporal variations of gravity can be used to infer terrestrial water storage (TWS) changes, which include soil moisture, groundwater, snow and ice, and surface water. Reflecting water storage change in the full surface–soil–aquifer profile, TWS data are well suited for studying hydrological extremes, especially anomalous conditions that last months to years. The effectiveness of GRACE/FO data for detecting and quantifying drought has been demonstrated in numerous contexts, including the 2003 European heat wave (Andersen et al. 2005), the Millennium Drought in Australia (Leblanc et al. 2009; van Dijk et al. 2013), the multiyear California drought in the 2000s (Weiler 2014; Thomas et al. 2017), and drought events in the Yangtze, Amazon, and other river basins (Chen et al. 2010; Thomas et al. 2014; Zhang et al. 2015; Getirana 2016). Zhao et al. (2017), Li et al. (2019), and Gerdener et al. (2020) developed GRACE/FO-based indicators for detecting hydrological drought globally. Other studies have used positive anomalies of GRACE/FO TWS to detect floods in the Amazon and a small basin in Cambodia (Chen et al. 2010; Tangdamrongsub et al. 2016) and to map flood vulnerability (Reager and Famiglietti 2009; Reager et al. 2014; Singh et al. 2021; Shah and Mishra 2021). As TWS changes represent net fluxes between precipitation, ET and runoff, GRACE/FO TWS data have been used to examine global increases in ET (Pascolini-Campbell et al. 2021) and intensification of the water cycle (Eicker et al. 2016; Chandanpurkar et al. 2021).
In this study, we employed a spatial-temporal clustering algorithm to identify wet/dry extreme events without predefined boundaries. Clustering in an objective way is necessary for examining changes in the characteristics of extreme events such as extent and frequency without bias or preconceived notions. We filled the observation gap (11 months) between the two GRACE/FO missions and other missing months using TWS simulated by a GRACE/FO data assimilating land surface model (Li et al. 2019). To better understand TWS-based extremes, we examined the most intense events in each continent, determined using an intensity metric (Thomas et al. 2014), which is based on spatially and temporally accumulated TWS anomalies. Mann-Kendall trends were calculated to examine changes in the characteristics of extremes including frequency, extent, duration, severity, and intensity. Correlation analysis was further performed to identify major controls of climate variability and climate change on changes in extremes in four major climate classes. In an earlier study, Rodell and Li (2023) employed the same clustering approach and intensity metric to rank the most intense wet and dry events during 2002–21. They determined that the global total intensity of hydroclimatic extreme events was strongly and significantly correlated with global mean temperature. The purpose of this study was to perform a more thorough analysis of how hydrological extremes have been changing in different climate zones and regions around the world, including severity–area–duration analysis, with the goal of providing a more detailed understanding that will be valuable for anticipating future change and mitigating impacts.
In section 2 we describe the data sources, the data processing, the clustering algorithm, and the metrics used to characterize extreme events. In section 3 we present the most intense events in five continents, changes in hydrological extremes, and correlation analysis results. In section 4 we discuss the hydroclimatic and anthropogenic drivers behind the detected changes.
2. Data and methods
a. GRACE/FO data
The GRACE/FO data used in this study are the mascon solutions developed by the Center for Space Research (CSR) at the University of Texas (Save et al. 2016). CSR applies a time-variable regularization matrix, derived from GRACE/FO information only, to constrain the inversion of satellite ranging data to gravity fields. This approach has been shown to reduce GRACE/FO signal damping associated with spherical harmonic solutions and related postprocessing (Landerer and Swenson 2012; Save et al. 2016). The data are available on a 0.25° global grid, which facilitates delineation of study regions and signal contamination from ocean signals. However, the effective resolution of GRACE/FO observations is lower, on the order of 150 000 km2 at midlatitudes (Rowlands et al. 2005; Swenson et al. 2006) due to the limitations of the measurement system.
We excluded regions that have shown persistent, long-term TWS declines not associated with drought, because these would otherwise be identified as extreme events by the clustering algorithm. These include regions of climate change driven ice loss, such as Greenland, glaciers along the Gulf Coast of Alaska, and Patagonia, and agricultural areas where groundwater withdrawals for irrigation, in some cases in conjunction with frequent droughts, have led to groundwater depletion: California’s Central Valley, northern India and the North China Plain. We followed Rodell et al. (2018) in delineating these regions. Our study period is August 2002–December 2021, which excludes problematic GRACE observations in April and May 2002 (June and July 2002 data are absent).
b. Global GRACE/FO data assimilation
There is an 11-month gap (July 2017–May 2018) between the two GRACE/FO missions, in addition to 18 other missing months of data that occur mostly between 2011 and 2017. To create a continuous record, we used simulated TWS from a global GRACE/FO data assimilating instance (Li et al. 2019) of the Catchment land surface model (CLSM; Koster et al. 2000) driven by the Land Information System (Kumar et al. 2006) to fill the data gaps. CLSM-simulated TWS consists of soil moisture, groundwater, snow, and canopy interception. Surface water and permanent ice are not simulated by CLSM. The data assimilation method incorporates surface water storage changes into groundwater (which is tightly coupled with surface water in the real world; Winter et al. 1998), and only 18 of 221 months were filled using the data assimilation output, hence the repercussions for the results are likely to be trivial. Regions with permanent ice cover were excluded from the analysis, as previously stated. Due to CLSM’s inability to simulate groundwater withdrawals, in regions where groundwater pumping intensifies TWS losses during droughts, CLSM simulated TWS may underestimate TWS dynamics. However, the effects on our results are likely small because regions with intense groundwater withdrawals were excluded as stated earlier. Nevertheless, to be cautious, we removed from our analysis any extreme event that began and ended within the 11-month gap period.
c. Data processing and clustering algorithm
We then used a clustering algorithm, ST-DBSCAN (Birant and Kut 2007), to delineate wet and dry events from spatially and temporally continuous groups of flagged wet and dry data points. Note that wet events only comprise wet points and dry events only comprise dry points. The version of ST-DBSCAN used in this study employs a spatial radius (R) and a time interval to define a search domain for a given wet/dry cell. We set the spatial radius R to 250 km, long enough to encompass two adjacent 2° grid cells anywhere on Earth, and we set the time interval to 1 month, so that data points (wet or dry cells) from the 2 months adjacent to the current month would be within the search domain.
If the number of neighboring wet/dry data points for a given wet/dry data point was larger than a minimum threshold (Nmin), this data point was marked as a core point and was given a cluster label. The process was repeated for all the neighboring points of the core point, and the same cluster label was given to any neighboring core points. With the spatial radius, R = 250 km, and the one-month interval, we tested three values for Nmin, 10, 12, and 14, to determine an optimal set of clusters based on the intraclustering distance, the averaged distance of all wet/dry cells to the centroid of a cluster, and intercluster distance, the distance between the centroids of two clusters. Ultimately, we chose Nmin = 12 based on sensitivity analysis (to be discussed in section 3a). As stated earlier, any clusters that fell entirely within the 11-month gap (July 2017–May 2018) between the two satellite missions were excluded from this study.
d. Characterization of extreme events
e. Severity–area–duration analysis
To examine how severity changes with spatial extent and duration, we calculated the severity–area–duration (SAD) curve for all the events. SAD was originally developed for upscaling point-scale precipitation to regional averages but has been adopted for analyzing hydrological extremes in previous studies (Andreadis et al. 2005; Sheffield et al. 2009; He et al. 2020).
We began the process of deriving a SAD curve by calculating the average severity [standardized TWS anomaly; Eq. (1)] of an event for a predefined duration (3, 6, and 12 months in this study). The maximum severity (absolute value for dry events) and its location were then identified. To examine how severity changes with increasing area, the next maximum severity was sought in the surrounding area, equal to 200 000 km2 in this study, from the location of the previous maximum severity. This process was repeated for increasingly larger areas (200 000 km2 increments) until the extent of the event at this predefined duration was reached. Since the maximum severity may occur at any month, the aforementioned process was repeated for all the overlapping time periods (each equal to the predefined duration) within the event and the maximum severity for each area and period was selected to form the SAD curve. Therefore, a SAD curve represents the maximum severities occurring within an event over increasingly larger areas and over the predefined durations. For additional details on SAD, we refer to Andreadis et al. (2005) and Sheffield et al. (2009).
3. Results
a. Sensitivity to clustering parameters
With the spatial radius fixed (R = 250 km in this case), the number of wet and dry clusters (events) typically decreases as the minimum number of neighboring points required to define a core point Nmin (top row in Fig. S1 in the online supplemental material) increases. This is because larger Nmin values cause clusters with fewer data points to be eliminated. The total number of clusters in Eurasia is more sensitive to Nmin than in other continents because it is large, making more wet and dry grid cells available for clustering.
Increases in Nmin lead to smaller intracluster distances, producing denser clusters (bottom row in Fig. S1). Intercluster distances decrease with increasing Nmin in most cases (bottom row in Fig. S1), which is likely due to reduced numbers of clusters including those located near the continental boundaries. In general, the intercluster and intracluster distances are not sensitive to the choice of Nmin given a 250-km spatial radius and the monthly interval. We also tested a larger R value (350 km), which increased intracluster distances considerably (for Nmin = 10, 12 only; Fig. S1).
In setting the parameters we preferred large intercluster and small intracluster distances because they are associated with more distinct clusters. However, while R = 250 km and Nmin = 14 yielded denser clusters than R = 250 km and Nmin = 12, it also resulted in too few wet clusters in Australia (Fig. S1). Along with visual examination of the most intense events (which are discussed in section 3b), we determined that R = 250 km and Nmin = 12 (and a one-month time interval) yielded appropriate numbers of wet and dry clusters for this study.
Using these parameters, the clustering algorithm yielded 505 wet clusters and 551 dry clusters (excluding those that fell entirely in the GRACE/FO gap period; Table 1). About 70% of these events lasted 6 months or shorter and less than 10% of them lasted 12 months or longer. The maximum extents and durations of events in each continent are much larger than the averages for each continent, suggesting that the events are highly variable overall (Table 1).
Summary of wet and dry events in the five continents.
b. Most intense events
The intensity metric [Eq. (3)] integrates three aspects of an event, TWS gains or losses, extent, and duration. Thus, it offers an alternative to assessing events based on spatial extent or duration, which are more common approaches (e.g., Andreadis et al. 2005; Sheffield et al. 2009; He et al. 2020). Further, hydrological extremes identified using TWS differ from those identified using soil moisture or streamflow data, in that TWS has a wider range of variability than soil moisture and its anomalies can be more persistent than those of either soil moisture or streamflow. Therefore, examining the most intense TWS events can provide new insight into the severity and how they relate to other hydroclimatic cycles and trends. In this section, we discuss the most intense wet and dry events (six each for Eurasia and three each for the other continents) ranked over each continent based on their intensity and link them with reported extremes when available.
1) Eurasia
(i) Top ranked wet events
The three most intense wet events in Eurasia during 2002–21 were located in high-latitude regions (Fig. 1a). The most intense wet event (5810 km3 month) was associated with floods that occurred in Far East Russia, southern Siberia, and southern European Russia caused by fast spring snowmelt, ice jams, and heavy snowfalls during 2017/18 (Fig. 1e and Fig. S2; Davies 2017; Relief Web 2017). There was also a storm surge in coastal Arctic Russia caused by delayed sea ice formation in 2018 due to warm temperature (NOAA 2018). Although the event spread to a large area, wet conditions occurred mainly in Far East Russia (Fig. S2).
Spatial extents of (a),(c) top six most intense wet and (b),(d) top six most intense dry events in Eurasia and (e),(f) the associated TWS anomaly (km3) time series [the inner sum of Eq. (3)]. The intensities of these events, in descending order, are given in the inset tables in (e) and (f) which also contain the color key for each event.
Citation: Journal of Climate 36, 24; 10.1175/JCLI-D-23-0199.1
The second most intense wet event, which overlapped in time with the most intense wet event, combined a record-breaking spring snowstorm and freezing rain in western Russia (Blašković 2018) and a record-breaking fall precipitation event in Norway and the Nordic countries (Figs. 1a,e; Views and News from Norway 2018). The third most intense wet event, which lasted more than 2 years (2007–09), included severe flooding caused by substantial snowfall, fast spring snowmelt, and ice jams in major rivers of central and eastern Russia including the Lena, Yenisei, and Ob (Figs. 1a,e; Landerer et al. 2010; Velicogna et al. 2012).
The fourth wet event was associated with severe floods in large river basins of northeastern China and the adjacent Russian and Mongolian regions during 2020/21 (Figs. 1c,e; FloodList 2021). The fifth event, occurring during 2019–21, spans the Kunlun and Qilian Mountains in western China where annual snowfall has been increasing in recent years (Figs. 1c,e; de Kok et al. 2019). The sixth event was associated with 2015/16 floods caused by spring snowmelt and ice jams in central and northwestern Russia, which caused significant property damages (Figs. 1c,e; Davies 2016b).
(ii) Top ranked dry events
As with the most intense wet events, the top three most intense dry events were located in the more northern latitudes (Fig. 1b). The first and fifth most intense dry events were collocated with known extensive droughts in central Russia during 2012–14 and 2010/11 that significantly impacted regional wheat production and global wheat prices (Figs. 1b,d,f; Sternberg 2011; Cherenkova et al. 2013; USDA 2012). During these periods, heat waves and wildfires were also observed in Siberia, where the top event stayed the longest (Fig. S3; NASA Earth Observatory/Event 77712 2013). The heat wave and the associated spike of global food prices in 2012 has been linked to amplified Rossby waves, which can lead to simultaneous heat waves in midlatitude regions including eastern Europe (Kornhuber et al. 2020).
The second most intense dry event includes the recent European drought that ended in early 2021 with an intensity of −5689 km3 month (Figs. 1b,f). In addition to abnormally low precipitation, the drought was likely exacerbated by the warmest winter temperature recorded in Europe (NASA Earth Observatory/Event 146888 2020).
The third most intense dry event was associated with hot and dry conditions in Far East Russia and the Sakhalin Island, where numerous forest fires occurred during the period of the event, 2002–04 (Figs. 1b,f; NASA Earth Observatory/Event 11787 2003). The fourth most intense dry event encompassed severe droughts in Afghanistan and Pakistan in 2021, which triggered a humanitarian crisis in the area (Figs. 1d,f; FAO 2019; New Humanitarian 2019). The sixth intense dry event is associated with droughts in Norway, Sweden, and the Baltic Republics in 2003 due to long-term (12 month) precipitation deficits (Figs. 1d,f; Spinoni et al. 2015). Increases in occurrences of severe droughts and heat waves in northwestern Europe may not be a coincidence: recent studies have linked them to enhanced Northern Hemisphere polar jet stream anomalies since the 1960s (e.g., Trouet et al. 2018).
2) North America
(i) Top ranked wet events
The most intense wet event in North America was evidenced by flooding in the major river basins of the United States east of the Rockies (Fig. 2a and Fig. S2). The flooding started in 2019 in the Missouri and Upper Mississippi basins following record-breaking January–March precipitation and fast snowmelt (Fig. 2c; NOAA 2019a). It then spread to the lower Mississippi, where the water levels were already high due to antecedent wet conditions (Gasparini and Yuill 2020). The event continued in the spring of 2020 with more floods in the Midwest and the southern states (NOAA 2020) and eventually merged with wet conditions in the mid-Atlantic states (NOAA 2019b, 2021). The magnitude of this event (11 896 km3 month) was the second largest of all events, globally.
As in Fig. 1, but for the top three most intense wet and dry events in North America and South America with maximum extents in (a) and (b) and associated TWS anomaly (km3) time series in (c)–(f).
Citation: Journal of Climate 36, 24; 10.1175/JCLI-D-23-0199.1
The second intense wet event in North America was associated with several winter storms that hit the mountains of the western United States in 2010/11 during a strong La Niña episode (Figs. 2a,c; Wikipedia 2023a). The third most intense wet event was notable for the heavy precipitation that central Canada received in 2005 (Figs. 2a,c; Wikipedia 2023b).
(ii) Top ranked dry events
The most intense dry event extended from Alaska to western Canada, but mainly affected northern Alberta, British Columbia, and the western Northwest Territories of Canada (Fig. 2b and Fig. S3). This event began in the northwest with abnormally dry conditions and worsened in early 2019 due to high temperatures and below normal precipitation (Fig. 2e; Canadian Drought Monitor 2019; AGCanada 2019). The interior of western Canada including the Arctic region has become more conducive to more intense and frequent extreme events including droughts due to shifts of atmospheric circulation patterns (Stewart et al. 2019).
The second intense dry event in North America was associated with recent droughts in southwestern North America, which were the worst in the last 22 years and partly influenced by anthropogenic climate change (Figs. 2b,e; Williams et al. 2022). The drought condition improved in late 2021 due to atmospheric river events (WMO 2022) but continued into 2022 until later that year when record precipitation brought widespread floods to California and surrounding areas, abruptly ending the drought (CalMatters 2023).
The third ranked dry event encompassed a Canadian Prairie drought in the early 2000s (Figs. 2b,e; Hanesiak et al. 2011) and dry conditions in southern Ontario and Quebec during 2002/03 and in the Great Lakes region of the United States during August–October 2003 (see drought maps in the Canadian Drought Monitor and the U.S. Drought Monitor).
3) South America
(i) Top ranked wet events
The top three most intense wet events in South America were associated with damaging floods caused by extreme precipitation in Brazil (Wikipedia 2023c), the Amazon (Espinoza et al. 2022) and the area encompassing Paraguay, Argentina, Uruguay, and southern Brazil (Figs. 2a,d; Davies 2016a; Wikipedia 2023d). The top two most intense wet events were associated with the 2009 and 2021 La Niña, and La Niñas are associated with wet December–February weather in the eastern Amazon basin (Mason and Goddard 2001). The third most intense wet event occurred during the 2016 El Niño, and El Niños are associated with wet summer and winter conditions in Uruguay and surrounding regions (Mason and Goddard 2001). It is noteworthy that a recent tree-ring study posited that hydrological extremes have increased since the mid-twentieth century in these areas, in part due to increases in greenhouse gas emissions (Morales et al. 2020).
(ii) Top ranked dry events
The top three most intense dry events all occurred in recent years (Figs. 2b,f). They affected mainly Brazil, where below normal annual precipitation had been observed in many areas since 2012 (Cunha et al. 2019). The top ranked event—also the top ranked dry event globally (Rodell and Li 2023)—was a severe Brazilian drought associated with the 2015/16 El Niño, which was exacerbated by the highest annual temperature recorded in Brazil in 2015 and the record-setting global temperature in 2016 (Berenguer et al. 2021; Jiménez-Muñoz et al. 2016; World Bank 2021).
The second most intense event represents the recent record-setting drought in central and southern Brazil where water levels in some reservoirs decreased by more than 8 m (NASA Earth Observatory/Event 148468 2021). The third most intense dry event occurred in eastern Brazil where precipitation deficits were observed throughout 2011–19 (Cunha et al. 2019). With a semiarid climate, TWS changes in this region may lag precipitation changes and thus, TWS did not reach the lowest value until 2020. While droughts are frequent in Brazil due to strong coupling with El Niño–Southern Oscillation (ENSO) and sea surface temperatures in the tropical Atlantic, land degradation and consequent desertification may have contributed to the intensity of these events (Marengo et al. 2017; Cunha et al. 2015).
4) Africa
(i) Top ranked wet events
The most intense wet event in Africa, which was massive with an intensity of 31 354 km3 month, was also the most intense wet event globally (Figs. 3a,e; Rodell and Li 2023). It spread across the Sahel and tropical Africa and was punctuated by severe floods during 2019/20 (FloodList 2019; Phys.org 2019). The flood in eastern Africa in late 2019 began with record-breaking precipitation in association with a positive phase in the Indian Ocean dipole (Wainwright et al. 2020), which is a major control on rainfall in the region (Behera et al. 2005). In the western Sahel, wet conditions were likely caused by a strong tropical easterly jet (Akinsanola and Zhou 2020). Increases in extreme precipitation in the area in the last decade have been reported in association with changes in atmospheric conditions (Vischel et al. 2019).
As in Fig. 1, but for the top three most intense wet and dry events in Africa and Australia with maximum extents in (a)–(d) and associated TWS anomaly (km3) time series in (e)–(h).
Citation: Journal of Climate 36, 24; 10.1175/JCLI-D-23-0199.1
The second and third most intense wet events occurred in southern Africa, where weather is strongly influenced by ENSO (Figs. 3a,e; Nicholson and Selato 2000). Both started during the strong La Niña of 2010/11, which brought heavy precipitation and flooding that caused significant loss of life and property damages (Wikipedia 2023e).
(ii) Top ranked dry events
The most intense dry event encompassed the 2005 severe eastern equatorial African drought (Figs. 3c,g; Geotimes 2005; Hastenrath et al. 2007), which killed 40% of Kenya’s cattle (Wakabi 2006), and droughts in the Congo basin and southeastern Africa (Zhou et al. 2014; NASA Earth Observatory/Event 15849 2005). The second ranked dry event, beginning around the end of 2018, featured severe droughts in southern Africa, with significant impacts on food supply, agriculture, and livestock production (Figs. 3c,g; NASA Earth Observatory/Event 146015 2019). The third ranked dry event was located in the western Sahelian region (Figs. 3c,g). Precipitation gauge data from Côte d’Ivoire indicate that 2002/03 was one of the driest periods since 1970 (Santé et al. 2019).
5) Australia
(i) Top ranked wet events
The most intense wet event in Australia during 2002–21 spread across the entire continent but mainly affected eastern Australia (Figs. 3b,f and Fig. S2). Heavy rain, including that from two cyclones, caused a series of floods between November 2010 and March 2011 and in January 2013 (Trewin 2011; Lorrey et al. 2012; Australian Institute for Disaster Resilience 2012; NASA Earth Observatory/Event 49779 2011; Wikipedia 2023f). Southern Australia also had the wettest December on record in 2011, which helped raise water levels in lakes and reservoirs back to normal after years of decline (NASA Earth Observatory/Event 76680 2011; Jensen 2011). Parts of Western Australia were also flooded in early 2011 (NASA Earth Observatory/Event 49779 2011). A major driver of the heavy rains was an unusually strong La Niña in 2010/11 and a strong negative phase of the Indian Ocean dipole (Ng et al. 2018). Synchronization of different climate modes is known to cause extreme precipitation and droughts in Australia (Cleverly et al. 2016). These widespread floods ended the Millennium Drought that had affected the continent since the early 2000s (Heberger 2012; van Dijk et al. 2013).
The second and third most intense wet events were caused in part by heavy rainfall from two tropical cyclones, Blanche in March 2017 and Monty in March 2004 (Figs. 3b,f; Levinson and Salinger 2005; Aljazeera 2017; ABM 2004). Short-term but wet tropical storm events can cause significant TWS increases in areas that are otherwise dry in winter, such as northwestern Australia (ABM 2015).
(ii) Top ranked dry events
The top three most intense dry events occurred within the Millennium Drought period, 2001–09 (Figs. 3d,h). The top dry event encompassed central and southeastern Australia, where the 2008 precipitation was the lowest in nearly a century. Southeastern Australia was hit the hardest by the Millennium Drought based on satellite-derived vegetation water content (van Dijk et al. 2013). Considering that croplands are extensive in the region, withdrawals for irrigation likely exacerbated the TWS depletion (van Dijk et al. 2013).
The second most intense top dry event was most prominent in northern Australia (Fig. 3d and Fig. S3). The minimum TWS anomaly (−840 km3) was reached in 2005, when northern Australia received low to near-record low precipitation (Fig. 3h; ABM 2006).
The third most intense dry event was centered in south-central Australia (Fig. 3d and Fig. S3). The year of its minimum TWS (2009) coincided with large precipitation deficits in central Australia according to Australian Bureau of Meteorology (Fig. 3h).
The intensity of these dry events was lower than that of the top wet event, owing in part to the strength of the 2010/11 La Niña and associated heavy precipitation. Further, the magnitudes of the TWS anomalies during the Millennium Drought may have been underestimated precisely because that drought persisted for nearly half of the GRACE/FO data record, thus causing the baseline mean TWS, from which anomalies were calculated, to be on the low side.
c. Severity–area–duration analysis
The intensity of an event, as defined here with units of km3 month [see Eq. (3)], differs from severity [unitless in this case; see Eq. (2)], which measures the degree of deviation from the mean instantaneously or over a set period. To examine severity and its relationship with intensity, we calculated SAD curves for all wet and dry events over 3-, 6- and 12-month durations (Figs. S4 and S5).
In each continent, maximum wet and dry severities occurred over the smallest areas and attained roughly the same magnitudes, between 4 and 6, with slightly higher values in Eurasia. For a given area, the most intense wet events tended to be more severe than the most intense dry events in Eurasia, Africa, and Australia, while they had similar severity in the Americas (Fig. S5). Severity generally decreases with increasing area, especially for small extent events at the 3-month duration. At longer durations (6 and 12 months), severity decreases more slowly with increasing area, indicating that the location of maximum severity can migrate as an event evolves. Note that there are fewer SAD curves for the 12-month duration subplots because most events last less than 12 months (Table 1).
In Africa and Australia, dry events were generally more extensive than wet events and their severity decreased more slowly with increasing area than that of wet events. This suggests that the atmospheric regimes that cause droughts are more expansive and uniform in Africa and Australia than in other continents. However, when they occur on these two continents, large-scale wet events can spread to large areas with high intensity as shown in the most intense wet (highlighted in yellow lines) events (Fig. S4).
These SAD curves are less smooth (i.e., a consistent and gentle decrease in severity from small to large scales) than those based on simulated soil moisture (Andreadis et al. 2005; Sheffield et al. 2009; He et al. 2020). Possible explanations for this include the coarse spatial and temporal scales of GRACE/FO TWS, the fact that TWS integrates multiple processes (those involved in the evolution of surface waters, snow, soil moisture, and groundwater), each of which may exhibit different temporal variability, and possibly various modeling deficiencies that could cause soil moisture variability across scales to be underrepresented.
The three most intense events in each continent (highlighted using the same color scheme as Figs. 1–3) are generally more severe than others regardless of duration and area (Figs. S4 and S5). The exception is Australia, where the top three dry events were middling in terms of severity for areas less than 106 km2. This indicates that the high intensity of these three events was driven more by area and duration than by severity. Severity of top events in each continent tended to decrease more gradually with increasing area compared to other events. This is a consequence of the most intense events generally covering the largest spatial scales, and hence more potential subareas with great severity.
d. Temporal variability of extreme events
1) Annual tendencies
To examine changes in extreme events, we calculated the Mann-Kendall trend in annual characteristics (number of events, intensity, extent, duration, and severity, excluding the partial year 2002) of extreme events in four Köppen–Geiger climate classes (Fig. 4). Annual quantities were calculated by averaging (summing for the number of events) over all events in a given year, determined based on the time of the maximum/minimum TWS anomaly (km3) of a wet/dry event. The location where an event stayed the longest (see Figs. S2 and S3) determined the climate class to which it belonged.
(a2) Spatial distribution of the four major Köppen–Geiger climate classes, (b2)–(g2) annual time series averaged over all climates, and (a1)–(f1),(a3)–(f3) annual tendencies of extreme event characteristics for the four and all climate classes. Polar climate regions and those with significant TWS declines due to human-induced groundwater depletion and losses of permanent ice are masked out in the climate class map and excluded from the analysis. Shaded bars represent statistically significant trends with p = 0.1.
Citation: Journal of Climate 36, 24; 10.1175/JCLI-D-23-0199.1
Significant changes in frequency of occurrence occurred in the Dry climate (Figs. 4a1,a3), where the number of wet events decreased while the number of dry events increased and in the Continental climate where the number of dry events increased. The intensity of wet events increased significantly when averaged over all climates, but changes in individual climates were not significant (Fig. 4b1). This seeming discrepancy is due to the most intense events skewing the global averages. Annual intensity averaged over all climates for both wet and dry events exhibited large interannual variability during and after the 2010/11 La Niña episode, which was associated with several of the most intense events as discussed earlier (Fig. 4c2; note that positive SOI is associated with La Niña and negative SOI is associated with El Niño).
Similarly, the extent and duration for all climates showed strong interannual variability in 2010/11 (Figs. 4d2,e2) and therefore, because that is the middle of the time series, in most cases they exhibited no significant trends (Figs. 4c1,d1,c3, and d3). Longer records will be needed to identify climate change–related trends in these quantities, particularly in regions strongly influenced by ENSO.
In contrast, average annual wet and dry event severity increased significantly in all cases except for wet events in the Dry climate (Figs. 4e1,e3). Standardization used in calculating severity makes it comparable across different sizes and durations of events. Therefore, average severity was not dominated by large events as with the other metrics. As shown in Fig. 4f2, average annual severity was not affected by the 2010/11 La Niña event.
TWS anomalies generally did not show consistent significant changes except in the Continental climate where wet TWS anomalies increased significantly while dry TWS anomalies decreased significantly, suggesting increases in the amplitude of TWS changes (Figs. 4f1,f3).
2) Correlation analysis
To examine the impacts of climate and climate change on TWS variability, we calculated lagged correlations (up to 12 months) between five climate indicators/signals (listed in the caption of Fig. 5) and total monthly TWS anomalies in each climate class. For each climate class and event type (wet or dry), the strongest correlated climate indicator is plotted in Fig. 5 along with the time series of TWS anomalies summed over all associated extreme events.
Monthly time series of TWS anomalies summed over (left) wet and (right) dry extreme events in a given climate class and the climate signal that shows the largest magnitude of correlation with those anomalies (r is the maximum correlation). The number in parentheses indicates the time lag (in months) of the maximum correlation. A 6-month running average was applied to smooth all plotted time series. The climate indices used for correlation analysis are the Goddard Institute for Space Studies (GISS) global surface temperature (GISTemp), the Southern Oscillation index (SOI), the TransNiño Index (TNI), the North Atlantic Oscillation (NAO), and the Dipole Mode Index (DMI).
Citation: Journal of Climate 36, 24; 10.1175/JCLI-D-23-0199.1
TWS anomalies of wet events in the Tropical climate class were most strongly correlated with a global average temperature index (GISTemp; r = 0.67; Fig. 5a). This result may be due to the rapid rise of TWS after 2019 associated with the worldwide top wet event that is centered in tropical Africa (Figs. 3a,e). In the Dry climate, TWS anomalies of wet events were most strongly correlated with ENSO (represented by SOI; r = 0.62; Fig. 5c), which is a known influence on precipitation variability in this climate (Dai and Wigley 2000; Ng et al. 2018). The TWS of wet events in the Temperate climate were also most strongly correlated with ENSO, but the correlation is much lower than that in the Dry climate (r = 0.4; Fig. 5e).
In the Continental climate, TWS of wet events were strongest correlated with GISTemp (r = 0.67; Fig. 5g). This result is consistent with observed changes in this climate such as faster than global average increases in temperature (Trenberth et al. 2007). In addition, discharge to the Arctic Ocean from major rivers of Eurasia and North America has been increasing in the past several decades due to thawing of permafrost which releases large quantities of soil water into surface runoff (Peterson et al. 2002; McClelland et al. 2006; Ahmed et al. 2020), with implications for other processes in the region such as ocean salinity and temperature (Timmermans and Marshall 2020). Increases in precipitation due to global warming have also been reported for northern Eurasia and may have contributed to this strong correlation (Ye 2001; Zolotokrylin and Cherenkova 2017).
The strong correlation (r = −0.64) between dry event TWS anomalies and GISTemp in the Tropical climate class is partly explained by the worldwide most intense dry event (in South America), occurred in 2016, when global average temperature reached its record high and, in another part, by the apparent trends in both time series (Fig. 5b). Similarly, the strong correlation (r = −0.61) between dry event TWS anomalies and GISTemp in the Dry climate may be due to the apparent trends in both time series (Fig. 5d). On the other hand, the apparent trends in TWS anomalies in the above cases are caused by drastic decreases in TWS in recent years when global temperature increased substantially.
The strongest correlations for dry events in the other two climate classes were lower than 0.4, so the influence of the plotted climate indicators appears to be weak. Nevertheless, it is worth noting that TWS anomalies of dry events in the Temperate climate exhibited the strongest correlation with the North Atlantic Oscillation (NAO), which has been linked to drought occurrence in this climate zone (McCabe et al. 2004; López-Moreno and Vicente-Serrano 2008).
Wet event characteristics in Tropical and Continental exhibit the strongest correlation with GISTemp in 9 out of 10 cases (Figs. 6a,d). In the Tropical climate, wet event characteristics also show strong correlations (r > 0.5) with DMI in 4 of 5 cases (Fig. 6a). In contrast, wet event characteristics in the Dry and Temperate show weak correlations with the five indicators except a few cases such as frequency of occurrence in the Dry climate and duration in the Temperate climate, which correlated strongly with DMI (r > 0.5), and intensity and extent in the Dry climate, which correlated strongly with NAO (r > 0.55; Figs. 6b,c).
Heat maps of correlations between monthly characteristics of (left) wet and (right) dry events and the five climate indicators in the four major classes and in all climate classes. Monthly intensity, frequency (number of events), and extent were calculated as the sum of these characteristics over all events in a climate while monthly severity and duration were averaged over all events in a climate class.
Citation: Journal of Climate 36, 24; 10.1175/JCLI-D-23-0199.1
The severity of dry events shows the strongest correlation (r > 0.58) with GISTemp across all four climate classes (Figs. 6f–i). Note that the negative correlations are due to severity being negative for dry events. Strong correlations are also observed between GISTemp and other characteristics such as frequency of occurrence of dry events in the Dry and Continental climates and intensity of dry events in Tropical and Dry climates (Figs. 6g,i). Consistent with some of these strong correlations, wildfires and severe droughts have been increasingly observed in high-latitude regions of Russia and Canada (Kelly et al. 2013; Sherstyukov and Sherstyukov 2014; Whitman et al. 2019).
Among all characteristics, severity is the only metric that is consistently most strongly correlated with GISTemp, regardless of climate and the type of extremes (wet or dry), except for wet events in Tropical climate (Fig. 6a). As explained earlier, severity, represented by standardized TWS anomalies, can better capture changes across all events while other characteristics, when averaged, may be dominated by few large events that may be associated with climate variability and thus exhibit large interannual variability.
Averaged over four climates, characteristics of both wet and dry extremes show the strongest and much higher correlation with global temperature than with other climate indicators, except for frequency of wet events, which is more strongly correlated with ENSO (represented by TNI) (Figs. 6e,j).
4. Summary and discussion
We identified 505 wet and 551 dry hydrological extremes globally using the GRACE/FO TWS data for 2002–21 and a temporal–spatial clustering algorithm. The most intense events in each continent, determined based on temporally and spatially accumulated TWS anomalies, reflected hydroclimatic variability associated with large-scale climate phenomena. These include the most intense dry event in South America, which is also the most intense one globally, in Brazil in association with the 2015/16 El Niño and the most intense wet event in Australia due to an unusually strong La Niña in 2010/11 and a strong negative phase of the Indian Ocean Dipole in 2011. The occurrence of most intense events also showed direct impacts of global warming and climate change. These include the top three most intense wet events in Eurasia which were associated with increases in fall–winter precipitation, fast spring snowmelt and, in some cases, thawing of permafrost. Severity–area–duration (SAD) analysis showed that the most intense events are generally more severe (represented by standardized TWS anomalies) than others and remained so throughout their duration. Across continents, these most intense extremes often caused substantial damage to properties and agricultural production as well as loss of human life.
Trend analysis showed that both wet and dry extreme events became more severe over the course of the study period (2002–21) across all four major climate classes. Increases in severity imply that the amplitudes of TWS changes, on average, have increased, based on the definition of severity of an event [Eq. (2)]. Increases in TWS amplitudes suggest that maximum TWS has been increasing while minimum TWS has been decreasing. These changes are likely due to the intensification of wet and dry weather regimes, with warmer air supporting higher evaporation rates during droughts and increased moisture transport during pluvials, such that wet (dry) spells become wetter (drier), hence expanding the dynamic range of TWS (Douville et al. 2021; Benestad et al. 2022; Rodell and Li 2023). Anthropogenic effects such as groundwater withdrawals and changes within the terrestrial water cycle such as degradation of permafrost, which releases subsurface water into streams, may further amplify the dynamic range of regional TWS changes. Increasing severity of both wet and dry events across all climates implies that the severity of weather whiplash has also increased. Identification of drought–pluvial seesaw events and their tendencies has been challenging because previous studies relied on soil moisture (He and Sheffield 2020). Due to its small dynamic range, soil moisture cannot represent the full depth and extent of severe droughts and floods as well as GRACE/FO TWS data do. Consistent with these trends, severity of both wet and dry events showed the strongest correlation with global mean temperature in most cases.
Robust concurrent changes in the frequency of occurrence were only observed in the Dry climate where the number of wet (dry) events decreased (increased) significantly, suggesting a drying trend in that climate. This apparent trend may have been influenced by the internal variability of climate. For instance, positive SOI (corresponding to La Niña events) as shown in the later study period, is known to cause dry conditions in southwestern United States, which is located in the Dry climate zone (Mason and Goddard 2001). As discussed earlier, global warming and associated hydroclimatic changes and anthropogenic effects may have also contributed to this drying trend.
Correlation analysis suggests strong impacts of global warming on TWS changes and all characteristics of wet extreme events in the Continental climate. These results are consistent with observed changes in northern Eurasia such as permafrost degradation, earlier and fast spring snowmelt, and increases in fall–winter precipitation (Frolova et al. 2017; Zolotokrylin and Cherenkova 2017). Seasonal precipitation is projected to increase in this climate, which may further increase wet TWS anomalies and severity of wet extremes (Konapala et al. 2020). A corresponding effect was seen in the warmer climates (Tropical and Dry), where TWS anomalies and characteristics of dry extremes were often most strongly correlated with global average temperature. Increases in water demands and ET in a warmer world may be a factor [WWAP (United Nations World Water Assessment Programme)/UN-Water 2018]. In the case of TWS anomalies of dry events in the Tropical climate, the strong correlation is influenced by the fact that the most intense dry event globally occurred in the warmest year on record (2016) and was located in the Tropical climate (Brazil). Averaged over four climate classes, which smooths out the climate variability of individual climates, the characteristics of hydrological extremes show the strongest correlation with global temperature in 9 out of 10 metrics. As discussed earlier, although these strong correlations may be affected by the apparent trends in the time series of global temperature and TWS anomalies, this does not diminish the impact of global warming on changes in TWS and hydrological extremes as trends in TWS were caused by abrupt increases/decreases in TWS in the last few years in responses to accelerated global warming.
A limitation of this study is the relatively short record of GRACE/FO data, which makes it difficult to put extreme events in the proper context, especially in regions with large interannual TWS variability. The short record also makes it difficult to separate the impacts of global warming from those of long-term climate variability. Another source of uncertainty was the use of simulated TWS output to fill gaps in the GRACE/FO data record, considering that both the model and atmospheric forcing data are imperfect. To minimize the impact of these uncertainties, we excluded events that fell entirely within the 11-month gap between the two missions. Of course, this may have affected our results. We also excluded regions with significant TWS declines associated with groundwater withdrawals. That was necessary in order to focus on nonanthropogenic extremes, but it necessitated nonglobal coverage, as did exclusion of ice-covered regions.
Nevertheless, this study clearly demonstrates the value of GRACE/FO data for identifying and quantifying hydrological extremes in a manner that is not possible using any other observation type. In particular, GRACE/FO data provide an integrated measure of water gains and losses at all depths on and below the surface and thus can reflect full impacts of global warming and climate change on hydrological extremes. As the global water cycle continues to evolve in response to climate change, extending the satellite-based TWS record will be essential to tracking variations in TWS changes and extremes, with implications for water security and natural disasters.
Acknowledgments.
This study was supported by GRACE-FO Science Team.
Data availability statement.
CSR GRACE/GRACE-FO data can be downloaded at http://www2.csr.utexas.edu/grace/RL06_mascons.html and simulated TWS from the GRACE/FO data assimilation into CLSM can be downloaded at https://disc.gsfc.nasa.gov/datasets/GLDAS_CLSM025_DA1_D_2.2/summary?keywords=GLDAS_CLSM025_DA1_D_2.2. Oscillation signals can be downloaded at https://psl.noaa.gov/data/climateindices/list/, https://psl.noaa.gov/gcos_wgsp/Timeseries/DMI/, and https://data.giss.nasa.gov/gistemp/. The python code for the ST-DBSCAN clustering algorithm was obtained from https://github.com/gitAtila/ST-DBSCAN.
REFERENCES
ABM, 2004: Severe tropical cyclone Monty. Australian Bureau of Meteorology, accessed 23 August 2023, http://www.bom.gov.au/cyclone/history/monty.shtml.
ABM, 2006: Australia’s hottest year on record. Australian Bureau of Meteorology accessed 18 August 2023, http://www.bom.gov.au/announcements/media_releases/climate/change/20060104.shtml.
ABM, 2015: Recent rainfall, drought and southern Australia’s long-term rainfall decline. Australian Bureau of Meteorology, accessed 18 August 2023, http://www.bom.gov.au/climate/updates/articles/a010-southern-rainfall-decline.shtml.
AGCanada, 2019: Prairie dryness concerns to persist. Accessed 17 August 2023, https://www.agcanada.com/daily/prairie-dryness-concerns-to-persist.
Ahmed, R., T. Prowse, Y. Dibike, B. Bonsal, and H. O’Neil, 2020: Recent trends in freshwater influx to the Arctic Ocean from four major Arctic-draining rivers. Water, 12, 1189, https://doi.org/10.3390/w12041189.
Akinsanola, A. A., and W. Zhou, 2020: Understanding the variability of West African summer monsoon rainfall: Contrasting tropospheric features and monsoon index. Atmosphere, 11, 309, https://doi.org/10.3390/atmos11030309.
Aljazeera, 2017: Tropical Cyclone Blanche hits northern Australia. Accessed 18 August 2023, https://www.aljazeera.com/news/2017/3/6/tropical-cyclone-blanche-hits-northern-australia.
Andersen, O. B., S. I. Seneviratne, J. Hinderer, and P. Viterbo, 2005: GRACE-derived terrestrial water storage depletion associated with the 2003 European heat wave. Geophys. Res. Lett., 32, L18405, https://doi.org/10.1029/2005GL023574.
Andreadis, K. M., E. A. Clark, A. W. Wood, A. F. Hamlet, and D. P. Lettenmaier, 2005: Twentieth-century drought in the conterminous United States. J. Hydrometeor., 6, 985–1001, https://doi.org/10.1175/JHM450.1.
Australian Institute for Disaster Resilience, 2012: Queensland and Brisbane 2010/11 floods. Accessed 17 August 2023, https://knowledge.aidr.org.au/resources/flood-queensland-201011/.
Behera, S. K., J.-J. Luo, S. Masson, P. Delecluse, S. Gualdi, A. Navarra, and T. Yamagata, 2005: Paramount impact of the Indian Ocean dipole on the East African short rains: A CGCM study. J. Climate, 18, 4514–4530, https://doi.org/10.1175/JCLI3541.1.
Benestad, R. E., and Coauthors, 2022: Global hydro-climatological indicators and changes in the global hydrological cycle and rainfall patterns. PLOS Climate, 1, e0000029, https://doi.org/10.1371/journal.pclm.0000029.
Berenguer, E., and Coauthors, 2021: Tracking the impacts of El Niño drought and fire in human-modified Amazonian forests. Proc. Natl. Acad. Sci. USA, 118, e2019377118, https://doi.org/10.1073/pnas.2019377118.
Birant, D., and A. Kut, 2007: ST-DBSCAN: An algorithm for clustering spatial-temporal data. Data Knowl. Eng., 60, 208–221, https://doi.org/10.1016/j.datak.2006.01.013.
Blašković, T., 2018: Snowfall of the century: Record-breaking snow and freezing rain wreak havoc across Moscow. The Watchers, accessed 17 November 2023, https://watchers.news/2018/02/04/snowfall-of-the-century-record-breaking-snow-freezing-rain-wreak-havoc-across-moscow/.
Blöschl, G., and Coauthors, 2019: Changing climate both increases and decreases European river floods. Nature, 573, 108–111, https://doi.org/10.1038/s41586-019-1495-6.
CalMatters, 2023: Is California’s drought over? Here’s what you need to know about rain, snow, reservoirs and drought. CalMatters, accessed 17 November 2023, https://calmatters.org/environment/2023/01/california-drought-snow-rain/.
Canadian Drought Monitor, 2019: Canadian Drought Monitor: Conditions as of March 31, 2019. Accessed 17 August 2023, https://svma.sk.ca/wp-content/uploads/2019/12/SVMA_AGRICULTURE-CANADA_Canadian-Drought-Monitor_End-of-March-2019.pdf.
Chandanpurkar, H. A., J. T. Reager, J. S. Famiglietti, R. S. Nerem, D. P. Chambers, M.-H. Lo, B. D. Hamlington, and T. H. Syed, 2021: The seasonality of global land and ocean mass and the changing water cycle. Geophys. Res. Lett., 48, e2020GL091248, https://doi.org/10.1029/2020GL091248.
Chen, J. L., C. R. Wilson, and B. D. Tapley, 2010: The 2009 exceptional Amazon flood and interannual terrestrial water storage change observed by GRACE. Water Resour. Res., 46, W12526, https://doi.org/10.1029/2010WR009383.
Cherenkova, E. A., N. K. Kononova1, and N. R. Muratova, 2013: Summer drought 2010 in the European Russia. Geogr. Environ. Sustainability, 6, 55–66, https://doi.org/10.24057/2071-9388-2013-6-1-81-92.
Ciavarella, A., and Coauthors, 2021: Prolonged Siberian heat of 2020 almost impossible without human influence. Climatic Change, 166, 9, https://doi.org/10.1007/s10584-021-03052-w.
Cleverly, J., and Coauthors, 2016: The importance of interacting climate modes on Australia’s contribution to global carbon cycle extremes. Sci. Rep., 6, 23113, https://doi.org/10.1038/srep23113.
Cui, W., X. Dong, B. Xi, and A. Kennedy, 2017: Evaluation of reanalyzed precipitation variability and trends using the gridded gauge-based analysis over the CONUS. J. Hydrometeor., 18, 2227–2248, https://doi.org/10.1175/JHM-D-17-0029.1.
Cunha, A. P. M. A., R. C. S. Alvalá, P. Y. Kubota, and R. M. S. P. Vieira, 2015: Impacts of land use and land cover changes on the climate over Northeast Brazil. Atmos. Sci. Lett., 16, 219–227, https://doi.org/10.1002/asl2.543.
Cunha, A. P. M. A., and Coauthors, 2019: Extreme drought events over Brazil from 2011 to 2019. Atmosphere, 10, 642, https://doi.org/10.3390/atmos10110642.
Dai, A. G., and T. M. L. Wigley, 2000: Global patterns of ENSO-induced precipitation. Geophys. Res. Lett., 27, 1283–1286, https://doi.org/10.1029/1999GL011140.
Davies, R., 2016a: Uruguay—Over 10,000 remain displaced by floods as further heavy rain hits. Accessed 17 August 2023, http://floodlist.com/america/uruguay-thousands-displaced-floods-april-2016.
Davies, R., 2016b: Russia—Spring floods leave 8,000 homes damaged. Accessed 17 August 2023, https://floodlist.com/europe/russia-spring-floods-leave-8000-homes-damaged.
Davies, R., 2017: Russia—Hundreds displaced as floods in Far East damage homes and roads. FloodList, accessed 18 August 2023, http://floodlist.com/asia/russia-floods-far-east-august-2017.
de Kok, R. J., P. D. A. Kraaijenbrink, O. A. Tuinenburg, P. N. J. Bonekamp, and W. W. Immerzeel, 2019: Snowfall increase counters glacier demise in Kunlun Shan and Karakoram. Cryosphere Discuss., https://doi.org/10.5194/tc-2019-228, preprint.
Douville, H., and Coauthors, 2021: Water cycle changes. Climate Change 2021: The Physical Science Basis, V. Masson-Delmotte et al., Eds., Cambridge University Press, 1055–1210, https://doi.org/10.1017/9781009157896.010.
Eicker, A., E. Forootan, A. Springer, L. Longuevergne, and J. Kusche, 2016: Does GRACE see the terrestrial water cycle “intensifying”? J. Geophys. Res. Atmos., 121, 733–745, https://doi.org/10.1002/2015JD023808.
Espinoza, J., J. A. Marengo, J. Schongart, and J. C. Jimenez, 2022: The new historical flood of 2021 in the Amazon River compared to major floods of the 21st century: Atmospheric features in the context of the intensification of floods. Wea. Climate Extremes, 35, 100406, https://doi.org/10.1016/j.wace.2021.100406.
FAO, 2019: Afghanistan drought response. Accessed 17 August 2023, http://www.fao.org/fileadmin/user_upload/emergencies/docs/Afghanistan_Drought_response_October_2018.pdf.
FloodList, 2019: West Africa—Death toll rises in Niger, homes destroyed in Nigeria, hundreds displaced in Chad. Accessed 17 August 2023, http://floodlist.com/africa/westafrica-floods-nigeria-niger-chad-september-2019.
FloodList, 2021: China—20,000 evacuate floods in Heilongjiang. Accessed 17 August 2023, https://floodlist.com/asia/china-floods-heilongjiang-june-2021.
Frolova, N. L., and Coauthors, 2017: Hydrological hazards in Russia: Origin, classification, changes and risk assessment. Nat. Hazards, 88, 103–131, https://doi.org/10.1007/s11069-016-2632-2.
Gasparini, N. M., and B. Yuill, 2020: High water: Prolonged flooding on the Deltaic Mississippi River. Eos, 101, https://doi.org/10.1029/2020EO141465.
Geotimes, 2005: Drought in the horn of Africa. Accessed 17 August 2023, http://www.geotimes.org/apr05/WebExtra042905.html.
Gerdener, H., O. Engels, and J. Kusche, 2020: A framework for deriving drought indicators from the Gravity Recovery and Climate Experiment (GRACE). Hydrol. Earth Syst. Sci., 24, 227–248, https://doi.org/10.5194/hess-24-227-2020.
Getirana, A., 2016: Extreme water deficit in Brazil detected from space. J. Hydrometeor., 17, 591–599, https://doi.org/10.1175/JHM-D-15-0096.1.
Hanesiak, J. M., and Coauthors, 2011: Characterization and summary of the 1999–2005 Canadian Prairie drought. Atmos.–Ocean, 49, 421–452, https://doi.org/10.1080/07055900.2011.626757.
Hastenrath, S., D. Polzin, and C. Mutai, 2007: Diagnosing the 2005 drought in equatorial East Africa. J. Climate, 20, 4628–4637, https://doi.org/10.1175/JCLI4238.1.
He, X., and J. Sheffield, 2020: Lagged compound occurrence of droughts and pluvials globally over the past seven decades. Geophys. Res. Lett., 47, e2020GL087924, https://doi.org/10.1029/2020GL087924.
He, X., M. Pan, Z. Wei, E. F. Wood, and J. Sheffield, 2020: A global drought and flood catalogue from 1950 to 2016. Bull. Amer. Meteor. Soc., 101, E508–E535, https://doi.org/10.1175/BAMS-D-18-0269.1.
Heberger, M., 2012: Australia’s millennium drought: Impacts and responses. The World’s Water, Island Press, 97–125, https://doi.org/10.5822/978-1-59726-228-6_5.
Hodgkins, G. A., and Coauthors, 2017: Climate-driven variability in the occurrence of major floods across North America and Europe. J. Hydrol., 552, 704–717, https://doi.org/10.1016/j.jhydrol.2017.07.027.
Hu, G., and C. L. E. Franzke, 2020: Evaluation of daily precipitation extremes in reanalysis and gridded observation-based data sets over Germany. Geophys. Res. Lett., 47, e2020GL089624, https://doi.org/10.1029/2020GL089624.
Huang, X., and D. L. Swain, 2022: Climate change is increasing the risk of a California megaflood. Sci. Adv., 8, eabq0995, https://doi.org/10.1126/sciadv.abq0995.
Jensen, A., 2011: South Australian River Murray floods 2010–2011 breathe life back into river ecosystem! Accessed 18 August 2023, https://blogs.adelaide.edu.au/environment/2011/04/05/south-australian-river-murray-floods-2010-2011-breathe-life-back-into-river-ecosystem-guest-blogger-dr-anne-jensen/.
Jiménez-Muñoz, J. C., C. Mattar, J. Barichivich, A. Santamaría-Artigas, K. Takahashi, Y. Malhi, J. A. Sobrino, and G. van der Schrier, 2016: Record-breaking warming and extreme drought in the Amazon rainforest during the course of El Niño 2015–2016. Sci. Rep., 6, 33130, https://doi.org/10.1038/srep33130.
Kato, H., M. Rodell, F. Beyrich, H. Cleugh, E. van Gorsel, H. Liu, and T. P. Meyers, 2007: Sensitivity of land surface simulations to model physics, land characteristics, and forcings, at four CEOP sites. J. Meteor. Soc. Japan, 85A, 187–204, https://doi.org/10.2151/jmsj.85A.187.
Kelly, R., M. L. Chipman, P. E. Higuera, I. Stefanova, L. B. Brubaker, and F. S. Hu, 2013: Recent burning of boreal forests exceeds fire regime limits of the past 10,000 years. Proc. Natl. Acad. Sci. USA, 110, 13 055–13 060, https://doi.org/10.1073/pnas.1305069110.
Konapala, G., A. K. Mishra, Y. Wada, and M. E. Mann, 2020: Climate change will affect global water availability through compounding changes in seasonal precipitation and evaporation. Nat. Commun., 11, 3044, https://doi.org/10.1038/s41467-020-16757-w.
Kornhuber, K., D. Coumou, E. Vogel, C. Lesk, J. F. Donges, J. Lehmann, and R. M. Horton, 2020: Amplified Rossby waves enhance risk of concurrent heatwaves in major breadbasket regions. Nat. Climate Change, 10, 48–53, https://doi.org/10.1038/s41558-019-0637-z.
Koster, R. D., M. J. Suarez, A. Ducharne, M. Stieglitz, and P. Kumar, 2000: A catchment‐based approach to modeling land surface processes in a general circulation model: 1. Model structure. J. Geophys. Res., 105, 24 809–24 822, https://doi.org/10.1029/2000JD900327.
Kumar, S. V., and Coauthors, 2006: Land information system: An interoperable framework for high resolution land surface modeling. Environ. Modell. Software, 21, 1402–1415, https://doi.org/10.1016/j.envsoft.2005.07.004.
Landerer, F. W., and S. C. Swenson, 2012: Accuracy of scaled GRACE terrestrial water storage estimates. Water Resour. Res., 48, W04531, https://doi.org/10.1029/2011WR011453.
Landerer, F. W., J. O. Dickey, and A. Güntner, 2010: Terrestrial water budget of the Eurasian pan‐Arctic from GRACE satellite measurements during 2003–2009. J. Geophys. Res., 115, D23115, https://doi.org/10.1029/2010JD014584.
Landerer, F. W., and Coauthors, 2020: Extending the global mass change data record: GRACE follow-on instrument and science data performance. Geophys. Res. Lett., 47, e2020GL088306, https://doi.org/10.1029/2020GL088306.
Leblanc, M., P. Tregoning, G. Ramillien, S. O. Tweed, and A. Fakes, 2009: Basin-scale, integrated observations of the early 21st century multiyear drought in southeast Australia. Water Resour. Res., 45, W04408, https://doi.org/10.1029/2008WR007333.
Lehner, F., E. R. Wahl, A. W. Wood, D. B. Blatchford, and D. Llewellyn, 2017: Assessing recent declines in Upper Rio Grande runoff efficiency from a paleoclimate perspective. Geophys. Res. Lett., 44, 4124–4133, https://doi.org/10.1002/2017GL073253.
Levinson, D. H., and M. J. Salinger, 2005: [The tropics] South Pacific and Australia region tropical cyclones [in “State of the Climate in 2004”]. Bull. Amer. Meteor. Soc., 86 (6), S35, https://doi.org/10.1175/1520-0477-86.6s.1.
Li, B., and Coauthors, 2019: Global GRACE data assimilation for groundwater and drought monitoring: Advances and challenges. Water Resour. Res., 55, 7564–7586, https://doi.org/10.1029/2018WR024618.
López-Moreno, J. I., and S. M. Vicente-Serrano, 2008: Positive and negative phases of the wintertime North Atlantic Oscillation and drought occurrence over Europe: A multitemporal-scale approach. J. Climate, 21, 1220–1243, https://doi.org/10.1175/2007JCLI1739.1.
Lorrey, A. M., S. McGree, J. A. Renwick, and S. Hugony, 2012: [Tropics] Southwest Pacific Basin [in “State of the Climate in 2011”]. Bull. Amer. Meteor. Soc., 93 (7), S112, https://doi.org/10.1175/2012BAMSStateoftheClimate.1.
Marengo, J. A., L. M. Alves, R. C. S. Alvala, A. P. Cunha, S. Brito, and O. L. L. Moraes, 2017: Climatic characteristics of the 2010–2016 drought in the semiarid Northeast Brazil region. Ann. Acad. Bras. Cienc., 90, 1973–1985, https://doi.org/10.1590/0001-3765201720170206.
Martin, J. T., and Coauthors, 2020: Increased drought severity tracks warming in the United States’ largest river basin. Proc. Natl. Acad. Sci. USA, 117, 11 328–11 336, https://doi.org/10.1073/pnas.1916208117.
Mason, S. J., and L. Goddard, 2001: Probabilistic precipitation anomalies associated with ENSO. Bull. Amer. Meteor. Soc., 82, 619–638, https://doi.org/10.1175/1520-0477(2001)082<0619:PPAAWE>2.3.CO;2.
McCabe, G. J., M. A. Palecki, and J. L. Betancourt, 2004: Pacific and Atlantic Ocean influences on multidecadal drought frequency in the United States. Proc. Natl. Acad. Sci. USA, 101, 4136–4141, https://doi.org/10.1073/pnas.0306738101.
McClelland, J. W., S. J. Déry, B. J. Peterson, R. M. Holmes, and E. F. Wood, 2006: A pan-Arctic evaluation of changes in river discharge during the latter half of the 20th century. Geophys. Res. Lett., 33, L06715, https://doi.org/10.1029/2006GL025753.
Milly, P. C. D., and K. A. Dunne, 2020: Colorado River flow dwindles as warming-driven loss of reflective snow energizes evaporation. Science, 367, 1252–1255, https://doi.org/10.1126/science.aay9187.
Miyan, M. A., 2015: Droughts in Asian least developed countries: Vulnerability and sustainability. Wea. Climate Extremes, 7, 8–23, https://doi.org/10.1016/j.wace.2014.06.003.
Morales, M. S., and Coauthors, 2020: Six hundred years of South American tree rings reveal an increase in severe hydroclimatic events since mid-20th century. Proc. Natl. Acad. Sci. USA, 117, 16 816–16 823, https://doi.org/10.1073/pnas.2002411117.
NASA Earth Observatory/Event 11787, 2003: Forest fires in eastern Russia. Accessed 17 August 2023, https://earthobservatory.nasa.gov/images/11787/forest-fires-in-eastern-russia.
NASA Earth Observatory/Event 15849, 2005: Food shortages in southeast Africa. Accessed 17 August 2023, https://earthobservatory.nasa.gov/images/15849/food-shortages-in-southeast-africa.
NASA Earth Observatory/Event 49779, 2011: Flooding in western Australia. Accessed 17 August 2023, https://earthobservatory.nasa.gov/images/49779/flooding-in-western-australia.
NASA Earth Observatory/Event 76680, 2011: Lake Eyre floods, South Australia. Accessed 17 August 2023, https://earthobservatory.nasa.gov/images/76680/lake-eyre-floods-south-australia.
NASA Earth Observatory/Event 77712, 2013: Fires in Siberia. Accessed 17 August 2023, https://earthobservatory.nasa.gov/images/event/77712/fires-in-siberia.
NASA Earth Observatory/Event 146015, 2019: Drought threatens millions in southern Africa. Accessed 17 August 2023, https://earthobservatory.nasa.gov/images/146015/drought-threatens-millions-in-southern-africa.
NASA Earth Observatory/Event 146888, 2020: Signs of drought in European groundwater. Accessed 17 August 2023, https://earthobservatory.nasa.gov/images/146888/signs-of-drought-in-european-groundwater.
NASA Earth Observatory/Event 148468, 2021: Brazil battered by drought. Accessed 17 August 2023, https://earthobservatory.nasa.gov/images/148468/brazil-battered-by-drought.
New Humanitarian, 2019: In rural Pakistan, ‘worst drought in years’ drives displacement and hunger. Accessed 17 August 2023, https://www.thenewhumanitarian.org/news/2019/02/06/rural-pakistan-worst-drought-years-drives-displacement-and-hunger.
Ng, B., W. Cai, T. Cowan, and D. Bi, 2018: Influence of internal climate variability on Indian Ocean dipole properties. Sci. Rep., 8, 13500, https://doi.org/10.1038/s41598-018-31842-3.
Nicholson, S. E., and J. C. Selato, 2000: The influence of La Niña on African rainfall. Int. J. Climatol., 20, 1761–1776, https://doi.org/10.1002/1097-0088(20001130)20:14<1761::AID-JOC580>3.0.CO;2-W.
NOAA, 2018: Alaska’s 2018 early fall extremes. Accessed 17 August 2023, https://www.climate.gov/news-features/event-tracker/alaska%E2%80%99s-2018-early-fall-extremes.
NOAA, 2019a: Spring outlook: Historic, widespread flooding to continue through May. Accessed 17 August 2023, https://www.noaa.gov/media-release/spring-outlook-historic-widespread-flooding-to-continue-through-may.
NOAA, 2019b: Monthly National Climate Report for Annual 2019. Accessed 17 August 2023, https://www.ncei.noaa.gov/access/monitoring/monthly-report/national/201913.
NOAA, 2020: Spring flood outlook (updated March 26, 2020). Accessed 17 August 2023, https://www.weather.gov/dvn/2020_springfloodoutlook.
NOAA, 2021: Monthly National Climate Report for Annual 2020. Accessed 17 August 2023, https://www.ncei.noaa.gov/access/monitoring/monthly-report/national/202013.
Pascolini-Campbell, M., J. T. Reager, H. A. Chandanpurkar, and M. Rodell, 2021: A 10 per cent increase in global land evapotranspiration from 2003 to 2019. Nature, 593, 543–547, https://doi.org/10.1038/s41586-021-03503-5.
Peterson, B. J., R. M. Holmes, J. W. McClelland, C. J. Vörösmarty, R. B. Lammers, A. I. Shiklomanov, I. A. Shiklomanov, and S. Rahmstorf, 2002: Increasing river discharge to the Arctic Ocean. Science, 298, 2171–2173, https://doi.org/10.1126/science.1077445.
Phys.org, 2019: At least 265 dead in floods, landslides as rains batter East Africa. Accessed 17 August 2023, https://phys.org/news/2019-12-dead-landslides-batter-east-africa.html.
Reager, J. T., and J. S. Famiglietti, 2009: Global terrestrial water storage capacity and flood potential using GRACE. Geophys. Res. Lett., 36, L23402, https://doi.org/10.1029/2009GL040826.
Reager, J. T., B. S. Thomas, and J. S. Famiglietti, 2014: River basin flood potential inferred using GRACE gravity observations at several months lead time. Nat. Geosci., 7, 588–592, https://doi.org/10.1038/ngeo2203.
Relief Web, 2017: Russian Federation: Floods. Accessed 17 August 2023, https://reliefweb.int/disaster/fl-2017-000052-rus.
Rodell, M., and B. Li, 2023: Changing intensity of hydroclimatic extreme events revealed by GRACE and GRACE-FO. Nat. Water, 1, 241–248, https://doi.org/10.1038/s44221-023-00040-5.
Rodell, M., J. S. Famiglietti, D. N. Wiese, J. T. Reager, H. K. Beaudoing, F. W. Landerer, and M.-H. Lo, 2018: Emerging trends in global freshwater availability. Nature, 557, 651–659, https://doi.org/10.1038/s41586-018-0123-1.
Rowlands, D. D., S. B. Luthcke, S. M. Klosko, F. G. R. Lemoine, D. S. Chinn, J. J. McCarthy, C. M. Cox, and O. B. Anderson, 2005: Resolving mass flux at high spatial and temporal resolution using GRACE intersatellite measurements. Geophys. Res. Lett., 32, L04310, https://doi.org/10.1029/2004GL021908.
Santé, N., Y. A. N’Go, G. E. Soro, N. H. Meledje, and B. T. A. Goula, 2019: Characterization of meteorological droughts occurrences in Côte d’Ivoire: Case of the Sassandra Watershed. Climate, 7, 60, https://doi.org/10.3390/cli7040060.
Save, H., S. Bettadpur, and B. D. Tapley, 2016: High-resolution CSR GRACE RL05 mascons. J. Geophys. Res. Solid Earth, 121, 7547–7569, https://doi.org/10.1002/2016JB013007.
Shah, D., and V. Mishra, 2021: Strong influence of changes in terrestrial water storage on flood potential in India. J. Geophys. Res. Atmos., 126, e2020JD033566, https://doi.org/10.1029/2020JD033566.
Sheffield, J., K. M. Andreadis, E. F. Wood, and D. P. Lettenmaier, 2009: Global and continental drought in the second half of the twentieth century: Severity–area–duration analysis and temporal variability of large-scale events. J. Climate, 22, 1962–1981, https://doi.org/10.1175/2008JCLI2722.1.
Sherstyukov, B. G., and A. B. Sherstyukov, 2014: Assessment of increase in forest fire risk in Russia till the late 21st century based on scenario experiments with fifth-generation climate models. Russ. Meteor. Hydrol., 39, 292–301, https://doi.org/10.3103/S1068373914050021.
Shvidenko, A. Z., and D. G.