Abstract

How Great Plains climate will respond under global warming continues to be a key unresolved question. There has been, for instance, considerable speculation that the Great Plains is embarking upon a period of increasing drought frequency and intensity that will lead to a semipermanent Dust Bowl in the coming decades. This view draws on a single line of inference of how climate change may affect surface water balance based on sensitivity of the Palmer drought severity index (PDSI). A different view foresees a more modest climate change impact on Great Plains surface moisture balances. This draws on direct lines of analysis using land surface models to predict runoff and soil moisture, the results of which do not reveal an ominous fate for the Great Plains. The authors’ study presents a parallel diagnosis of projected changes in drought as inferred from PDSI and soil moisture indicators in order to understand causes for such a disparity and to shed light on the uncertainties. PDSI is shown to be an excellent proxy indicator for Great Plains soil moisture in the twentieth century; however, its suitability breaks down in the twenty-first century, with the PDSI severely overstating surface water imbalances and implied agricultural stresses. Several lines of evidence and physical considerations indicate that simplifying assumptions regarding temperature effects on water balances, especially concerning evapotranspiration in Palmer’s formulation, compromise its suitability as drought indicator in a warming climate. The authors conclude that projections of acute and chronic PDSI decline in the twenty-first century are likely an exaggerated indicator for future Great Plains drought severity.

1. Introduction

The future trajectory of climate change in the U.S. Great Plains (30°–50°N, 105°–95°W), particularly the frequency and intensity of future drought, is a matter of continuing interest and debate given its past history of severe drought episodes such as the 1930s “Dust Bowl.” While some have raised the specter of a shift to semipermanent 1930s type drought conditions on the Great Plains due to human-induced global warming, the special report of the Intergovernmental Panel on Climate Change regarding extreme events (Field et al. 2012) expresses only low confidence in a projected change in drought over the U.S. Great Plains as a whole and medium confidence for some increased dryness across the southern portion of the domain.

Why are there such substantial differences in expectations for drought in the Great Plains and why is there an overall low level of confidence in drought projections for this region as a whole? A possible explanation is an inconsistency in trends for surface water balances among analyses that use different drought indices. One approach has been to examine the sensitivity of the Palmer drought severity index (PDSI; Palmer 1965) to projected changes in monthly temperature and precipitation (e.g., Rind et al. 1990; Jones et al. 1996; Burke et al. 2006; Burke and Brown 2008; Dai 2011, 2012; Burke 2011; Wehner et al. 2011). Results from these studies, though derived from widely different generations of climate models, share the common feature that PDSI quickly approaches severe drought values with increasing greenhouse gas forcing.

Rind et al. (1990) suggested that increasing drought over the United States would become apparent in the 1990s, while Dai (2011) argues that the United States might see persistent droughts in the next 20–50 yr. The spatial pattern of U.S. drought increase was found to be highly reproducible among the 19 models contributing to phase 3 of the Coupled Model Intercomparison Project (CMIP3), largely because the response of PDSI to future temperature increase is very robust (Dai 2011, 2012; Wehner et al. 2011). Wehner et al. demonstrate that climate models having the largest projected temperature increase yield the largest increase in drought severity because of the strong dependency of evapotranspiration on surface air temperature in PDSI. Although this temperature sensitivity depends qualitatively on the formula used to compute potential evapotranspiration (e.g., Chen et al. 2005; Burke et al. 2006), Wehner et al. (2011) surmise that the rapid and severe decline of PDSI under the influence of future warming should be indicative of decreases in soil moisture, regardless of how mean precipitation may change. However, their supposition has not been confirmed by analyses of projected soil moisture changes simulated directly by climate model land surface schemes.

Alternatively, annual-mean soil moisture in CMIP3 models is projected to decline in some areas by the end of the twenty-first century, though not significantly so over the northern Great Plains (Sheffield and Wood 2008; Orlowsky and Seneviratne 2011). In some regions, such as the northern Great Plains, soil moisture does not change because projected increases in precipitation are offset by a warming-induced increase in evapotranspiration. Consistent with these changes, surface runoff projections do not exhibit appreciable change (cf. 2041–60 to 1901–70) for the Missouri and upper Mississippi River drainage basins in CMIP3 models (Milly et al. 2005). For the Great Plains as a whole, Sheffield and Wood’s (2008) diagnosis of simulated soil moisture reveals no statistically significant change in the frequency of short-term drought (4–6-month duration) for any emissions scenario (B1, A1B, and A2) through 2100, and the earliest year for detecting a statistically significant change in the frequency of long-term (>12 months) drought is about 2050. More recently, Winter and Eltahir (2012) analyzed regional climate model projections for the U.S. Midwest (an area slightly east of the Great Plains region considered herein) and found no reductions in future summer soil moisture, with increased evapotranspiration balanced by increased precipitation. Thus, questions remain on how Great Plains drought will respond to global warming. To better understand uncertainty sources, we present a side-by-side analysis of projected changes in drought using both a PDSI and a soil moisture metric.

2. Data and methods

a. Climate model simulations

We use historical simulations and climate projections based on the Community Climate System Model, version 4 (CCSM4; Gent et al. 2011). CCSM4 is one of numerous models that compose CMIP5. We analyze monthly output from an ensemble of three CCSM4 simulations forced with variations in greenhouse gases, aerosols, time-varying solar irradiance, and the radiative effects of volcanic activity for 1850–2005 (Taylor et al. 2012). Projections of climate conditions for 2006–2100 are based upon the representative concentration pathway 8.5 for individual greenhouse gases and aerosols (Moss et al. 2010).

The Community Land Model version 4 (CLM4) used in CCSM4 represents a significant advance over prior versions (Lawrence et al. 2011; Lawrence et al. 2012). Advances include improved canopy treatment, a plant functional type dependency on soil moisture stress function, improved surface hydrology and runoff, and improved representation of evapotranspiration that provides more realistic treatment of plant transpiration, evaporation from soils, and canopy evaporation. The land model includes a prognostic carbon–nitrogen cycle with time-varying vegetation phenology. Soil water is predicted from a multilayer model in which time variability in soil moisture is a function of infiltration, runoff, gradient diffusion, gravity, canopy transpiration through root extraction, and interactions with groundwater (Oleson et al. 2010). Over the U.S. Great Plains region, the land model resolves 10 soil layers to ~3 m, with five additional bedrock layers to a depth of ~35 m.

b. Drought indices

The monthly PDSI (Palmer 1965) is calculated from the CCSM4 archive of monthly temperature and precipitation. The potential evapotranspiration (PET) is calculated using the Thornthwaite (1948) formulation, which depends upon only temperature and latitude. Similar approaches were used in the recent studies of Dai (2011, 2012) and Wehner et al. (2011). Burke et al. (2006) and Dai (2011, 2012) also calculated PET using the Penman–Monteith (PM) equation (Shuttleworth 1993), which, while providing a more realistic estimate of the meteorological control on PET, continues to be strongly temperature driven (Fennessey and Kirshen 1994). A recognized limitation of PDSI as an index for drought over semiarid regions including portions of the Great Plains is its non-Gaussian distribution with a tendency to yield more severe negative PDSI values. We used the statistical normalization method of Wells et al. (2004) to recalibrate the PDSI, which has the effect of altering the statistical distribution of extreme wet and dry occurrences in some areas so as to be more consistent with the expected frequency of rare events, to allow for more accurate comparisons of the index across regions, and to facilitate intercomparison with other standardized drought indices. Here, the parameters for calculating the PDSI values and the subsequent calibrations are determined from the monthly model output of 1901–2000 for each of the three simulations separately, and the calibrated PDSI values are obtained for the entire 1850–2100 period for each run using the respective set of parameters.

The second drought index is derived from the CCSM4 monthly column integrated soil moisture for three model layers having a total of ~10-cm depth, normalized, also using the standard deviation of its annual mean during 1901–2000. Sensitivity analyses indicated that our results do not differ materially when using a shallower (1 cm) or deeper (1 m) soil layer.

3. Results

Individual PDSI traces averaged over the Great Plains (30°–50°N, 105°–95°W) for the period 1850–2100 for each of three model integrations (gray curves) and the three-run ensemble average (yellow curve) are shown in Fig. 1. Two significant features of these time series are (i) a progressive reduction in PDSI commencing in the late twentieth century and accelerating in the twenty-first century occurring in all simulations due to increasing greenhouse gas (GHG) forcing and (ii) a randomly occurring decadal-long drought event during the late nineteenth century in one simulation due to internal model variability.

Fig. 1.

Annual time series of PDSI averaged over the Great Plains (30°–50°N, 105°–95°W) for 1851–2100 (the calibration period is 1901–2000). Individual traces correspond to each of the three CCSM4 model integrations (gray curves) while the three-run ensemble average is shown by the yellow curve. The time series are smooth with a nine-point Gaussian filter.

Fig. 1.

Annual time series of PDSI averaged over the Great Plains (30°–50°N, 105°–95°W) for 1851–2100 (the calibration period is 1901–2000). Individual traces correspond to each of the three CCSM4 model integrations (gray curves) while the three-run ensemble average is shown by the yellow curve. The time series are smooth with a nine-point Gaussian filter.

The randomly occurring severe drought in the late-nineteenth-century simulation has many of the meteorological characteristics of severe drought over the Great Plains during the 1930s. Figure 2 (left) shows the spatial patterns of decadally averaged climate and land surface conditions, a period during which the simulated PDSI (1865–75) was consistently indicating drought. Negative index values of the PDSI span the Great Plains region from North Dakota to Texas (Fig. 2, top left), similar to the observed 1930s pattern (e.g., Hoerling et al. 2009). The pattern of soil moisture deficits (Fig. 2, second row, left) largely mimics the indications for drought provided by PDSI. Further, the spatial pattern and intensity of drought given by both indicators appear largely determined by the simulated precipitation deficits (Fig. 2, third row, left). Widespread reductions in rainfall with average annual deficits of 1 standardized departure span much of the Great Plains, similar to the magnitude of precipitation deficits observed during the decade of the Dust Bowl (e.g., Schubert et al. 2004; Hoerling et al. 2009). Figure 2 (bottom left) also indicates that the drought region experienced elevated surface temperatures, though generally less than +0.5°C. While such warmth is consistent with an overall inverse relationship between rainfall and temperature in CCSM4 (not shown) analogous to that occurring in observations (e.g., Madden and Williams 1978), neither the pattern of PDSI nor soil moisture indicators of this simulated drought event were determined strongly by the temperature conditions. Overall, this randomly occurring moisture deficiency over the Great Plains is well described by PDSI using only monthly temperature and precipitation.

Fig. 2.

Spatial patterns of decadally averaged land surface conditions for (top)–(bottom) PDSI, 10-cm soil moisture, precipitation, and temperature for (left) 1865–75 and (right) 2045–55 for the CCSM4 model integrations. Anomalies are computed from the 1901–2000 reference period. The 1865–75 results are based on a single model run, and the results for 2045–55 are based on the three-run ensemble average. The box shown in (bottom left) denotes the Great Plains domain over which spatial averages are computed for time series analysis in Figs. 1 and 3 and scatterplot relationships in Fig. 4.

Fig. 2.

Spatial patterns of decadally averaged land surface conditions for (top)–(bottom) PDSI, 10-cm soil moisture, precipitation, and temperature for (left) 1865–75 and (right) 2045–55 for the CCSM4 model integrations. Anomalies are computed from the 1901–2000 reference period. The 1865–75 results are based on a single model run, and the results for 2045–55 are based on the three-run ensemble average. The box shown in (bottom left) denotes the Great Plains domain over which spatial averages are computed for time series analysis in Figs. 1 and 3 and scatterplot relationships in Fig. 4.

In contrast, under projected future climate change, modeled soil moisture and PDSI respond very differently. Spatial maps of conditions by the mid-twenty-first century (Fig. 2, right) reveal a North American–wide pattern of much reduced PDSI (Fig. 2, top) implying that drought conditions exceed both the scale and intensity of that occurring during the Dust Bowl era. The PDSI pattern deviates radically from the spatial pattern of projected soil moisture departures (Fig. 2, second row). Whereas the soil moisture departures are more coherent with the simulated pattern of rainfall change (Fig. 2, third row), the PDSI values are more coherent with the simulated pattern of temperature change (Fig. 2, bottom).

In addition, the simulated time series of Great Plains meteorological and land surface conditions (Fig. 3) reveal that soil moisture exhibits a decline during the twenty-first century, but the magnitude of land surface drying is less than a standard deviation of typical interannual variability. By contrast, PDSI declines on order of 5–10 times its standard deviation by the latter decades of the twenty-first century. The comparatively modest amount of Great Plains soil moisture depletion in CCSM4 is consistent with the multimodel results based on the land surface models of CMIP3 (e.g., Sheffield and Wood 2008). This relatively small signal of soil moisture reduction is consistent with the limited detectability for a change in drought frequency over the Great Plains until the latter portions of the twenty-first century, as noted in other studies and assessments (Sheffield and Wood 2008; Orlowsky and Seneviratne 2011; Field et al. 2012). Nonetheless, such soil moisture deficits sustained over a long period could have appreciable consequences.

Fig. 3.

Standardized annual time series of Great Plains surface temperature (red), evapotranspiration (purple), precipitation (green), soil moisture (blue), and PDSI (yellow) as simulated in the ensemble of CCSM4 forced integrations. Also shown are the Palmer model–derived annual time series of evapotranspiration (yellow) and potential evapotranspiration (orange). The reference period for the standardization is 1901–2000, and each of the curves is smoothed with a nine-point Gaussian filter.

Fig. 3.

Standardized annual time series of Great Plains surface temperature (red), evapotranspiration (purple), precipitation (green), soil moisture (blue), and PDSI (yellow) as simulated in the ensemble of CCSM4 forced integrations. Also shown are the Palmer model–derived annual time series of evapotranspiration (yellow) and potential evapotranspiration (orange). The reference period for the standardization is 1901–2000, and each of the curves is smoothed with a nine-point Gaussian filter.

A physically based understanding for these dramatic differences in drought projections is provided by comparing Palmer and CCSM4 projected evapotranspiration (ET) changes. The ET in Palmer’s formulation (Fig. 3, yellow curve) initially increases rapidly with the projected warming, with the moisture loss determined largely by the increase in PET-driven demand, the latter being a close proxy for the magnitude of projected temperature rise (cf. red and orange curves in Fig. 3). Since ET’s upper bound is dictated by precipitation and available soil moisture, a decline in the rate of ET rise in Palmer’s model ensues in the latter half of the twenty-first century as the bucket model’s soil moisture becomes substantially depleted. At this point in Palmer’s projections, PDSI falls to unprecedented negative values. By contrast, ET in CCSM4 (Fig. 3, purple curve) does not increase nearly as much as ET in Palmer’s model. The overall modest positive ET departure in CCSM4 projections is consistent with its projected modest soil moisture decline, whereas higher-frequency variability of ET in CCSM4 is mostly driven by fluctuations in precipitation.

Finally, the simulations indicate that covariability of annual departures in Great Plains PDSI and soil moisture drastically changes in the warming twenty-first-century climate. Figure 4 compares the relationship during the twentieth century (blue circles) with that during the twenty-first century (red circles) based on diagnosis of all three CCSM4 simulations. PDSI explains about 40% of the interannual variations in soil moisture during the twentieth century but explains only about 20% of the variance during the twenty-first century. A −1 PDSI value (an indication for mild drought) corresponds to about a −1 standardized departure of soil moisture during the twentieth century; however, a −6 PDSI value (which is far beyond the range of the most extreme PDSI value during the calibration period) also corresponds to a −1 standardized departure of soil moisture during the twenty-first century. The magnitude of the PDSI during the projection period of a warming climate thus ceases to be a reliable indicator of land surface water deficiencies and implied drought severity, at least within the simulations of CCSM4.

Fig. 4.

Comparison of the relationship between annual 10-cm soil moisture and PDSI for the 1901–2000 period (blue circles; R = +0.61) with that of the 2001–2100 period (R = +0.48) based on the diagnosis of all three CCSM4 simulations.

Fig. 4.

Comparison of the relationship between annual 10-cm soil moisture and PDSI for the 1901–2000 period (blue circles; R = +0.61) with that of the 2001–2100 period (R = +0.48) based on the diagnosis of all three CCSM4 simulations.

4. Conclusions and discussion

Assessments of how climate change may affect the frequency and severity of drought need to consider various drought indicators (e.g., Field et al. 2012). Two such indicators, PDSI and the soil moisture anomalies, each presents very different views of how surface water balances may evolve over the Great Plains under global warming. To understand these differences, we analyzed both drought indicators using CCSM4 simulations. Our analysis reproduces disparate outlooks for Great Plains drought; projections using the PDSI suggest that the Great Plains will be in a semipermanent state of severe Dust Bowl–like drought in coming decades, whereas soil moisture projections reveal modest drying and comparatively low detectability for changes in Great Plains drought frequency. Our analyses illustrate that the PDSI exaggerates future drought severity over the Great Plains because of unrealistic sensitivity to the projected warming of surface temperatures.

Scale analysis by Hu and Willson (2000) shows the PDSI is about equally affected by temperature and precipitation anomalies having similar magnitude but that the temperature sensitivity during the twentieth century is generally masked by the strong negative correlation between temperature and rainfall variability. A GHG-forcing change in surface temperature without a corresponding change in rainfall leads to projected values of PDSI that are effectively proxies for the projected increases in temperature but are not suitable indicators for soil moisture. The simple parameterization of potential evapotranspiration using a Thornthwaite formula is not appreciably remedied using a more sophisticated Penman–Monteith formulation (e.g., Dai 2011, 2012). Estimates of evapotranspiration based solely on temperature have been regarded as problematic (e.g., Lockwood 1999; Burke et al. 2006; Burke 2011). Palmer’s formula for how PET relates to a specific location’s radiation climate, as inferred from temperature alone, becomes violated in the presence of appreciable warming and leads to unrealistic extreme drought indications (Lockwood 1999). Additionally, the sensitivity of the PET calculation is strongly affected by the choice of the calibration period (Karl 1986).

The PDSI was developed to define drought using a minimum number of widely observed variables and provides a good snapshot of observed drought conditions in a variety of climates. However, Dai’s (2011) review of factors affecting drought raised several limitations of PDSI under global warming. Dai noted that the Palmer formula overestimates potential evapotranspiration under a warming climate and “the fact that [the PDSI index] may not work for the twenty-first-century climate itself is a troubling sign.” While past performance has been useful in appreciating the value of the PDSI, it is no guarantee for future success, especially when applied to the nonstationary climate system currently unfolding.

Our result for severe drought indication over the Great Plains based on PDSI is entirely consistent with a host of prior PDSI-based drought projections but derived from an earlier generation of climate models (e.g., Dai 2011, Wehner et al. 2011). The similar results are due to the robustness in projected surface warming combined with the strong dependency of PDSI on temperature, rather than the soundness of Palmer’s formulation of the complex surface water balance. Palmer’s drought index avoids the complicating biological factors and the complex processes of soil–vegetation–atmosphere transfer, which are essential in representing how land surface moisture balances evolve in a warming climate. With the advent of model-based reanalyses and the need to examine the GHG-forced climate response of drought, it is clear that drought indices need to be developed that incorporate all relevant variables at the proper time resolution to consistently define drought conditions across the full range of climates.

An extensive comparison of the Thornthwaite method with pan-based estimates of evapotranspiration in China (Chen et al. 2005) illustrates severe deficiencies that are only somewhat alleviated by properly calibrated PM estimates. However, we stress the fact that, while the use of the PM formulation to estimate PET reduces the bias associated with the strong temperature dependence in Palmer’s formulation (see, e.g., Burke et al. 2006), the impact of other simplifications remain. For instance, the PM formulation lacks a land surface model with plant phenology from which transpiration is physically derived and still uses simple accounting assumptions of how ET balances either precipitation or PET in Palmer’s water balance calculations.

In the context of our study, it is important to consider the implications that a nonstationary, warming climate has for misinterpreting drought severity based on model projections of PDSI. The limitations are that 1) the absolute value of PDSI should be viewed as an unreliable measure of drought severity in a warming climate (and for these same reasons in an extreme heat wave event like over the Great Plains in 2012) and 2) inferences of possible amplifying effects of temperature alone (separate from precipitation effects) on drought severity derived from analysis of Palmer’s model should likewise be viewed as unreliable. For real-time drought monitoring, the first limitation could be mitigated by annually updating the calibration of the Palmer equations and their coefficients using the latest data. By updating to include the data through 2011–12, one can derive a more representative estimate of the “climatological water balance,” given the known appreciable variability and emergent trends; however, the problem in using PDSI for assessing future drought conditions in climate model projections remains.

So is a transition to semipermanent drought imminent for the Great Plains? While our results cannot rule out the possibility of a Dust Bowl–like period in the near future, the physical processes for such an occurrence are likely to be related to prolonged deficiencies in Great Plains precipitation, rather than to the local temperature effect of projected surface warming alone.

Acknowledgments

The authors thank the editor and three anonymous reviewers for their helpful reviews of the manuscript. NCAR provided the data for the CCSM4 simulations.

REFERENCES

REFERENCES
Burke
,
E. J.
,
2011
:
Understanding the sensitivity of different drought metrics to the drivers of drought under increased atmospheric CO2
.
J. Hydrometeor.
,
12
,
1378
1394
.
Burke
,
E. J.
, and
S. J.
Brown
,
2008
:
Evaluating uncertainties in the projection of future drought
.
J. Hydrometeor.
,
9
,
292
299
.
Burke
,
E. J.
,
S. J.
Brown
, and
N.
Christidis
,
2006
:
Modeling the recent evolution of global drought and projections for the twenty-first century with the Hadley Centre climate model
.
J. Hydrometeor.
,
7
,
1113
1125
.
Chen
,
D.
,
G.
Gao
,
C.-Y.
Xu
,
J.
Guo
, and
G.
Ren
,
2005
:
Comparison of the Thornthwaite method and pan data with the standard Penman-Monteith estimates of reference evapotranspiration in China
.
Climate Res.
,
28
,
123
132
.
Dai
,
A.
,
2011
:
Drought under global warming: A review
.
Wiley Interdiscip. Rev. Climate Change
,
2
,
45
65
,
doi:10.1002/wcc.81
.
Dai
,
A.
,
2012
:
Increasing drought under global warming in observations and models
.
Nat. Climate Change
,
2
,
doi:10.1038/nclimate1633
.
Fennessey
,
N. M.
, and
P. H.
Kirshen
,
1994
:
Evaporation and evapotranspiration under climate change in New England
.
J. Water Resour. Plann. Manage.
,
120
,
48
69
.
Field
,
C. B.
, and
Coauthors
,
2012
: Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. Cambridge University Press, 582 pp.
Gent
,
P. R.
, and
Coauthors
,
2011
:
The Community Climate System Model version 4
.
J. Climate
,
24
,
4973
4991
.
Hoerling
,
M.
,
X.-W.
Quan
, and
J.
Eischeid
,
2009
:
Distinct causes for two principal U.S. droughts of the 20th century
.
Geophys. Res. Lett.
,
36
,
L19708
,
doi:10.1029/2009GL039860
.
Hu
,
Q.
, and
G. D.
Willson
,
2000
:
Effects of temperature anomalies on the Palmer drought severity index in the central United States
.
Int. J. Climatol.
,
20
,
1899
1911
.
Jones
,
P.
,
M.
Hulme
,
K.
Briffa
, and
C.
Jones
,
1996
:
Summer moisture availability over Europe in the Hadley Centre general circulation model based on the Palmer drought severity index
.
Int. J. Climatol.
,
16
,
155
172
.
Karl
,
T.
,
1986
:
The sensitivity of the Palmer drought severity index and Palmer’s Z-index to their calibration coefficients including potential evapotranspiration
.
J. Climate Appl. Meteor.
,
25
,
77
86
.
Lawrence
,
D. M.
, and
Coauthors
,
2011
:
Parameterization improvements and functional and structural advances in version 4 of the Community Land Model
.
J. Adv. Model. Earth Syst.
,
3
,
M03001
,
doi:10.1029/2011MS000045
.
Lawrence
,
D. M.
,
K. W.
Oleson
,
M. G.
Flanner
,
C. G.
Fletcher
,
P. J.
Lawrence
,
S.
Levis
,
S. C.
Swenson
, and
G. B.
Bonan
,
2012
:
The CCSM4 land simulation, 1850–2005: Assessment of surface climate and new capabilities
.
J. Climate
,
25
,
2240
2260
.
Lockwood
,
J. G.
,
1999
:
Is potential evapotranspiration and its relationship with actual evapotranspiration sensitive to elevated atmospheric CO2 levels?
Climatic Change
,
41
,
193
212
.
Madden
,
R.
, and
J.
Williams
,
1978
:
The correlation between temperature and precipitation in the United States and Europe
.
Mon. Wea. Rev.
,
106
,
142
147
.
Milly
,
P.
,
K.
Dunne
, and
A.
Vecchia
,
2005
:
Global pattern of trends in streamflow and water availability in a changing climate
.
Nature
,
438
,
347
350
.
Moss
,
R. H.
, and
Coauthors
,
2010
:
The next generation of scenarios for climate change research and assessment
.
Nature
,
463
,
747
756
.
Oleson
, and
Coauthors
,
2010
: Technical description of version 4.0 of the Community Land Model (CLM). NCAR Tech. Note TN-478+STR, 238 pp.
Orlowsky
,
B.
, and
S.
Seneviratne
,
2011
:
Global changes in extreme events: Regional and seasonal dimensions
.
Climatic Change
,
110
,
669
696
,
doi:10.1007/s10584-011-0122-9
.
Palmer
,
W. C.
,
1965
: Meteorological drought. U.S. Department of Commerce Research Paper 45, 58 pp.
Rind
,
D.
,
R.
Goldberg
,
J.
Hansen
,
C.
Rosenzweig
, and
R.
Ruedy
,
1990
:
Potential evapotranspiration and the likelihood of future drought
.
J. Geophys. Res.
,
95
(D7
9983
10 004
.
Schubert
,
S. D.
,
M. J.
Suarez
,
P. J.
Pegion
,
R. D.
Koster
, and
J. T.
Bacmeister
,
2004
:
On the cause of the 1930s dust bowl
.
Science
,
303
,
1855
1859
,
doi:10.1126/science.1095048
.
Sheffield
,
J.
, and
E.
Wood
,
2008
:
Projected changes in drought occurrence under future global warming from multi-model, multi-scenario, IPCC AR4 simulations
.
Climate Dyn.
,
31
,
79
105
.
Shuttleworth
,
W.
,
1993
: Evaporation. Handbook of Hydrology, D. R. Maidment, Ed., McGraw-Hill, 4.1–4.53.
Taylor
,
K. E.
,
R. J.
Stouffer
, and
G. A.
Meehl
,
2012
:
An overview of CMIP5 and the experiment design
.
Bull. Amer. Meteor. Soc.
,
93
,
485
498
.
Thornthwaite
,
C.
,
1948
:
An approach toward a rational classification of climate
.
Geogr. Rev.
,
38
,
55
94
.
Wehner
,
M.
,
D.
Easterling
,
J.
Lawrimore
,
R.
Heim
,
R.
Vose
, and
B.
Santer
,
2011
:
Projections of future drought in the continental Unites States and Mexico
.
J. Hydrometeor.
,
12
,
1359
1377
.
Wells
,
N.
,
S.
Goddard
, and
M.
Hayes
,
2004
:
A self-calibrating Palmer drought severity index
.
J. Climate
,
17
,
2335
2351
.
Winter
,
J. M.
, and
E. A. B.
Eltahir
,
2012
:
Modeling the hydroclimatology of the midwestern United States. Part 2: Future climate
.
Climate Dyn.
,
38
,
595
611
,
doi:10.1007/s00382-011-1183-1
.