1. Introduction
Flash droughts are intense agricultural or hydrologic droughts characterized by the speed with which they are established and the widespread damage they cause to unprepared ecosystems, crops, grazing animals, and people (Basara et al. 2019; Otkin et al. 2022; Svoboda et al. 2002; Yuan et al. 2023; Walker et al. 2024). Unlike conventional droughts, which evolve relatively slowly over several months to years, flash droughts can develop during a few weeks, often faster than the time scales associated with drought monitoring products and early warning systems, undermining drought risk management. Because they occur on such fast subseasonal time scales, flash droughts also pose a substantial challenge for subseasonal-to-seasonal prediction, prompting the wider scientific effort to improve their predictability (Pendergrass et al. 2020).
As with other kinds of droughts, flash droughts occur in different hydroclimatological regimes, resulting in varied impacts on ecosystems and society. The negative agricultural, environmental, and economic impacts of flash drought are significant. For example, the 2012 flash drought in the U.S. Midwest resulted in U.S. $30 billion in agricultural losses (Jin et al. 2019; Hoffmann et al. 2021), while the 2017 flash drought in the U.S. northern plains and southern Canada caused more than $2.6 billion in damages to the U.S. agricultural sector and resulted in fires burning 4.8 million acres across the region (He et al. 2019; Hoell et al. 2020).
The main physical drivers of flash drought are precipitation deficits coupled with enhanced evaporation. As such, Mo and Lettenmaier (2016) have classified flash droughts into heat-wave-driven and precipitation-deficit-driven flash droughts, highlighting the relative dominance of each process in flash drought emergence. While both conditions are common to nearly all drought types, it is the speed at which they occur that sets flash drought apart. Rapid reductions in precipitation and enhancements in evaporation are caused by specific meteorological and land surface conditions. Ridging in the mid- to upper-level troposphere promotes both a reduction in precipitation and increased evaporative demand via lower cloud coverage, increased solar radiation, increased air temperature, and larger vapor pressure deficits (Ford and Labosier 2017; Otkin et al. 2018). Though the elevated temperature is not necessarily required, it is often present during flash drought due to amplifying land–atmosphere feedbacks (Mo and Lettenmaier 2015, 2016). The depletion of soil moisture can lead to a transition from an energy-limited regime to a water-limited regime causing vegetation stress and damage (Otkin et al. 2019). Consequently, despite increased evaporative demand, evaporation eventually decreases due to a lack of moisture, increasing sensible heat flux, near-surface air temperature, and vapor pressure deficit in a positive feedback (Seneviratne et al. 2006).
Vegetation can also play a role in exacerbating flash drought occurrence by increasing evapotranspiration (ET) and amplifying soil moisture decline (Chen et al. 2021). For example, relatively wet conditions across the Great Plains during the spring of 2012 led to abundant vegetation at the start of the summer season. Soon thereafter, a precipitation deficit coupled with above-average temperatures led to high transpiration rates in the densely vegetated region, prompting rapid soil moisture decline and the emergence of flash drought (Sun et al. 2015). It is estimated that with lower rates of transpiration, soil moisture would have declined more slowly.
Chen et al. (2021) performed a sensitivity analysis with the Community Earth System Model, version 2, to investigate the impact of vegetation greening [in this case, a doubling of leaf area index (LAI)] on flash drought occurrence across the United States. They showed amplified flash drought frequency in the Great Plains and the western United States attributable to increased evapotranspiration and depleted soil moisture. At the same time, they showed that flash drought in the eastern United States is not sensitive to vegetation greening, indicating regional heterogeneity in the response. Overall, despite the considerable role that vegetation can have on water availability, for example, through soil moisture feedbacks and land–atmosphere coupling (Chen et al. 2020; Diffenbaugh and Ashfaq 2010; Dirmeyer et al. 2012; Koster et al. 2019; Miralles et al. 2014), its role in flash drought development has received less attention.
Greenhouse gas–driven increases in evaporative demand and precipitation variability are enhancing the risk of flash drought (Parker et al. 2021). In fact, the amplification of evaporative demand and precipitation deficit is increasing the frequency of flash droughts at the expense of traditional, slower-developing drought events (Koster et al. 2019; Yuan et al. 2023). Furthermore, flash droughts themselves are exhibiting shorter onset periods (Qing et al. 2022). Given the severe effects of flash droughts and the projected increase in the frequency of conditions that drive their occurrence with further global warming (Koster et al. 2019; Yuan et al. 2019; Zeng et al. 2023; Walker and Van Loon 2023), it is critical that we monitor flash drought occurrence and development.
Despite extensive research in the past decade, there is no consensus definition of flash drought. Several indices have emerged, designed to detect flash droughts based on evaporation, potential evaporation, and vegetation stress, among others (Hobbins et al. 2016; McEvoy et al. 2016; Anderson et al. 2016; Otkin et al. 2016; Basara et al. 2019; Anderson et al. 2013; Otkin et al. 2014, 2015; L. Chen et al. 2019; L. G. Chen et al. 2019; Ford et al. 2015; Lorenz et al. 2018; Pendergrass et al. 2020). Root-zone soil moisture is a direct measure of plant water availability and is therefore another common choice for flash drought monitoring (Ford and Labosier 2017; Liu et al. 2020; Yuan et al. 2019). However, direct measurements of root-zone soil moisture are not widely available, and soil-moisture-based flash drought detection often requires the use of land surface models.
In the United States, soil water availability, and therefore drought, can be monitored with the use of the North American Land Data Assimilation System, version 2 (NLDAS2) (Xia et al. 2012). The NLDAS produces a quality-controlled and spatially contiguous estimate of land surface water and energy fluxes, streamflow, and water storage using four land surface models (LSMs) forced by observed and reanalyzed meteorological data (more information on NLDAS is provided in the methods section). Notably, NLDAS was developed to reduce soil moisture biases in the land surface models employed in numerical weather prediction. As such, it provides one of the most complete estimates of root-zone soil moisture (Xia et al. 2012) and is therefore a powerful tool for monitoring and studying flash drought.
In its operational mode, the land surface models in NLDAS use a monthly climatology to diagnose vegetation characteristics (such as LAI and greenness). Given the influential role of vegetation on soil moisture and flash drought, it is possible that inaccuracies in the vegetation climatology relative to actual characteristics for each year/location will influence flash drought characteristics and predictability in the NLDAS. Recent studies (Ahmad et al. 2022; Kumar et al. 2019; Mocko et al. 2021) have examined the influence of assimilated satellite data, including vegetation characteristics, on simulated surface hydrology in the NLDAS environment. Kumar et al. (2019) investigated the impact of assimilating Global Land Surface Satellite (GLASS) LAI data into the prognostic vegetation scheme of the Noah LSM with multiparameterization options (Noah-MP). The LAI data assimilation results in the improvement of various hydrologic variables, and hence hydrological drought monitoring, especially over cropland regions due to its benefits in unmodeled processes like irrigation (Mocko et al. 2021). Mocko et al. (2021) further examined the impact of the assimilation of GLASS LAI on Noah-MP soil moisture and drought. Their findings indicate that assimilation of LAI data into Noah-MP improves drought characteristics, including a brief case study of the 2012 Midwest flash drought, particularly over irrigated areas. Ahmad et al. (2022) examined the influence of LAI and soil moisture assimilation on the characteristics of two U.S. Great Plains flash droughts and found improvement in drought onset timing and intensification. While these works suggest that the assimilation of near-real-time vegetation may have an important influence on flash drought detection and characteristics, a systematic climatological analysis of such influence on flash drought has not been conducted.
The objective of this study is to explicitly investigate the influence of LAI assimilation on flash droughts in the NLDAS environment. We examine the flash droughts’ characteristics, including frequency, intensity, and trend in two simulations: one with LAI assimilation implemented into Noah-MP and one without LAI assimilation implemented into Noah-MP. We examine the robustness of our soil-moisture-based flash drought analysis by comparing flash drought detection, characteristics, and trends with those based on the U.S. Drought Monitor (USDM) category definition. The climatological analysis presented here includes an examination of all flash drought events from 1981 to 2017 across the entire CONUS, elucidating robust features of flash drought response to LAI assimilation across hydroclimate regimes and times of the year that are not necessarily apparent in the case study analysis. Results of this work can lead to improved flash drought detection and process understanding, and provide helpful information for flash drought preparation and mitigation efforts in a rapidly changing climate.
2. Methods
a. NLDAS2 data
The NLDAS (Mitchell et al. 2004; Xia et al. 2012) is an operational system which utilizes multiple land surface models in an uncoupled manner with various meteorological forcing data from observations and reanalysis datasets. The forcing data include daily gauge-based precipitation, bias-corrected shortwave radiation, and surface meteorology reanalyses. The NLDAS provides operational near-real-time surface data (4-day lag) on a 1/8° spatial resolution with an hourly time step extending back to 1979 from 25° to 53°N over central North America.
b. Noah-MP and data assimilation
Noah-MP (Niu et al. 2011; Yang et al. 2011) is a comprehensive land surface model used in a variety of research settings, including as a stand-alone model, coupled with the Weather Research and Forecasting (WRF) Model (Barlage et al. 2015; L. Chen et al. 2019), and as a part of the National Water Model (Li et al. 2022). As the name suggests, it was developed to allow users to vary the representations of different surface hydrologic processes to quantify uncertainty, improve process understanding, and identify the most appropriate combination of parameterization schemes for specific research questions. In the simulations with Noah-MP presented here, only a single set of parameterizations are utilized.
Noah-MP includes a full suite of physical processes, including a multilayer snowpack with melt/refreeze capability, surface water infiltration, runoff and groundwater movement, and an unconfined aquifer. The model integrates various vegetation features such as canopy and leaf properties and a representation of the carbon cycle through modeling prognostic vegetation growth and senescence. Noah-MP uses a Ball–Berry photosynthesis-based stomatal resistance model (Ball et al. 1987), which regulates carbon and water fluxes between the plant and the atmosphere. This feature allows Noah-MP to simulate various biophysical processes over time such as photosynthesis, respiration, carbon allocation, mortality, and LAI.
Prognostic vegetation in the land surface model allows the assimilation of LAI into Noah-MP’s dynamic vegetation scheme. To investigate the impact of the assimilated LAI into Noah-MP on soil moisture and flash drought, we examine two separate simulations: the open-loop (OL) simulation and the data assimilation (DA) simulation (Mocko et al. 2021). The two simulations are identical except for their treatment of LAI. In the OL simulation, LAI is simulated using the standard Noah-MP prognostic vegetation model. In the DA simulation, GLASS LAI from the Advanced Very High Resolution Radiometer (AVHRR) is assimilated into Noah-MP (Mocko et al. 2021). The satellite-derived LAI data are available from 1981 to the present with a spatial resolution of 0.05° and a temporal resolution of 8 days. The influence of MODIS LAI data assimilation (available since 2000) and that of AVHRR LAI data assimilation in Noah-MP are similar (Kumar et al. 2019), and AVHRR provides a longer time series of data appropriate for drought studies (Mocko et al. 2021). The land cover classification used in both simulations is the same and comes from the University of Maryland land-cover classification dataset.
The data assimilation algorithm implemented in the DA simulation is a 1D ensemble Kalman filter (EnKF) algorithm (Reichle et al. 2002a,b) applied in the NASA-developed Land Information System (LIS) (Kumar et al. 2016; Peters-Lidard et al. 2007). The ensemble Kalman filter is a sequential method that utilizes an ensemble of simulations and a Monte Carlo approach to estimate the mean and covariance of a Gaussian probability distribution. Here, a uniform observation error standard deviation has been utilized (Kumar et al. 2019). In this process, to establish the ensemble spread and model uncertainty, perturbations are utilized in both the meteorological and model prognostic fields. For more information regarding the assimilation process, please refer to Kumar et al. (2019).
c. Soil-moisture-based flash drought definition
Flash droughts are analyzed for the growing season, chosen here as the months of April–October for the period of 1981–2017. The soil-moisture-based flash drought definition used in this analysis is based on that from Qing et al. (2022) and Yuan et al. (2019). The definition captures the rapid intensification of flash drought and requires that dry conditions persist long enough to cause adverse impacts. To compare the flash drought characteristics yielded from different definitions, we have modified the original definitions of Qing et al. (2022) and Yuan et al. (2019) to use weekly time steps rather than pentad time steps, as outlined next. Based on this definition, a flash drought is detected when weekly root-zone soil moisture (top 1 m) in a grid cell declines from above the weekly climatological 40th percentile to below the weekly climatological 20th percentile in 3 weeks or less and soil moisture remains below the 20th percentile for at least 2 weeks. The flash drought terminates when soil moisture rises above the 20th percentile or when the growing season ends (31 October) (Fig. 1a). Note that since the flash drought detection in April relies on the prior soil moisture conditions in March, flash droughts with onset in the first few weeks of April are not detected in this analysis. The use of climatological soil moisture thresholds rather than fixed soil moisture thresholds reduces the impact of systematic soil moisture biases on flash drought detection. Note that percentiles are calculated separately for the OL and DA simulations.
Application of the modified (a) soil-moisture-based (Qing et al. 2022) and (b) USDM-based (Otkin et al. 2018; Pendergrass et al. 2020) flash drought definitions utilized in this study to a single grid cell in northern Arkansas during the 2012 growing season (see methods). Lines in (a) indicate the (blue) 1-m soil moisture value, (green) climatological 40th percentile 1-m soil moisture value, and (orange) climatological 20th percentile 1-m soil moisture value from DA. The blue line in (b) indicates the USDM-based drought category. Red dashed lines in (a) and (b) indicate the start and end of the flash drought event. Months with no USDM categories represent the time with no drought condition.
Citation: Journal of Hydrometeorology 25, 9; 10.1175/JHM-D-23-0219.1
In our analysis, flash droughts cover the entire period of aridity, not just the short period when soil moisture declines rapidly. For example, a flash drought initiated in April could persist for the entire growing season if soil moisture fails to climb back above the 20th percentile. As such, many of the flash droughts identified here transition to longer, traditional drought episodes. To examine the influence of LAI assimilation on the fraction of flash droughts that transition to longer droughts, and to assess the impact of LAI assimilation on just the period immediately following rapid soil moisture decline in a flash drought, we also compare characteristics of the onset period of flash droughts in the DA and OL simulations. Specifically, we define the onset period of flash droughts as the 2–4-week period after the soil moisture rapidly declines below the 20th percentile.
The soil-moisture-based definition allows us to calculate several flash drought characteristics, including the frequency, mean duration, and mean severity of the flash droughts. The frequency of flash droughts is defined as the average number of events per growing season computed across all the study years. The mean duration is calculated as the average number of weeks in each event. The mean severity is defined as the mean accumulated soil moisture percentile deficit from the 40th percentile. The severity metric incorporates not only the number of weeks in flash drought but also the magnitude of deviation from the 40th percentile.
d. USDM-based definitions
To contextualize the results of the soil-moisture-based flash drought definition, we also utilize a commonly used definition based on the USDM. The USDM is produced by the National Drought Mitigation Center (NDMC) at the University of Nebraska–Lincoln, the U.S. Department of Agriculture (USDA), and the National Oceanic and Atmospheric Administration (NOAA). These data are released every Thursday and show the U.S. regions that are in drought. The map provides five classifications: abnormally dry (D0) and four levels of drought: moderate (D1), severe (D2), extreme (D3), and exceptional (D4). Many inputs, authors, and experts contribute to its production, and many sectors benefit from the map, including farmers, energy producers, and water suppliers.
The USDM is an operational dataset generated based on the climatological (precipitation deficit), agricultural (soil moisture deficit and crop damage), and hydrological (decrease in snow, streamflow, reservoirs, and groundwater level) conditions. The USDM maps have been generated since 2000 in weekly time intervals and are available in ArcGIS shapefile format. To perform flash drought analyses and compare with the soil-moisture-based definition, we have transformed the shapefile maps into a gridded dataset with a spatial resolution of 1/8°, matching that of NLDAS.
The USDM-based flash drought definitions used in this analysis are adopted from Pendergrass et al. (2020) and Otkin et al. (2018), with additional new criteria that specify an initial drought status and allow for a determination of flash drought duration (Fig. 1b). Based on the definitions put forth by Pendergrass et al. (2020), flash drought occurs when there is at least a two-category degradation in USDM categories in 2 weeks which persists at least for another 2 weeks. Here, we suggest that the initial USDM category should be none (normal or wet conditions) or D0 to reflect nondrought starting conditions. Additionally, to allow for a comparison with the soil-moisture-based definition, we suggest that flash drought terminates when there is one category improvement above the category associated with the initial flash drought condition or when the growing season ends. In addition to the Pendergrass et al. (2020) version of flash drought, we examine the definition proposed by Otkin et al. (2018), which detects a moderate flash drought when there is a two-category degradation in USDM categories in the span of 6 weeks (Otkin et al. 2018).
3. Results
We first examine the magnitude and direction of mean growing season LAI biases in the simulations and how those biases influence the simulation of mean growing season soil moisture and flash drought characteristics. We then explore how the evolution of these biases during the growing season impacts flash drought development. We end with an analysis of how these biases influence estimated trends in flash drought during the 1980–2017 period and contextualization of the results within other definitions of flash drought.
a. LAI and soil moisture climatology
Mean growing season LAI is generally similar for the OL and DA simulations for the 1981–2017 period (Figs. 2a–d). But while spatial patterns in LAI are similar between the two simulations, modeling vegetation prognostically as in the OL simulation underestimates the magnitude of mean growing season LAI across most of the United States. The largest absolute underestimations are found in the eastern United States where climatological LAI values are highest. The OL run slightly overestimates LAI in agriculturally intensive regions in the Ohio Valley and southern Great Plains. Relativizing the difference between the OL and DA experiments reveals the spatial pattern of LAI changes due to the data assimilation (Fig. 2d). The largest changes occur in regions with tight soil moisture and vegetation status coupling, like the western United States (Alexander et al. 2022).
The growing season mean (a),(b) LAI and (e),(f) 1-m volumetric soil moisture values in the OL and DA simulations. The mean difference and percent difference between OL and DA for (c),(d) LAI and (g),(h) soil moisture for the 1981–2017 period.
Citation: Journal of Hydrometeorology 25, 9; 10.1175/JHM-D-23-0219.1
Differences in LAI between OL and DA translate directly into root-zone soil moisture differences over much of the domain (Figs. 2e–h). Locations with underestimated LAI in the OL simulation, for example, tend to have higher soil moisture values, and vice versa, suggesting the strong control surface vegetation has on water availability. The differences in vegetation activity and soil moisture are most apparent in the water-limited regions of the western United States (Kumar et al. 2020; Seneviratne et al. 2006; Li et al. 2023). Here, volumetric soil moisture differs by as much as 0.03 m3 m−3 across a broad swath of California, Arizona, and New Mexico. In contrast, in some energy-limited regions, such as the northeast United States and the northern Midwest, where absolute LAI differences are large, growing season mean soil moisture is not impacted (percent changes near 0) by the vegetation differences in the two simulations.
Consistent with previous work (Chen et al. 2021), the magnitude of LAI appears to have a role in initiating flash drought. Two weeks prior to the onset of flash drought, LAI anomalies in both experiments are generally greater than average for the time of year (Figs. S1a,b in the online supplemental material). This is especially the case in the southern, central, and northern plains and portions of the Midwest and coastal Southeast. This suggests that enhanced vegetation activity may accelerate water losses from the soil via transpiration (Zhao et al. 2022), further motivating the examination of LAI assimilation on flash drought characteristics. We note that while LAI appears to be an important driver of flash drought, there are examples where precipitation deficit drives vegetation stress and lower LAI prior to flash drought onset. Interestingly, there are differences between the OL and DA simulations, both in terms of the magnitude of the LAI anomaly and the sign of the LAI anomaly prior to flash drought onset. For example, across much of the Northeast, the LAI anomaly prior to onset is negative in OL and positive in DA. In California and the Great Plains, the LAI anomalies are much higher prior to drought onset in the OL simulation.
b. Growing season mean flash drought climatology
Most of the United States has experienced flash drought at some point during the study period of 1981–2017, and some regions have averaged almost one flash drought per growing season, according to the DA simulation (Fig. 3b). Hotspots of flash drought activity occur across a diverse set of hydroclimatological regimes, including the semiarid Great Plains and humid Southeast, Northwest, Northeast, and northern Michigan, all of which show areas with an average of 0.7–0.9 flash droughts per growing season. Based on a comparison with soil texture (Fig. S2), flash drought detection in several of these hotspot areas appears to be heavily influenced by the rapid drainage of water through sandy soil (e.g., Nebraska, northern Michigan, Florida, and the coastal Southeast) (Patel et al. 2021).
(a)–(d) A comparison of the average number of flash droughts per growing season, (e)–(h) duration average, (i)–(l) severity average, and (m)–(p) the percentage of time in flash drought between the OL and DA simulations.
Citation: Journal of Hydrometeorology 25, 9; 10.1175/JHM-D-23-0219.1
The representation of vegetation in the two experiments plays an important role in flash drought characteristics, including duration, severity, and percentage of the growing season (Figs. 3b–p). Flash drought duration is 1–5 weeks shorter on average in OL across portions of the Midwest, south, Intermountain West, and central and northern Great Plains. This represents up to a 100% reduction in average flash drought duration in these regions (Fig. 3h). Isolated areas of longer OL flash droughts are scattered across the Great Plains and western United States, though coherent spatial patterns of longer OL flash drought duration are difficult to discern.
The shorter flash drought duration in OL contributes, in part, to more frequent flash drought occurrence (the number of times a flash drought is initiated) compared with DA. There are about 10%–70% more individual flash drought events across much of the western and central United States. Indeed, flash drought duration is inversely related to the frequency of flash droughts (cf. Figs. 3a,b,e,f). Given that OL detects more flash drought events even in regions without large duration differences between the simulations, it is clear that event duration is not the only factor driving greater OL flash drought frequency.
The average severity of flash droughts is greater in DA, particularly in the Midwest, northern plains, and the South, consistent with generally longer duration flash drought events (Figs. 3i–l). This represents up to a 70% reduction in average flash drought severity in OL compared with DA in these regions. The percentage of time in flash drought (Figs. 3m–p) combines flash drought occurrence and duration into one metric. The OL model run simulates more total time in flash drought across much of the United States, especially across central and western portions of the domain. There are large differences (>7% more of the total time) in the Central Valley of California and eastern Washington where LAI assimilation in DA reflects the use of irrigation and associated healthy vegetation status measured from satellite data (further details on the influence of irrigation is presented in the discussion and conclusions section). Across much of the eastern United States, relatively small differences in flash drought frequency and duration between DA and OL lead to little to no difference in the fraction of time in flash drought (from −2% to 2%) between the simulations.
Next, we examine the influence of LAI differences on the percentage of flash droughts that transition to “traditional” longer-lasting drought events. Here, we consider a flash drought to have transitioned to a traditional drought if the period in which the soil moisture remains below the 20th percentile persists for 5 weeks or longer (see methods; e.g., Basara et al. 2019). Flash droughts in the western United States, Midwest, and Ohio Valley have a higher chance of transitioning to drought than in other parts of the country (around 80% in OL and 85% in DA) (Fig. 4). The percentage of flash droughts that do not transition to drought (i.e., 2-, 3-, or 4-week duration) is more likely to occur (around 65% in OL and 60% in DA) in the Great Plains, the coastal Southeast, and parts of northern New England. In general, flash droughts in the DA simulation are slightly more likely to transition into longer traditional droughts (this is true in 57% of grid cells) (Fig. 4c). However, flash droughts in large portions of Texas, Kansas, Wyoming, and Montana are more likely to transition to drought in OL though this figure shows little spatial coherency for drought detection. Further consideration of the factors underlying the lack of spatial coherency and the potential for spatially aggregating the data would be useful topics for future work. The differences in the simulation of the onset period of flash droughts (2–4 weeks) are similar to the differences in the full flash drought events (Fig. S3).
(a),(b) A comparison of the percentage of flash droughts that transition to drought (i.e., duration of 5 or more weeks) and (c) the difference between OL and DA.
Citation: Journal of Hydrometeorology 25, 9; 10.1175/JHM-D-23-0219.1
c. Monthly analysis of flash droughts and the role of vegetation
Given that simulated LAI biases at a particular location can vary in magnitude and sign within the growing season, we next examine model output at the monthly scale to better understand the influence of assimilated LAI on flash drought. First, we examine the simulation of flash drought in OL and DA for each of the growing season months. Then, we explore the ways in which LAI influences the monthly scale differences in the two simulations.
The overall spatial patterns and monthly evolution of flash drought in OL and DA resemble one another (Fig. 5). In April, for example, flash droughts are primarily confined to the eastern United States, with a hotspot over the southern Mississippi River basin. As early spring transitions to late spring and summer, flash droughts increase across much of the CONUS. In May and June, flash droughts occur most often in the Northwest United States, the Rocky Mountains, the Great Lakes region, portions of the Northeast, and the southern Mississippi River basin, with some locations averaging 15%–20% of weeks in flash drought at this time. As summer progresses to July, the areal extent of flash drought increases to include the Great Plains, and the average percentage of weeks in flash drought reaches its maximum for the CONUS as a whole. Flash drought occurrence begins to subside in August and, with the exception of Florida, continues to decrease into September. By October, flash droughts are somewhat uncommon, comprising less than 10% of weeks across all of the United States, except in Florida where they occur roughly 15% of weeks.
The percentage of time under flash drought in (left) OL and (right) DA for each of the growing season months.
Citation: Journal of Hydrometeorology 25, 9; 10.1175/JHM-D-23-0219.1
Next, we examine the climatological monthly mean differences between OL and DA (OL − DA) for LAI, soil moisture, and the percentage of time in flash drought (Fig. 6). Climatological monthly mean differences in LAI values reveal that the slight overestimation of mean growing season LAI across the Ohio Valley and southern Great Plains in OL (Fig. 2c) mostly occurs in the early growing season (April–June) and in October (note the brown shading in this region indicates DA has lower LAI values than OL), with LAI values up to 1.2 m2 m−2 smaller in DA. The overestimation of OL LAI early in the growing season in these regions leads to a relative reduction (0.005–0.02 m3 m−3, i.e., up to 10%) in volumetric soil moisture due to enhanced transpiration rates. The percentage of time in flash drought is greater in OL by roughly 7% across much of the Ohio Valley, Midwest, and Great Plains during this time.
The climatological monthly mean differences between OL and DA (OL–DA) for (left) LAI, (center) soil moisture, and (right) flash drought during April–October.
Citation: Journal of Hydrometeorology 25, 9; 10.1175/JHM-D-23-0219.1
Elsewhere during the early growing season, LAI values are generally lower in OL, with the largest absolute underestimations (between 2 and 3 m2 m−2) in the eastern United States. As noted earlier, the underestimation of LAI leads to relatively greater soil moisture in OL across the interior West, northern Great Plains, and Southeast. In June, the relatively lower soil moisture in DA coincides with a greater fraction of time in flash drought across the Southeast, the western Dakotas, and large portions of Colorado. However, the impact of lower DA soil moisture on flash drought is relatively minor across most regions during the early growing season.
As summer progresses, the simulation of flash drought diverges between the two model runs. The persistently greater LAI in the DA simulation throughout the growing season eventually results in noticeably reduced soil moisture values compared with OL near the end of the growing season (note the lack of blue shading in August through September in Figs. 6n,q). This tends to bring on slightly more flash droughts in the DA compared with OL during September and October across the central United States and the eastern United States, which is largely not the case in the early growing season months of May and June. While it is clear soil moisture is influenced by LAI biases, the impact on flash drought is not always straightforward. For example, lower soil moisture in DA relative to OL does not always lead to greater flash drought. Throughout the spring and summer, soil moisture is considerably lower (−0.02 m3 m−3) across irrigated regions of the Central Valley of California and eastern Washington and Oregon in DA. In these areas, reduced soil moisture in DA coincides with a greater than 12% reduction in the percentage of time in flash drought. Similarly, slightly above-average soil moisture in July across the Ohio Valley and Midwest in DA is associated with a 7% greater likelihood of time in flash drought. Unlike conventional agricultural drought, a reduction in soil moisture could in fact make flash drought less likely because soil moisture remains below the 40th percentile and therefore, by definition, will not meet the criteria for a flash drought event (see methods).
d. LAI and soil moisture trends
Given the widespread biases in the simulation of climatological LAI, it is likely that the OL simulation also misrepresents the response of LAI, and therefore flash drought, to recent observed changes in climate. Indeed, trend analysis shows that modeling vegetation prognostically in the OL simulation leads to a misrepresentation of the historical evolution of LAI changes during the past several decades (Fig. 7). Satellite measurements indicate a positive LAI trend across most of the central and eastern United States, as well as the coastal Pacific Northwest, and negative trends across portions of California, the northern Rockies, the northern Midwest, and the Southwest during this time, driven by CO2 fertilization, land cover change, and precipitation trends, as previously reported in Zhu et al. (2016). However, the OL simulation misses many of these changes and, in fact, often simulates a trend in the opposite direction (Fig. 7b), especially across the central and eastern United States.
The trend in (a),(b) LAI; (d),(e) soil moisture; and (g),(h) flash drought occurrence over time (growing season 1981–2017) in OL and DA and (c),(f),(i) whether the sign of that trend agrees.
Citation: Journal of Hydrometeorology 25, 9; 10.1175/JHM-D-23-0219.1
The misrepresentation of the LAI trend in OL helps to explain the divergence in root-zone soil moisture trends between the OL and DA simulations in the central and southeastern United States (Figs. 7d,e). According to the DA simulation, much of the United States experienced a negative trend in root-zone soil moisture, with notable exceptions in the Northeast, northern Great Plains, and portions of the Midwest. Meanwhile, the OL simulation shows increases in soil moisture across much of the central and southern United States, in disagreement with DA (Fig. 7f). Elsewhere, the sign of the soil moisture trend in OL agrees with DA, but the magnitude of the trend varies, especially in the southwest United States where DA simulates much larger reductions in soil moisture over time. Figure S5 shows the statistically significant trends.
e. Flash drought trends
Disagreement in the trends of LAI and soil moisture has the potential to influence agreement on the magnitude and sign of flash drought trends in OL and DA. Figures 7g–i show the trend in flash drought occurrence over time in OL and DA and whether the sign of that trend agrees. Specifically, for each grid cell, we calculate whether a flash drought occurred during each year, assigning the year a binary value of 1 or 0, and then calculate the trend in those binary values across the 37-yr period. In the DA simulation, the number of years with a flash drought has been increasing across much of the United States, especially in the far southern and northern Great Plains, Midwest, portions of the Rockies, and Southwest United States. The OL simulation also exhibits a positive trend in the number of years with a flash drought in portions of the Southwest and Midwest but shows negative trends across much of the Great Plains, in disagreement with DA. Other regions of notable disagreement are found in Montana, Georgia, and South Carolina and along southern areas of the Mississippi River. Additionally, analysis conducted with the Mann–Kendall significance test shows that the trend in flash drought occurrence aggregated across the United States is not significant (p value: 0.97) in OL, with a very small positive slope, but is clearly positive and significant in DA (p value: 0.05) (Fig. 8). These trends are not significant if only focusing on the onset period of flash drought events though DA still exhibits an increasing trend (Fig. S4).
Comparison of the trend in flash drought occurrence averaged across the continental United States (growing season 1981–2017) in the DA and OL simulations.
Citation: Journal of Hydrometeorology 25, 9; 10.1175/JHM-D-23-0219.1
f. Comparison of soil-moisture-based flash drought with USDM-based flash drought
Overall, the comparison of soil-moisture- and USDM-based definitions shows consistency in terms of the area percentage in flash drought though the level of agreement varies year to year (Fig. 9). Here, we compare the percentage of the continental United States under flash drought during the growing season (April–October) for the years 2000–2017 from the soil-moisture-based flash drought definition and the USDM-based flash drought definitions put forth in Pendergrass et al. (2020) and Otkin et al. (2018). Notable deviations between the definitions occur in 2001, 2004, 2009, and 2014 when the soil-moisture-based definition detects widespread flash droughts, but the USDM-based definitions do not. Regarding the USDM-based definitions, it appears that the Otkin definition, with its 6-week category degradation criteria (see methods) and focus on detecting strong- and moderate-intensity flash drought events, shows better agreement with the soil-moisture-based definition than the Pendergrass definition, which uses a 2-week category degradation criterion (see methods) and as such detects substantially fewer events.
The percentage of the CONUS under flash drought during the growing season (April–October) for the years 2000–17 in the DA and OL simulations based on the soil-moisture-based and USDM-based flash drought definitions.
Citation: Journal of Hydrometeorology 25, 9; 10.1175/JHM-D-23-0219.1
As shown in Fig. 10, flash drought definition substantially influences flash drought frequency, duration, and the percentage of time in flash drought. Flash drought frequency is higher across nearly all of the United States using the soil-moisture-based definition in the DA simulation. During the 2000–2017 period, the soil-moisture-based definition identifies several locations with an average of almost 1 flash drought per year. The USDM-based definitions suggested by Pendergrass et al. (2020) and Otkin et al. (2018), respectively, show locations with a maximum frequency of 0.3 and 0.6 flash droughts per year. While the spatial pattern of flash drought frequency is broadly similar between the definitions, the soil-moisture-based definition identifies high frequency in the Northwest, eastern Great Lakes, portions of the Northeast, and Florida that are not present in the USDM-based definitions. As noted earlier, the higher number of flash droughts captured by the soil-moisture-based definition could be partly associated with the definition’s sensitivity to certain soil textures, which leads to rapid drainage in soil moisture and the initiation of flash drought.
Flash drought characteristics in (a),(d),(g) the soil-moisture-based definition (DA) and (b),(e),(h),(c),(f),(i) the USDM-based definitions (Otkin et al. 2018; Pendergrass et al. 2020) over the growing season in the 2000–17 time period. Event duration is plotted white in locations where flash droughts are not detected.
Citation: Journal of Hydrometeorology 25, 9; 10.1175/JHM-D-23-0219.1
The duration of flash droughts also varies in the two different types of definitions (Figs. 10d–f) such that the USDM-based definitions generally exhibit longer drought events (5–15 weeks longer) across most of the United States, notably seen in the Great Plains and portions of the Intermountain West. Last, the percentage of time in flash drought reaches as high as 20% of the growing season in some areas using the soil-moisture-based and Otkin definitions, whereas the Pendergrass definition maxes out at close to 10% of the growing season. Overall, though considerable differences exist between the soil-moisture-based definition and the USDM-based definitions, there is better agreement between the soil-moisture-based definition and the definition suggested by Otkin et al. (2018). The USDM-based definition proposed by Pendergrass et al. (2020) appears to detect only the most intense flash droughts consistent with the findings of Schwartz et al. (2023).
Assimilation of updated LAI information impacts the agreement of flash drought detection between the soil-moisture-based and USDM-based definitions (Otkin) (Fig. 11). Temporal and spatial correlation analyses over the 2000–17 period between flash drought in OL and USDM and DA and USDM show that LAI assimilation into the Noah-MP LSM generally leads to better agreement (up to 0.6 and 0.2 higher temporal and spatial correlation, respectively) of flash drought detection with the USDM-based definition. Figure 11 indicates that improvements in the agreement are especially prominent in irrigated regions with groundwater pumping such as in the Corn Belt and Ogallala Aquifer regions. However, in some years (2003 and 2004) and locations (e.g., portions of the Interior West), LAI assimilation results in less agreement with the USDM-based definition. It is important to note that the USDM-based definition is not necessarily the “truth,” and the comparison presented here is used to simply assess agreement between the different flash drought definitions.
The agreement of flash drought detection between the soil-moisture-based (DA) and USDM-based (Otkin et al. 2018) definitions based on the temporal and spatial correlation analyses over the growing season in the 2000–17 time period.
Citation: Journal of Hydrometeorology 25, 9; 10.1175/JHM-D-23-0219.1
4. Discussion and conclusions
Averaged across the growing season, the Noah-MP simulation with prognostic vegetation (OL) consistently underestimates LAI across much of the United States, causing relatively high soil moisture values compared to the simulation with LAI assimilation (DA). An important deviation from this pattern is the overestimation of modeled LAI in the Midwest and portions of the eastern Great Plains in the spring season, which drives relatively lower soil moisture values. The widespread LAI underestimation in the prognostic Noah-MP land surface model is likely due to a lack of or misrepresentation of land surface processes like irrigation and vegetation dynamics (Mocko et al. 2021), and the Midwest spring season overestimation is due to the modeled growth and phenology of vegetation as affected by temperature, soil moisture, and radiation which can cause an earlier spring leaf-out or greening (Hosseini et al. 2022). Despite the clear improvement in the representation of LAI with assimilation, it is challenging to say definitively whether the inclusion of LAI assimilation improves the rate of soil moisture decline, and hence flash drought characteristics in NLDAS given the lack of widespread soil moisture observations and little consensus in the literature on flash drought definition.
Previous annual mean soil moisture evaluation efforts conducted by Mocko et al. (2021) indicate that root-zone soil moisture values in DA compare more favorably with many individual soil moisture observation sites than OL. However, in some locations, the OL soil moisture values match better. More research is needed to understand how changes to other land surface processes, including transpiration, photosynthesis, and the movement of water through the soil column, may need to be altered to accommodate the most successful implementation of LAI assimilation. Additionally, soil moisture can be assimilated directly into the land surface model (Crow et al. 2024). This may be especially beneficial to the study of flash droughts in regions where LAI is not the primary driver of drought onset. Despite the uncertainty, the soil-moisture-based flash droughts in the simulation with LAI assimilation generally match those of the USDM-based flash droughts better than those from the prognostic vegetation simulation (Fig. 11). This is consistent with findings from Mocko et al. (2021) investigating the impact of LAI assimilation into Noah-MP on traditional drought across the CONUS and suggests the added value of vegetation assimilation in NLDAS for flash drought representation (Ahmad et al. 2022; Mocko et al. 2021).
While a systematic evaluation of LAI assimilation on traditional drought has been conducted (Mocko et al. 2021), little work has focused on the effects of LAI assimilation on flash drought (Ahmad et al. 2022). Because flash droughts, by definition, involve the rapid drawdown of soil moisture from near-normal soil moisture conditions, their response to LAI assimilation can differ from that of traditional drought. While LAI-driven reductions (increases) in soil moisture increase (decrease) the likelihood of traditional agricultural drought development, the impact on flash drought is not always straightforward. For example, if a model with prognostic vegetation slightly overestimates LAI compared with reality, soil moisture in the model may fall below the climatological 40th percentile value prior to the onset of atmospheric conditions that drive flash drought. By definition, a flash drought will not be detected in the model even if soil moisture declines rapidly because the starting soil moisture value did not exceed the 40th percentile. This behavior helps to explain why, in some instances, lower (higher) soil moisture in the DA simulation does not lead to greater (fewer) flash droughts (Fig. 6). This type of behavior may also be present in some irrigated regions where LAI and soil moisture are inconsistent with one another in the DA simulation. For instance, the relatively higher LAI in the Central Valley of California in DA (Fig. 2) is the result of irrigation, yet the DA simulation does not incorporate the effects of irrigation on soil moisture. Consequently, the relatively high LAI draws soil moisture below the climatological 40th percentile, resulting in fewer flash drought events compared with OL (Fig. 3). This suggests that the joint assimilation of soil moisture and LAI could be particularly useful for flash drought detection in some agricultural areas (e.g., Sabater et al. 2008; Kumar et al. 2015; Ahmad et al. 2022; Yin et al. 2023). Despite these complexities, our analysis shows that LAI assimilation has a substantial influence on simulated soil moisture content and flash drought characteristics, and several general observations about the effect of LAI data assimilation on flash drought in NLDAS can be made.
A notable finding of this assessment of LAI assimilation on flash drought climatology is that LAI assimilation generally makes flash drought events longer (Fig. 3) as a result of enhanced LAI, and therefore transpiration, observed by the satellites during much of the growing season, in line with studies of Sun et al. (2015) and Chen et al. (2021). This slightly increases the likelihood that flash droughts transition to more traditional longer-lasting droughts in the simulation with LAI assimilation. The tendency that the flash drought would transition to a longer drought is potentially helpful in drought early warning (Otkin et al. 2022). For instance, if we know that 80% of flash drought events transition to drought events of at least 5-week duration, as found in the Pacific Northwest and Midwest United States, we can hopefully plan accordingly and effectively manage water resources once we know the flash drought has begun.
Our results also have interesting implications for our understanding of flash drought behavior in NLDAS over the past several decades. The prognostic vegetation simulation with Noah-MP does not capture the observed trends in LAI shown in the vegetation assimilation simulation (Fig. 7). As a result, the prognostic vegetation and data assimilation simulations do not agree on the trend in root-zone soil moisture across much of the Great Plains and Southeast United States and therefore indicate opposite flash drought trends across much of these regions as well (Fig. 7). Like our findings, Osman et al. (2021) indicate an overall increasing trend in historical flash droughts across the United States. Understanding how flash droughts are changing in the current climate is needed for confidence in model projections of future flash droughts, and the discrepancy between the trends in the DA and OL simulations highlights the critical need to better understand this response. This is especially the case given that future trends of flash droughts from models show increasing severity and risk in populated places like China (Yuan et al. 2019) and an increasing number of flash droughts with accelerated onset (i.e., less time to prepare) across the globe (Qing et al. 2022).
Though the NLDAS data and soil-moisture-based flash drought definition presented here are commonly used in flash drought research (e.g., Ahmad et al. 2022; Chen et al. 2021; Mishra et al. 2021; Qing et al. 2022), considerable variability in the choice of drought monitoring variable and in flash drought definition remains. Comparison of our results with those from a global study that utilized a soil-moisture-based flash drought definition and ERA5 data (Mukherjee and Mishra 2022) shows that there are some discrepancies in flash drought hotspots (mostly visible in the northern Rockies), driven by different simulations of soil moisture in the datasets, highlighting the large uncertainty around this variable. We find similar flash drought location hotspots as those presented in Koster et al. (2019). However, the frequency of flash droughts in their study is, in general, much lower, with at most 0.2 events per year across North America. The reasons for disagreement in the number of flash droughts could be related to both the dataset (MERRA-2) and the flash drought definition, which is based on precipitation deficit and excess ET. Employing the standardized evaporative stress index, the study of Christian et al. (2021) also exhibits some similarities and differences in flash drought hotspots with our study. They also find high flash drought frequency in the Great Plains and coastal Southeast, but unlike our study, they find a high frequency of flash droughts in the Midwest (up to 0.8 events per year). Mismatches between our results and other studies are consistent with the findings of Qing et al. (2022), who point out that inconsistencies in global flash drought detection and characteristics are widespread due to variations in input data and flash drought definition. Altogether, this highlights the need for observation-based spatially and temporally contiguous soil moisture data and greater awareness of the strengths and weaknesses of different flash drought definitions (Sehgal et al. 2021; Lisonbee et al. 2022; Ford et al. 2023).
While the assimilation of LAI into Noah-MP does not fix all land surface model biases (Kumar et al. 2019), the results presented here suggest that our current statistics and understanding of flash drought in the NLDAS system may be incomplete without the proper representation of vegetation. As such, we recommend further research into LAI assimilation and its influence on drought in NLDAS.
Acknowledgments.
Ali Fallah, Christopher Skinner, Matthew Barlow, and Laurie Agel acknowledge funding support by Grant NA20OAR4310424 from NOAA’s Climate Program Office’s Modeling, Analysis, Predictions, and Projections program. Justin Mankin acknowledges funding support from DOE DESC0022302 and NOAA MAPP NA20OAR4310425. We would like to acknowledge high-performance computing support from Cheyenne (https://doi.org/10.5065/D6RX99HX) provided by NCAR’s Computational and Information Systems Laboratory, sponsored by the National Science Foundation.
Data availability statement.
USDM maps are from the collaboration between the NDMC, NOAA, and USDA (https://www.drought.gov/data-maps-tools/us-drought-monitor). The NLDAS2 forcing data used in this effort were acquired as part of the activities of NASA’s Science Mission Directorate and are archived and distributed by the Goddard Earth Sciences (GES) Data and Information Services Center (DISC) (https://ldas.gsfc.nasa.gov/nldas/nldas-get-data).
REFERENCES
Ahmad, S. K., and Coauthors, 2022: Flash drought onset and development mechanisms captured with soil moisture and vegetation data assimilation. Water Resour. Res., 58, e2022WR032894, https://doi.org/10.1029/2022WR032894.
Alexander, G. A., H. A. Holmes, X. Sun, D. Caputi, I. C. Faloona, and H. J. Oldroyd, 2022: Simulating land-atmosphere coupling in the Central Valley, California: Investigating soil moisture impacts on boundary layer properties. Agric. For. Meteor., 317, 108898, https://doi.org/10.1016/j.agrformet.2022.108898.
Anderson, M. C., C. Hain, J. Otkin, X. Zhan, K. Mo, M. Svoboda, B. Wardlow, and A. Pimstein, 2013: An intercomparison of drought indicators based on thermal remote sensing and NLDAS-2 simulations with U.S. drought monitor classifications. J. Hydrometeor., 14, 1035–1056, https://doi.org/10.1175/JHM-D-12-0140.1.
Anderson, M. C., and Coauthors, 2016: The evaporative stress index as an indicator of agricultural drought in Brazil: An assessment based on crop yield impacts. Remote Sens. Environ., 174, 82–99, https://doi.org/10.1016/j.rse.2015.11.034.
Ball, J. T., I. E. Woodrow, and J. A. Berry, 1987: A model predicting stomatal conductance and its contribution to the control of photosynthesis under different environmental conditions. Progress in Photosynthesis Research, J. Biggins, Ed., Vol. 4, Springer, 221–224, https://doi.org/10.1007/978-94-017-0519-6_48.
Barlage, M., M. Tewari, F. Chen, G. Miguez-Macho, Z.-L. Yang, and G.-Y. Niu, 2015: The effect of groundwater interaction in North American regional climate simulations with WRF/Noah-MP. Climatic Change, 129, 485–498, https://doi.org/10.1007/s10584-014-1308-8.
Basara, J. B., J. I. Christian, R. A. Wakefield, J. A. Otkin, E. H. Hunt, and D. P. Brown, 2019: The evolution, propagation, and spread of flash drought in the Central United States during 2012. Environ. Res. Lett., 14, 084025, https://doi.org/10.1088/1748-9326/ab2cc0.
Chen, L., Y. Li, F. Chen, M. Barlage, Z. Zhang, and Z. Li, 2019: Using 4-km WRF CONUS simulations to assess impacts of the surface coupling strength on regional climate simulation. Climate Dyn., 53, 6397–6416, https://doi.org/10.1007/s00382-019-04932-9.
Chen, L., T. W. Ford, and P. Yadav, 2021: The role of vegetation in flash drought occurrence: A sensitivity study using community earth system model, version 2. J. Hydrometeor., 22, 845–857, https://doi.org/10.1175/JHM-D-20-0214.1.
Chen, L. G., J. Gottschalck, A. Hartman, D. Miskus, R. Tinker, and A. Artusa, 2019: Flash drought characteristics based on U.S. drought monitor. Atmosphere, 10, 498, https://doi.org/10.3390/atmos10090498.
Chen, L. G., A. Hartman, B. Pugh, J. Gottschalck, and D. Miskus, 2020: Real-time prediction of areas susceptible to flash drought development. Atmosphere, 11, 1114, https://doi.org/10.3390/atmos11101114.
Christian, J. I., J. B. Basara, E. D. Hunt, J. A. Otkin, J. C. Furtado, V. Mishra, X. Xiao, and R. M. Randall, 2021: Global distribution, trends, and drivers of flash drought occurrence. Nat. Commun., 12, 6330, https://doi.org/10.1038/s41467-021-26692-z.
Crow, W. T., H. Kim, and S. Kumar, 2024: Systematic modeling errors undermine the application of land data assimilation systems for hydrological and weather forecasting. J. Hydrometeor., 25, 3–26, https://doi.org/10.1175/JHM-D-23-0069.1.
Diffenbaugh, N. S., and M. Ashfaq, 2010: Intensification of hot extremes in the United States. Geophys. Res. Lett., 37, L15701, https://doi.org/10.1029/2010GL043888.
Dirmeyer, P. A., and Coauthors, 2012: Evidence for enhanced land–atmosphere feedback in a warming climate. J. Hydrometeor., 13, 981–995, https://doi.org/10.1175/JHM-D-11-0104.1.
Ford, T. W., and C. F. Labosier, 2017: Meteorological conditions associated with the onset of flash drought in the Eastern United States. Agric. For. Meteor., 247, 414–423, https://doi.org/10.1016/j.agrformet.2017.08.031.
Ford, T. W., D. B. McRoberts, S. M. Quiring, and R. E. Hall, 2015: On the utility of in situ soil moisture observations for flash drought early warning in Oklahoma, USA. Geophys. Res. Lett., 42, 9790–9798, https://doi.org/10.1002/2015GL066600.
Ford, T. W., J. A. Otkin, S. M. Quiring, J. Lisonbee, M. Woloszyn, J. Wang, and Y. Zhong, 2023: Flash drought indicator intercomparison in the United States. J. Appl. Meteor. Climatol., 62, 1713–1730, https://doi.org/10.1175/JAMC-D-23-0081.1.
He, M., J. S. Kimball, Y. Yi, S. Running, K. Guan, K. Jensco, B. Maxwell, and M. Maneta, 2019: Impacts of the 2017 flash drought in the US Northern plains informed by satellite-based evapotranspiration and solar-induced fluorescence. Environ. Res. Lett., 14, 074019, https://doi.org/10.1088/1748-9326/ab22c3.
Hobbins, M. T., A. Wood, D. J. McEvoy, J. L. Huntington, C. Morton, M. Anderson, and C. Hain, 2016: The evaporative demand drought index. Part I: Linking drought evolution to variations in evaporative demand. J. Hydrometeor., 17, 1745–1761, https://doi.org/10.1175/JHM-D-15-0121.1.
Hoell, A., and Coauthors, 2020: Lessons learned from the 2017 flash drought across the U.S. Northern Great Plains and Canadian prairies. Bull. Amer. Meteor. Soc., 101, E2171–E2185, https://doi.org/10.1175/BAMS-D-19-0272.1.
Hoffmann, D., A. J. E. Gallant, and M. Hobbins, 2021: Flash drought in CMIP5 models. J. Hydrometeor., 22, 1439–1454, https://doi.org/10.1175/JHM-D-20-0262.1.
Hosseini, A., D. M. Mocko, N. A. Brunsell, S. V. Kumar, S. Mahanama, K. Arsenault, and J. K. Roundy, 2022: Understanding the impact of vegetation dynamics on the water cycle in the Noah-MP model. Front. Water, 4, 925852, https://doi.org/10.3389/frwa.2022.925852.
Jin, C., X. Luo, X. Xiao, J. Dong, X. Li, J. Yang, and D. Zhao, 2019: The 2012 flash drought threatened US Midwest agroecosystems. Chin. Geogr. Sci., 29, 768–783, https://doi.org/10.1007/s11769-019-1066-7.
Koster, R. D., S. D. Schubert, H. Wang, S. P. Mahanama, and A. M. DeAngelis, 2019: Flash drought as captured by reanalysis data: Disentangling the contributions of precipitation deficit and excess evapotranspiration. J. Hydrometeor., 20, 1241–1258, https://doi.org/10.1175/JHM-D-18-0242.1.
Kumar, S., M. Newman, D. M. Lawrence, M.-H. Lo, S. Akula, C.-W. Lan, B. Livneh, and D. Lombardozzi, 2020: The GLACE-hydrology experiment: Effects of land–atmosphere coupling on soil moisture variability and predictability. J. Climate, 33, 6511–6529, https://doi.org/10.1175/JCLI-D-19-0598.1.
Kumar, S. V., C. D. Peters-Lidard, J. A. Santanello, R. H. Reichle, C. S. Draper, R. D. Koster, G. Nearing, and M. F. Jasinski, 2015: Evaluating the utility of satellite soil moisture retrievals over irrigated areas and the ability of land data assimilation methods to correct for unmodeled processes. Hydrol. Earth Syst. Sci., 19, 4463–4478, https://doi.org/10.5194/hess-19-4463-2015.
Kumar, S. V., and Coauthors, 2016: Assimilation of gridded GRACE terrestrial water storage estimates in the North American land data assimilation system. J. Hydrometeor., 17, 1951–1972, https://doi.org/10.1175/JHM-D-15-0157.1.
Kumar, S. V., D. M. Mocko, S. Wang, C. D. Peters-Lidard, and J. Borak, 2019: Assimilation of remotely sensed leaf area index into the Noah-MP land surface model: Impacts on water and carbon fluxes and states over the continental United States. J. Hydrometeor., 20, 1359–1377, https://doi.org/10.1175/JHM-D-18-0237.1.
Li, J., C. Miao, G. Zhang, Y.-H. Fang, W. Shangguan, and G.-Y. Niu, 2022: Global evaluation of the Noah-MP land surface model and suggestions for selecting parameterization schemes. J. Geophys. Res. Atmos., 127, e2021JD035753, https://doi.org/10.1029/2021JD035753.
Li, W., and Coauthors, 2023: Contrasting drought propagation into the terrestrial water cycle between dry and wet regions. Earth’s Future, 11, e2022EF003441, https://doi.org/10.1029/2022EF003441.
Lisonbee, J., M. Woloszyn, and M. Skumanich, 2022: Making sense of flash drought: Definitions, indicators, and where we go from here. J. Appl. Serv. Climatol., 2021 (1), 1–19, https://doi.org/10.46275/JOASC.2021.02.001.
Liu, Y., Y. Zhu, L. Ren, J. Otkin, E. D. Hunt, X. Yang, F. Yuan, and S. Jiang, 2020: Two different methods for flash drought identification: Comparison of their strengths and limitations. J. Hydrometeor., 21, 691–704, https://doi.org/10.1175/JHM-D-19-0088.1.
Lorenz, D. J., J. A. Otkin, M. Svoboda, C. R. Hain, and Y. Zhong, 2018: Forecasting rapid drought intensification using the Climate Forecast System (CFS). J. Geophys. Res. Atmos., 123, 8365–8373, https://doi.org/10.1029/2018JD028880.
McEvoy, D. J., J. L. Huntington, M. T. Hobbins, A. Wood, C. Morton, M. Anderson, and C. Hain, 2016: The evaporative demand drought index. Part II: CONUS-wide assessment against common drought indicators. J. Hydrometeor., 17, 1763–1779, https://doi.org/10.1175/JHM-D-15-0122.1.
Miralles, D. G., A. J. Teuling, C. C. van Heerwaarden, and J. Vilà-Guerau 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.
Mishra, V., S. Aadhar, and S. S. Mahto, 2021: Anthropogenic warming and intraseasonal summer monsoon variability amplify the risk of future flash droughts in India. npj Climate Atmos. Sci., 4, 1, https://doi.org/10.1038/s41612-020-00158-3.
Mitchell, K. E., and Coauthors, 2004: The multi-institution North American Land Data Assimilation System (NLDAS): Utilizing multiple GCIP products and partners in a continental distributed hydrological modeling system. J. Geophys. Res., 109, D07S90, https://doi.org/10.1029/2003JD003823.
Mo, K. C., and D. P. Lettenmaier, 2015: Heat wave flash droughts in decline. Geophys. Res. Lett., 42, 2823–2829, https://doi.org/10.1002/2015GL064018.
Mo, K. C., and D. P. Lettenmaier, 2016: Precipitation deficit flash droughts over the United States. J. Hydrometeor., 17, 1169–1184, https://doi.org/10.1175/JHM-D-15-0158.1.
Mocko, D. M., S. V. Kumar, C. D. Peters-Lidard, and S. Wang, 2021: Assimilation of vegetation conditions improves the representation of drought over agricultural areas. J. Hydrometeor., 22, 1085–1098, https://doi.org/10.1175/JHM-D-20-0065.1.
Mukherjee, S., and A. K. Mishra, 2022: A multivariate flash drought indicator for identifying global hotspots and associated climate controls. Geophys. Res. Lett., 49, e2021GL096804, https://doi.org/10.1029/2021GL096804.
Niu, G.-Y., and Coauthors, 2011: The community Noah land surface model with multiparameterization options (Noah-MP): 1. Model description and evaluation with local-scale measurements. J. Geophys. Res., 116, D12109, https://doi.org/10.1029/2010JD015139.
Osman, M., B. F. Zaitchik, H. S. Badr, J. I. Christian, T. Tadesse, J. A. Otkin, and M. C. Anderson, 2021: Flash drought onset over the contiguous United States: Sensitivity of inventories and trends to quantitative definitions. Hydrol. Earth Syst. Sci., 25, 565–581, https://doi.org/10.5194/hess-25-565-2021.
Otkin, J. A., M. C. Anderson, C. Hain, and M. Svoboda, 2014: Examining the relationship between drought development and rapid changes in the evaporative stress index. J. Hydrometeor., 15, 938–956, https://doi.org/10.1175/JHM-D-13-0110.1.
Otkin, J. A., M. Shafer, M. Svoboda, B. Wardlow, M. C. Anderson, C. Hain, and J. Basara, 2015: Facilitating the use of drought early warning information through interactions with agricultural stakeholders. Bull. Amer. Meteor. Soc., 96, 1073–1078, https://doi.org/10.1175/BAMS-D-14-00219.1.
Otkin, J. A., and Coauthors, 2016: Assessing the evolution of soil moisture and vegetation conditions during the 2012 United States flash drought. Agric. For. Meteor., 218–219, 230–242, https://doi.org/10.1016/j.agrformet.2015.12.065.
Otkin, J. A., M. Svoboda, E. D. Hunt, T. W. Ford, M. C. Anderson, C. Hain, and J. B. Basara, 2018: Flash droughts: A review and assessment of the challenges imposed by rapid-onset droughts in the United States. Bull. Amer. Meteor. Soc., 99, 911–919, https://doi.org/10.1175/BAMS-D-17-0149.1.
Otkin, J. A., Y. Zhong, E. D. Hunt, J. Basara, M. Svoboda, M. C. Anderson, and C. Hain, 2019: Assessing the evolution of soil moisture and vegetation conditions during a flash drought–flash recovery sequence over the south-central United States. J. Hydrometeor., 20, 549–562, https://doi.org/10.1175/JHM-D-18-0171.1.
Otkin, J. A., and Coauthors, 2022: Getting ahead of flash drought: From early warning to early action. Bull. Amer. Meteor. Soc., 103, E2188–E2202, https://doi.org/10.1175/BAMS-D-21-0288.1.
Parker, T., A. Gallant, M. Hobbins, and D. Hoffmann, 2021: Flash drought in Australia and its relationship to evaporative demand. Environ. Res. Lett., 16, 064033, https://doi.org/10.1088/1748-9326/abfe2c.
Patel, K. F., and Coauthors, 2021: Soil texture and environmental conditions influence the biogeochemical responses of soils to drought and flooding. Commun. Earth Environ, 2, 127, https://doi.org/10.1038/s43247-021-00198-4.
Pendergrass, A. G., and Coauthors, 2020: Flash droughts present a new challenge for subseasonal-to-seasonal prediction. Nat. Climate Change, 10, 191–199, https://doi.org/10.1038/s41558-020-0709-0.
Peters-Lidard, C. D., and Coauthors, 2007: High-performance Earth system modeling with NASA/GSFC’s land information system. Innov. Syst. Software Eng., 3, 157–165, https://doi.org/10.1007/s11334-007-0028-x.
Qing, Y., S. Wang, B. C. Ancell, and Z.-L. Yang, 2022: Accelerating flash droughts induced by the joint influence of soil moisture depletion and atmospheric aridity. Nat. Commun., 13, 1139, https://doi.org/10.1038/s41467-022-28752-4.
Reichle, R. H., D. B. McLaughlin, and D. Entekhabi, 2002a: Hydrologic data assimilation with the ensemble Kalman filter. Mon. Wea. Rev., 130, 103–114, https://doi.org/10.1175/1520-0493(2002)130<0103:HDAWTE>2.0.CO;2.
Reichle, R. H., J. P. Walker, R. D. Koster, and P. R. Houser, 2002b: Extended versus ensemble Kalman filtering for land data assimilation. J. Hydrometeor., 3, 728–740, https://doi.org/10.1175/1525-7541(2002)003<0728:EVEKFF>2.0.CO;2.
Sabater, J. M., C. Rüdiger, J.-C. Calvet, N. Fritz, L. Jarlan, and Y. Kerr, 2008: Joint assimilation of surface soil moisture and LAI observations into a land surface model. Agric. For. Meteor., 148, 1362–1373, https://doi.org/10.1016/j.agrformet.2008.04.003.
Schwartz, C., T. Haigh, M. Svoboda, and M. Goebel, 2023: How is flash drought understood?—Experts’ definitions and decision-makers’ perceptions. Wea. Climate Soc., 15, 909–924, https://doi.org/10.1175/WCAS-D-22-0142.1.
Sehgal, V., N. Gaur, and B. P. Mohanty, 2021: Global flash drought monitoring using surface soil moisture. Water Resour. Res., 57, e2021WR029901, https://doi.org/10.1029/2021WR029901.
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.
Sun, Y., R. Fu, R. Dickinson, J. Joiner, C. Frankenberg, L. Gu, Y. Xia, and N. Fernando, 2015: Drought onset mechanisms revealed by satellite solar-induced chlorophyll fluorescence: Insights from two contrasting extreme events. J. Geophys. Res. Biogeosci., 120, 2427–2440, https://doi.org/10.1002/2015JG003150.
Svoboda, M., and Coauthors, 2002: The drought monitor. Bull. Amer. Meteor. Soc., 83, 1181–1190, https://doi.org/10.1175/1520-0477-83.8.1181.
Walker, D. W., and A. F. Van Loon, 2023: Droughts are coming on faster. Science, 380, 130–132, https://doi.org/10.1126/science.adh3097.
Walker, D. W., and Coauthors, 2024: Flash drought typologies and societal impacts: A worldwide review of occurrence, nomenclature, and experiences of local populations. Wea. Climate Soc., 16, 3–28, https://doi.org/10.1175/WCAS-D-23-0015.1.
Xia, Y., and Coauthors, 2012: Continental-scale water and energy flux analysis and validation for the North American Land Data Assimilation System project phase 2 (NLDAS-2): 1. Intercomparison and application of model products. J. Geophys. Res., 117, D03109, https://doi.org/10.1029/2011JD016048.
Yang, Z.-L., and Coauthors, 2011: The community Noah land surface model with multiparameterization options (Noah-MP): 2. Evaluation over global river basins. J. Geophys. Res., 116, D12110, https://doi.org/10.1029/2010JD015140.
Yin, J., and Coauthors, 2023: Assimilation of blended satellite soil moisture data products to further improve Noah-MP model skills. J. Hydrol., 621, 129596, https://doi.org/10.1016/j.jhydrol.2023.129596.
Yuan, X., L. Wang, P. Wu, P. Ji, J. Sheffield, and M. Zhang, 2019: Anthropogenic shift towards higher risk of flash drought over China. Nat. Commun., 10, 4661, https://doi.org/10.1038/s41467-019-12692-7.
Yuan, X., Y. Wang, P. Ji, P. Wu, J. Sheffield, and J. A. Otkin, 2023: A global transition to flash droughts under climate change. Science, 380, 187–191, https://doi.org/10.1126/science.abn6301.
Zeng, Z., and Coauthors, 2023: Increased risk of flash droughts with raised concurrent hot and dry extremes under global warming. npj Climate Atmos. Sci., 6, 134, https://doi.org/10.1038/s41612-023-00468-2.
Zhao, M., G. A., Y. Liu, and A. G. Konings, 2022: Evapotranspiration frequently increases during droughts. Nat. Climate Change, 12, 1024–1030, https://doi.org/10.1038/s41558-022-01505-3.
Zhu, Z., and Coauthors, 2016: Greening of the Earth and its drivers. Nat. Climate Change, 6, 791–795, https://doi.org/10.1038/nclimate3004.