Abstract

Simulations from 18 coupled atmosphere–ocean GCMs are analyzed to predict changes in the climatological Great Plains low-level jet (GPLLJ) and Midwest U.S. hydrology resulting from greenhouse gas increases during the twenty-first century. To build confidence in the prediction, models are selected for analysis based on their twentieth-century simulations, and their simulations of the future are diagnosed to ensure that the response is reasonable. Confidence in the model projections is also bolstered by agreement among models, in a so-called multimodel ensemble, and by analogy with present-day interannual variability.

The GCMs agree that the GPLLJ will be more intense in April, May, and June in the future. The selected models even agree on the reason for this intensification, namely, a westward extension and strengthening of the North Atlantic subtropical high (the Bermuda high) that occurs when greenhouse gas–induced warming over the continental United States exceeds that of the subtropical Atlantic in the spring. Accompanying the changes in the GPLLJ are springtime precipitation increases of 20%–40% in the upper Mississippi Valley, which are closely associated with intensified meridional moisture convergence by the jet, with decreases to the south, which results in reduced moist static stability in the region. The simulated differences in the Midwest circulation and hydrology in the spring for the twenty-first century are similar to the observed moisture balance and circulation anomalies for May and, especially, June of 1993, a year of devastating floods throughout the Mississippi Valley.

1. Introduction

Summertime convective precipitation over the central United States supports ecosystems and agriculture, and variations can have significant impacts on the environment, the economy, and people’s lives. Recent examples are the floods of 1993 and 2007, and the 1988 summer drought, but it is clear from longer-term records that the region also exhibits variability on many time scales. The Dust Bowl climate of the 1930s combined with an economic depression and forced a massive westward migration, and persistent warm, dry conditions reestablished in the 1950s, and again in the late 1980s. On longer time scales, tree ring data analysis reveals the occurrence of multidecadal megadroughts in past centuries (Gray et al. 2003; Stahle et al. 2007). This history of variability, along with the ever-growing need to manage water resources effectively, suggests a vulnerability to possible future changes in the region’s hydrology.

One prominent feature of the atmospheric dynamics related to summer precipitation distributions is the low-level southerly flow from the Gulf of Mexico. This flow is commonly referred to as “the” Great Plains low-level jet (GPLLJ) when seen in the climatology, but it is known to be composed of discrete, nocturnal, low-level jet (LLJ) events. Here we investigate potential changes in this flow and the associated Midwest hydrology in the twenty-first century resulting from increases in greenhouse gases. Output from 18 coupled ocean–atmosphere GCMs is examined. These models were run to contribute to the Fourth Assessment Report (AR4) of the Intergovernmental Panel on Climate Change (IPCC), and are made available to the community through the Program for Climate Model Diagnostics and Intercomparison (PCMDI).

A dynamical analysis, as opposed to a statistical analysis, is used to assess the reliability of the GCM simulations and produce a defensible prediction for the end of the twenty-first century. First, we select models that are best able to capture the present-day climate of the central United States, especially the low-level flow and its associated dynamics. Second, simulations of the future are diagnosed to expose the reasons for the simulated response to allow for judgment about how reasonable the response may be. Third, we evaluate the degree to which the models’ projections agree with each other because agreement among simulations suggests that processes that are modeled with less certainty and, therefore, by various methods are not dominating the response. Finally, we search for analogies in present-day climate variations. If a future climate state resembles, for example, a particular observed anomaly, then we are assured that such a state is physically possible in the real world.

Section 2 provides background on our current understanding, and section 3 discusses the model output and reanalysis products used in the analysis. Simulations from the coupled GCMs are evaluated in section 4, and we select GCMs that produce reasonable simulations for further investigation of changes in the twenty-first century for section 5. In section 6, the simulation and prediction of Midwest hydrology in the models is investigated for the twentieth- and twenty-first-century simulations. An analogy with present-day variability is presented in section 7, which also summarizes the conclusions.

2. Background

Meridional jets are common features of the summer climatology over the central United States. These GPLLJs generally form during the night and early morning (e.g., Mitchell et al. 1995; Hoxit 1975), and occur in 50%–60% of the warm-season nights (Song et al. 2005). Walters (2001) distinguishes 12 GPLLJ classifications and, while there are some similarities among the various jet types (e.g., a collocated humidity gradient and a thermal ridge southwest of the jet axis), other features vary considerably. Walters and Winkler (2001) find that about 37% of the jet cases can be classified as anticyclonic curving jets.

Because they develop so frequently, these LLJs contribute significantly to the monthly mean climatological flow. Figure 1 shows the monthly mean 850-hPa wind and geopotential height climatology over the central United States from the North American Regional Reanalysis (NARR; Mesinger et al. 2006) for April through September, averaged from 1979 to 1999. At its summertime peak in June and July, the low-level flow is southerly–southeasterly along the gulf coast, with strongest onshore winds between 90° and 100°W. The flow retains its southerly direction deep into the continental interior in the western portion of its domain (west of about 98°W). To the east of 98°W, the winds turn eastward (anticyclonic) and become zonal north of about 33°N. Recognizing the connection with frequent jet events, this low-level flow is often referred to as the “climatological GPLLJ” or, simply, the GPLLJ.

Fig. 1.

Monthly mean 850-hPa geopotential heights (gpm) and winds (m s−1) from the NARR, averaged from 1979 to 1999. Vector scale is indicated in m s−1, and contour interval is 10 gpm.

Fig. 1.

Monthly mean 850-hPa geopotential heights (gpm) and winds (m s−1) from the NARR, averaged from 1979 to 1999. Vector scale is indicated in m s−1, and contour interval is 10 gpm.

North of about 35°N, the flow is aligned perpendicular to the geopotential height gradient that forms between the eastern extension of the North Atlantic subtropical high (the Bermuda high), across Florida and the eastern Gulf of Mexico, and the broad continental low positioned over the Rockies. Magnitudes are supergeostrophic (not shown). South of 35°N, significant flow across geopotential height contours is evident, particularly over Texas and northern Mexico.

Between April and June, horizontal geopotential height gradients over the central United States transition from being primarily meridional (Fig. 1a) to primarily zonal (Figs. 1b,c). During the spring months, low heights over the Rockies weaken slightly, while the higher geopotential heights over the southeastern United Sates strengthen significantly. Low geopotential heights over the western United States remain fairly constant through the summer, and the weakening of the jet in August and September is primarily associated with a weakening of the subtropical high.

When the jet forms in April, maximum meridional wind velocities are centered near 910 hPa and 98°W, with maximum meridional velocities of nearly 4 m s−1 in the monthly mean. As the peak winds strengthen through the spring, reaching over 6 m s−1 in June and July in the southern plains, the jet core rises to about 860 hPa. Figure 2a displays the vertical cross section of the June mean meridional wind averaged from 30° to 38°N (to capture the meridional component of the flow across the southern plains) from the NARR. The jet is zonally asymmetric, with stronger zonal wind shears in the west along the mountains than in the east. Below the core, the jet tilts to the west with increasing altitude. Above and to the east of the jet, the flow is northerly.

Fig. 2.

Cross section of the June meridional wind speed (m s−1) from the (a) NARR and eight AR4 coupled GCMs (see Table 1), averaged over 30°–38°N and 1979–99. The contour interval is 1 m s−1, and white areas denote the presence of topography.

Fig. 2.

Cross section of the June meridional wind speed (m s−1) from the (a) NARR and eight AR4 coupled GCMs (see Table 1), averaged over 30°–38°N and 1979–99. The contour interval is 1 m s−1, and white areas denote the presence of topography.

The inertial oscillation is thought to be relevant for understanding the diurnal cycle of LLJs over the Great Plains as well as the supergeostrophic velocities of the climatological GPLLJ, because wind speeds accelerate when nighttime radiative cooling of the atmosphere very close to the surface creates a stable environment and nearly frictionless conditions aloft (e.g., Blackadar 1957; Hoxit 1975; Stensrud 1996).

Topography is also a factor in LLJ formation over the Great Plains. Holton (1967) suggested that the climatological GPLLJ is a geostrophic feature that arises (by the thermal wind relation) as a result of positive zonal surface temperature gradients that develop along the westward-sloping Great Plains topography, with higher geopotential heights to the east associated with the North Atlantic subtropical high and its westward extension, the Bermuda high. This idea does not explain the supergeostrophic velocity, but it does place the jet over the eastern foothills of the Rockies. Subsequent studies further develop the view that topography is relevant to the LLJ formation. Uccellini (1980) shows the importance of leeside cyclogenesis and troughing on the eastern slopes of the Rocky Mountains for the production of lower-tropospheric pressure gradients needed for the development of the LLJs over the central United States. The mountains also generate thermal forcing resulting from the elevation of the topography, though this forcing may be less important than the mechanical forcing that results when the westerly upper-level jet impinges on the Rockies, even in the summer months (e.g., Ringler and Cook 1995, 1999). Transient thermal and vorticity forcing resulting from the modification of transient eddy activity in the presence of topography may also be influential, especially in extending the LLJ farther northward and eastward (Ting and Wang 2006).

The southerly low-level flow has been identified as an important moisture source for convection and the development of MCCs (e.g., Tuttle and Davis 2006), and the LLJs are relevant to the development of dynamic instability through cyclonic shear, convergence, and ascent (Mitchell et al. 1995). Helfand and Schubert (1995) estimate that up to ⅓ of the moisture transported into the North America continent is delivered by this low-level flow. Ruiz-Barradas and Nigam (2005) find that variations in moisture converged within the atmospheric column are very important on interannual time scales, but on seasonal time scales evaporation exceeds the column moisture convergence (Nigam and Ruiz-Barradas 2006).

Intensification of the southerly low-level flow, along with more frequent and intense GPLLJ formation, has been associated with extreme hydrological events over the central United States. Studies of the 1993 floods in the Midwest show an association with a sustained period of intense jet formation (Arritt et al. 1997). Bell and Janowiak (1995) suggest that flooding conditions develop and persist in association with three circulation features, and one of these is a persistent wave pattern with strong southwesterly flow that sustains southerly moisture transport into the central United States.

The 1988 drought was associated with a northward displacement of the westerly subtropical jet stream in association with an anomalous ridge of high pressure over the northern Great Plains and a significant weakening of the moisture transport from the Gulf of Mexico (Trenberth and Guillemot 1996). Dirmeyer and Brubaker (1999) estimate that 41% of the rainfall that fell over the Mississippi basin during the summer of 1988 originated as evaporation from the same basin, while only 33% of the precipitation was recycled during the flood year of 1993.

Apart from such extreme events, the relationship between the low-level flow and midwestern precipitation is less clear, with a muddled cause and effect. Some authors (e.g., McCorcle 1988) find strong GPLLJ activity during dry years, and others (e.g., Sun et al. 2004) find that weaker jets are more typical. Ting and Wang (2006) report a close association between the overall interannual variability of rainfall over the continental United States and the strength of the low-level flow.

Atmospheric GCMs (e.g., Byerle and Paegle 2003; Sud et al. 2003; Ting and Wang 2006) and regional climate models (RCMs; e.g., Bosilovich and Sun 1999; Mo and Juang 2003; Sun et al. 2004; Miguez-Macho et al. 2005) have been used extensively to study the hydrodynamics of the central United States. Anderson et al. (2003) evaluated 13 RCMs simulating the floods of 1993 and found that all produce positive precipitation minus evaporation (P − E) values over the upper Mississippi basin, though most were weaker than in the observations because of low precipitation. All of the models had some difficulty simulating the diurnal cycle of precipitation. Ruiz-Barradas and Nigam (2006) find that atmospheric GCMs frequently overestimate water recycling and, therefore, underestimate the role of the low-level flow in converging moisture over the Great Plains.

3. Methodology: Reanalyses and the evaluation of GCM output

The analyzed model output is monthly mean values from coupled ocean–atmosphere GCM integrations performed for IPCC AR4. Output from 18 different GCMs from various groups worldwide were examined, as listed and named in Table 1.

Table 1.

Coupled GCM integrations examined. Naming conventions are from IPCC (to encourage uniformity across analyses of the IPCC integrations, these official names for each simulation are suggested online at http://www-pcmdi.llnl.gov/ipcc/model_documentation/ipcc_model_documentation.php).

Coupled GCM integrations examined. Naming conventions are from IPCC (to encourage uniformity across analyses of the IPCC integrations, these official names for each simulation are suggested online at http://www-pcmdi.llnl.gov/ipcc/model_documentation/ipcc_model_documentation.php).
Coupled GCM integrations examined. Naming conventions are from IPCC (to encourage uniformity across analyses of the IPCC integrations, these official names for each simulation are suggested online at http://www-pcmdi.llnl.gov/ipcc/model_documentation/ipcc_model_documentation.php).

At issue for examining the low-level flow over the continental United States is the treatment of “data in rock,” for example, on pressure levels into which the Rocky Mountain topography protrudes. Some of the GCM investigators provided output to the AR4 archive with variables interpolated into rock and others did not, leaving undefined values when a pressure level goes underground. Here we take the output as provided by the investigators, so some model output is presented with missing values and some is not.

Two reanalyses are used as a standard of comparison, namely, the NARR and the National Centers for Environmental Prediction (NCEP)/Department of Energy Global Reanalysis 2 (NCEP2; Kanamitsu et al. 2002). The NARR is at a higher resolution than NCEP2, at 32 km compared with 2.5° × 2.5°. The NARR also assimilates precipitation, while NCEP2 does not, so only NARR precipitation is used here. However, in some ways comparing the circulation from the coupled GCMs with NCEP2 is a fairer comparison, because the resolution is similar.

Normally one would interpolate all of the fields to be compared onto the same grid, but here we keep all of the displays of reanalysis and model output on their native grids to avoid losing information in the intermodel comparison, and to provide perspective on the extent to which resolution may influence the proper simulation of the jet.

In projections of future climate, we chose to examine results for the last two decades of the twenty-first century under the IPCC’s A2 scenario for future emissions of CO2 and other greenhouse gases. This is a business-as-usual scenario that predicts changes in the absence of effective control on emissions in the twenty-first century.

The analysis of the model output emphasizes an evaluation of the dynamics of the jet, and its consequences for moisture fields, for the relatively large space scales that are resolved well by the current generation of GCMs. This approach uses the predictions of future atmospheric flow and large-scale moist stability characteristics of the atmosphere, which are simulated well, to better understand precipitation, which is much more difficult to simulate. Because this method requires that we work with a subset of the 18 GCMs, we choose the ones that most closely represent the observations in the central United States in spring and summer.

4. Representation of the present-day GPLLJ dynamics in the current generation of coupled GCMs

All of the models (Table 1) produce a climatological GPLLJ, but with various degrees of accuracy. Cross sections of the June meridional wind averaged from 1979 through 1999 and over 30°–38°N from some of the GCMs listed in Table 1 are displayed in Figs. 2b–i. The models chosen for display and further analysis here are those that produce more realistic representations of the low-level meridional flow during June and July. The Goddard Institute for Space Studies Atmosphere–Ocean Model (GISS-AOM), GISS Model E-R (GISS-ER), and GISS Model E-H (GISS-EH) are not chosen because their representations of the GPLLJ are quite weak, reaching only about 2 m s−1 in June and July, for example. Similarly, the 3 m s−1 jet of the Institute of Atmospheric Physics (IAP) Flexible Global Ocean–Atmosphere–Land System Model gridpoint version 1.0 (FGOALS-g1.0) simulation, and the 4 m s−1 jet maxima of the Institute for Numerical Mathematics (INM) Coupled Model, version 3.0 (INM-CM3.0) simulation in June and the L’Institut Pierre-Simon Laplace (IPSL) Coupled Model, version 4 (CM4) simulation in both June and July cause us to eliminate those models from further analysis at this point. The Geophysical Fluid Dynamics Laboratory Climate Model version 2.1 (GFDL CM2.1) simulation is excluded because, while it is quite similar to the GFDL Climate Model version 2.0 (GFDL CM2.0) simulation for June with a jet maximum over 5 m s−1, the wind speed drops to 4 m s−1 in July, which is 50% below both the observed jet speed and the accurate jet speed of the GFDL CM2.0 simulation for July. The Canadian Centre for Climate Modelling and Analysis (CCCma) Coupled General Circulation Model, version 3.1 (CGCM3.1) simulation produces a jet that is quite strong, up to 9 m s−1 in July, so it is eliminated from the study at this point because a number of other models produce more realistic wind speeds. Finally, we chose not to include the Model for Interdisciplinary Research on Climate 3.2, high-resolution version [MIROC3.2(hires)] and Commonwealth Scientific and Industrial Research Organisation Mark version 3.0 (CSIRO Mk3.0) for further analysis because the jet maxima in their simulations is right at the surface, at 930 hPa at 97°W, and the CSIRO jet is also somewhat weak (barely 5 m s−1) and quite broad. (Note that while we justify the model selection here with reference to the cross section of the meridional wind, we also examined horizontal wind vectors and geopotential heights at the level of the simulated jet core to ensure that the selected models are also reasonable in this respect.)

Most of the models capture the climatological jet’s longitudinal position fairly well, although some place the core somewhat too far east [e.g., ECHAM5/Max Planck Institute Ocean Model (MPI-OM), IAP FGOALS-g1.0, GISS-AOM]. Some models, such as third climate configuration of the Met Office Unified Model (HadCM3), the Centre National de Recherches Météorologiques Coupled Global Climate Model, version 3 (CNRM-CM3), and GFDL CM2.0, capture the westward tilt with increasing elevation below the core. However, in every case the jet is too wide. This is possibly a resolution issue, because the models with a higher resolution tend to produce a more narrow jet, and because the jet in NCEP2 (not shown) is wider than the jet in the NARR.

A horizontal view of the low-level flow in June in these GCM simulations is provided in Fig. 3, for comparison with the June representation in the NARR (Fig. 1c). High geopotential heights over the southeastern United States are somewhat too strong in some of the simulations, most notably the National Center for Atmospheric Research (NCAR) Community Climate System Model, version 3 (CCSM3). In other months (not shown), the high may be a little too weak in some of the simulations, or misplaced 3°–6° latitude too far north. Overall, these models capture the general continental-scale geopotential height pattern seen in the NARR during the months when the GPLLJ exists.

Fig. 3.

June 850-hPa geopotential heights (gpm) and winds from eight selected AR4 GCMs, averaged from 1979 to 1999. Vector scale (m s−1), and the contour interval is 10 gpm.

Fig. 3.

June 850-hPa geopotential heights (gpm) and winds from eight selected AR4 GCMs, averaged from 1979 to 1999. Vector scale (m s−1), and the contour interval is 10 gpm.

The ability of the models to capture the seasonal cycle of the GPLLJ is also evaluated. Monthly mean geopotential heights and winds averaged over 1979–99 at 850 hPa from April through September from the models shown in Fig. 2 were compared with the NARR geopotential heights and winds shown in Fig. 1. For example, we find that CNRM-CM3 and ECHAM5–MPI-OM reproduce accurate seasonal cycles. In HadCM3 and GFDL CM2.0, the subtropical high is somewhat weak throughout the 6-month period examined, but the transitions are timed appropriately. In PCM, the subtropical high forms 1 month early, remains strong from May through July, and then quickly diminishes in August, breaking down the jet structure about 1 month early.

To provide an overview of the simulations of the jet’s seasonality and to establish a standard for comparing differences and changes, monthly mean 850-hPa meridional wind speeds in the area bounded by 30°–38°N, 102°–92°W are displayed in Figs. 4a,b for NARR, NCEP2, and the eight simulations displayed in Fig. 3. The averaging area chosen is appropriate for the GCMs, with their too-wide representation of the meridional flow, to support the analysis below that quantifies future changes in the jet in the GCMs. Compared to NARR, the GCMs appear to simulate a stronger jet, but this is an artifact of the averaging area chosen. Note that NCEP2, which has a horizontal resolution similar to the GCMs and a weaker jet, also seems stronger than the NARR with this averaging.

Fig. 4.

(a), (b) Meridional wind averaged over the southern Great Plains (30°–38°N, 102°–92°W) for 1979–99 in 11 coupled GCM simulations, NARR, and NCEP2. (c), (d) Differences in the meridional wind averaged over the southern Great Plains (30°–38°N, 102°–92°W) for 2079–99 minus 1979–99.

Fig. 4.

(a), (b) Meridional wind averaged over the southern Great Plains (30°–38°N, 102°–92°W) for 1979–99 in 11 coupled GCM simulations, NARR, and NCEP2. (c), (d) Differences in the meridional wind averaged over the southern Great Plains (30°–38°N, 102°–92°W) for 2079–99 minus 1979–99.

Maximum values of the meridional wind within the averaging area occur in June in the NARR. The ECHAM5–MPI-OM, CNRM-CM3, Meteorological Research Institute Coupled General Circulation Model, version 2.3.2a (MRI CGCM2.3.2a), and HadCM3 simulations all capture the June maximum, but the jet in the MIROC3.2, medium-resolution version MIROC3.2 (medres) is strongest in August, with a secondary maximum in June. The GFDL CM2.0 (and GFDL CM2.1, which is not shown) and NCAR CCSM3 simulations have well-defined peaks in August. The NCAR Parallel Climate Model (PCM) simulation is very similar to NCEP2 in April through August but, like all of the models except MRI CGCM2.3.2a and ECHAM5/MPI-OM, it simulates northerly flow that is too weak, or even southerly flow, in the winter months. (We also examined additional ensemble members from HadCM3 and ECHAM5–MPI-OM and find that they are very similar to each other.)

5. Simulated twenty-first-century changes in the GPLLJ

Figures 4c,d show differences between the average meridional wind (30°–38°N, 102°–92°W) for 2079–99 under the A2 scenario forcing and the 1979–99 means plotted in Figs. 4a,b from eight GCM simulations. All of the simulations except that of NCAR PCM exhibit stronger wind speeds in April, and all of the simulations project an increase in the meridional wind speed in this region in May. There is also agreement among the models, with the exception of MRI CGCM2.3.2a, that the increased wind speed will be sustained through June.

In contrast to the springtime intensification of the jet, July, August, and September wind speed differences are minimal. In the fall, particularly October, there is a tendency for increased wind speeds, which suggests that the GPLLJ structure stays in place longer into the fall in the twenty-first-century simulations as compared with those of the twentieth century, but the differences are not as consistent or pronounced as in April–June (AMJ).

Differences between the 2079–99 and 1979–99 mean AMJ meridional wind velocity in six GCM simulations, averaged over 30°–38°N, are displayed in Fig. 5. (To create a more manageable set of models for the dynamical analysis, the NCAR CCSM3 and MIROC3.2 models are dropped from the analysis at this point. These models were chosen to be eliminated because the NCAR CCSM3 jet intensity is the most unrealistic in the remaining set of models, and MIROC3.2 has an unrealistic double peak in its seasonal cycle.) This figure confirms that the increases in AMJ meridional wind speed seen in Figs. 4c,d occur only in the LLJ region. It also shows that the longitudinal placement of the jet is not changed, and that the jet intensification is somewhat greater below the jet core. Differences in the ECHAM5–MPI-OM, CNRM-CM3, and GFDL CM2.0 simulations (Figs. 5a–c) are virtually identical, and the intensification of the GPLLJ in HadCM3 (Fig. 5d) is only slightly weaker. The weak response in the NCAR PCM (Fig. 5e) amounts to a slight jet strengthening along the flank of the Rockies. The MRI CGCM2.3.2a difference (Fig. 5f) is relatively small because the intensification does not persist through June (Fig. 4d), but the structure of the anomaly is similar to the other model results.

Fig. 5.

Cross sections of differences in the predicted (under the A2 forcing scenario for 2079–99) and present-day (1979–99) meridional wind speed (m s−1) for AMJ from six AR4 coupled GCMs (see Table 1) averaged over 30°–38°N. The contour interval is 0.5 m s−1. White areas indicate data in rock.

Fig. 5.

Cross sections of differences in the predicted (under the A2 forcing scenario for 2079–99) and present-day (1979–99) meridional wind speed (m s−1) for AMJ from six AR4 coupled GCMs (see Table 1) averaged over 30°–38°N. The contour interval is 0.5 m s−1. White areas indicate data in rock.

What change in the continental-scale dynamics accompanies these changes in the low-level flow? In the twenty-first-century simulations, geopotential heights at 850 hPa increase globally. Averaged over the continental United States and the North Atlantic at 850 hPa, these differences range between 12 and 20 gpm in the GCMs examined. However, there is a change in the structure of the geopotential height fields as well. Simulated differences in the AMJ 850-hPa geopotential height in the selected six GCMs under the A2 forcing scenario are shown in Fig. 6, with the average geopotential height increase over the domain in each model subtracted to emphasize the change in structure. Positive geopotential height anomalies cover the eastern United States in each simulation and, as a result, zonal geopotential height gradients are increased across the Midwest. Decreases in geopotential heights to the west, over the Rockies, also contribute to the enhanced zonal gradients, consistent with the idea that troughing over the eastern foothill of the Rockies contributes to the formation of the jet. (Recall, however, that many of the models are displaying extrapolated data in rock over the high topography.)

Fig. 6.

Differences in the simulated AMJ anomalous 850-hPa geopotential heights for 2079–99 minus 1979–99 from (a) CNRM-CM3, (b) GFDL CM2.0, (c) ECHAM5/MPI-OM, (d) MRI CGCM2.3.2a, (e) NCAR PCM, and (f) HadCM3. The contour interval is 3 gpm, and negative values are shaded. In each panel, white contours indicate the 1520- and 1560-gpm geopotential heights in the 1979–99 simulations. White areas indicate data in rock.

Fig. 6.

Differences in the simulated AMJ anomalous 850-hPa geopotential heights for 2079–99 minus 1979–99 from (a) CNRM-CM3, (b) GFDL CM2.0, (c) ECHAM5/MPI-OM, (d) MRI CGCM2.3.2a, (e) NCAR PCM, and (f) HadCM3. The contour interval is 3 gpm, and negative values are shaded. In each panel, white contours indicate the 1520- and 1560-gpm geopotential heights in the 1979–99 simulations. White areas indicate data in rock.

The modeled increases in geopotential heights over the eastern United States are part of a larger-scale response that is similar in each model. A distinct quadrupole anomaly develops in the CNRM-CM3, GFDL CM2.0, MRI, and HADCM simulations (Figs. 6a,b,d and f, respectively), with positive normalized geopotential height anomalies in the east and west, and negative anomalies to the north and south. The anomaly pattern is similar both in the ECHAM5–MPI-OM simulation (Fig. 6c), but with a less distinct separation between the lows, and in the NCAR PCM simulation (Fig. 6e), but with only a weak positive anomaly over the eastern United States. [In the NCAR PCM simulation, the quadrupole anomaly emerges more clearly in May and June, consistent with this model’s jet intensification in these months (Fig. 4d), but not in April.]

The thick white contours (1520 and 1560 gpm) in Fig. 6 indicate the location and shape of the North Atlantic subtropical high and its westward extension, the Bermuda high, in the twentieth-century simulations. The patterns in Fig. 6 that place positive anomalies over the eastern United States are then seen to be due to an overall strengthening of the North Atlantic subtropical high and its westward extension in the GCMs, that is, an enhancement of the Bermuda high. The center of the North Atlantic subtropical high shifts to the west by about 5° longitude in the simulations, and also strengthens.

The simulated differences in the GPLLJ (Fig. 4 and 5) are primarily a geostrophic response to the geopotential height anomalies shown in Fig. 6. As an example, Fig. 7a shows differences in 850-hPa meridional wind speeds as simulated by the CNRM-CM3, and Fig. 7b shows the geostrophic wind difference calculated from the geopotential height anomalies shown in Fig. 6a. Figures 7c,d are the full and geostrophic wind fields, respectively, for the GFDL CM2.0 simulations. Over the central and southern Great Plains in both models the meridional wind anomaly is supergeostrophic, as is the case for the observed jet (Means 1952; Blackadar 1957). However, with only about 0.5 m s−1 difference between the geostrophic and actual wind speeds, the jet intensification is clearly related to the intensification of the zonal geopotential height gradients shown in Fig. 6. The response in the other GCMs examined is similar, agreeing with the CNRM-CM3 and GFDL CM2.0 simulations that the enhancement of the GPLLJ over the midwestern United States in the twenty-first century is related to a strengthening and westward extension of the North Atlantic subtropical high in spring (AMJ). This mechanism for jet intensification is related to Wexler’s (1961) analogy that the GPLLJ is similar to a western boundary current in the ocean, a consequence of the westward extension of the Bermuda high against blocking by the Rockies.

Fig. 7.

Differences in the AMJ 850-hPa (a), (c) full meridional wind and (b), (d) geostrophic meridional wind for 2079–99 minus 1979–99 as simulated by CNRM-CM3 and GFDL CM2.0, respectively. The contour interval is 0.5 m s−1, and white areas denote data in rock.

Fig. 7.

Differences in the AMJ 850-hPa (a), (c) full meridional wind and (b), (d) geostrophic meridional wind for 2079–99 minus 1979–99 as simulated by CNRM-CM3 and GFDL CM2.0, respectively. The contour interval is 0.5 m s−1, and white areas denote data in rock.

The simulated changes in AMJ low-level geopotential heights (Fig. 6) and, as a consequence, the climatological GPLLJ (Figs. 4 and 5), can be related to changes in low-level temperature. Figure 8 displays differences in surface temperature for the six model simulations, arranged as in Fig. 6. Over the North Atlantic, each model simulates either cooling or, as in the case of ECHAM5–MPI-OM and the flux-corrected MRI CGCM2.3.2a simulation, negligible warming north of about 40°N and warming to the south. This pattern is consistent with a weakening of the meridional overturning circulation in the Atlantic, which has been reported in many coupled GCMs in response to greenhouse forcing (e.g., Gregory et al. 2005; Schneider et al. 2007).

Fig. 8.

Differences in AMJ surface temperature (K) for 2079–99 minus 1979–99 as simulated by (a) CNRM-CM3, (b) GFDL CM2.0, (c) ECHAM5/MPI-OM, (d) MRI CGCM2.3.2a, (e) NCAR PCM, and (f) HadCM3. The contour interval is 1 K. Negative values are unshaded, and values >+4 K are darkly shaded.

Fig. 8.

Differences in AMJ surface temperature (K) for 2079–99 minus 1979–99 as simulated by (a) CNRM-CM3, (b) GFDL CM2.0, (c) ECHAM5/MPI-OM, (d) MRI CGCM2.3.2a, (e) NCAR PCM, and (f) HadCM3. The contour interval is 1 K. Negative values are unshaded, and values >+4 K are darkly shaded.

In each simulation displayed in Fig. 8 there is stronger warming over the continental United States than over the subtropical and midlatitude Atlantic Ocean to the east. These differences in zonal temperature gradients, and not the differences in meridional temperature gradients described above, are related to the enhancement of the North Atlantic subtropical high. While one would expect a weakening of the subtropical high in association with SST increases in the subtropical Atlantic in isolation, the warming is greater over the land surface to the west than it is over the subtropical Atlantic. This differential heating results in a shift in atmospheric mass off the continent and an increase in surface pressures over the western Atlantic.

The meridional geopotential height anomalies simulated for 2079–99 are also similar to the positive phase of the North Atlantic Oscillation (NAO), with anomalously low geopotential heights over the North Atlantic and high heights in the subtropics. Weaver and Nigam (2008) find that this phase of the NAO, which occurs in summer as well as winter, is related to a weakening of the GPLLJ and low (high) rainfall in the upper (lower) Midwest in July. In their analysis of the NARR, for example, this explains 12% of the variability of the jet in July. The response here is opposite, however, with a strengthening of the jet. A weakening of the jet in response to the NAO-like differences in meridional gradients is overwhelmed here by the differences in zonal gradients.

The agreement among the GCM projections for AMJ 2079–99 displayed in Figs. 4 –8 is striking when one considers how much GCM simulations can, and often do, differ, adding to our confidence in these projections of a springtime intensification of the GPLLJ.

6. Implications for Midwest U.S. hydrology

One reason that an understanding of potential changes in the GPLLJ is crucial is because the jet is related to Midwest hydrology, as discussed in section 2. In particular, a strong jet is related to high precipitation rates and flooding. However, precipitation is notoriously difficult to represent in GCMs, and most would agree that we can rely on GCM predictions of changes in the large-scale atmospheric dynamics more than predictions of regional precipitation. Also, the extent to which the GCMs can capture the physics that connects the LLJs and rainfall events is not clear. Lee et al. (2007) examined output from three atmospheric GCMs and found that, while the models were are able capture reasonable simulations of the low-level flow, including its diurnal cycle, they were unable to capture the relationship of the jet to precipitation.

Here we examine monthly mean simulations of the Midwest hydrology in the twentieth and twenty-first centuries to understand if they are consistent with our understanding of observed connections with the jet dynamics.

The quality of the simulated AMJ precipitation over the central United States in the twentieth-century integrations is first examined. Figure 9a displays the AMJ precipitation climatology for 1979–99 from the NARR. Precipitation rates are greatest along the central gulf coast and the lower Midwest, and over southern Florida, with relatively low values along the southern Atlantic coast and over the Atlantic Ocean. Between 95° and 105°W, a strong zonal precipitation gradient extends from southern Texas to South Dakota.

Fig. 9.

AMJ precipitation (mm day−1) averaged over 1979–99 from (a) NARR and (b) GFLD CM2.0, (c) ECHAM5/MPI-OM, (d) MRI CGCM2.3.2a, (e) CNRM-CM3, and (f) HadCM3 coupled GCM simulations. The contour interval is 0.5 mm day−1.

Fig. 9.

AMJ precipitation (mm day−1) averaged over 1979–99 from (a) NARR and (b) GFLD CM2.0, (c) ECHAM5/MPI-OM, (d) MRI CGCM2.3.2a, (e) CNRM-CM3, and (f) HadCM3 coupled GCM simulations. The contour interval is 0.5 mm day−1.

Twentieth-century AMJ precipitation climatologies from five of the six coupled model simulations evaluated in the previous section are displayed in Figs. 9b–f. None of the GCM simulations captures the observed precipitation with great accuracy, consistent with the findings of Ruiz-Barradas and Nigam (2006). Precipitation maxima are generated over North America, but not one model places a maximum along the gulf coast. While rainfall rates are similar to those observed, maxima are displaced. In the GFDL CM2.0, MRI CGCM2.3.2a, and CNRM-CM3 simulations, precipitation rates are excessive to the west of 95°W, and these models do not capture the zonal precipitation gradients along the eastern flank of the Rockies (95°–105°W). The HadCM3 and ECHAM5–MPI-OM simulations are more accurate in capturing the zonal precipitation gradient, but they also place the precipitation maximum too far inland over the central United States instead of along the gulf coast. Precipitation in the NCAR PCM is not shown and is excluded in the hydrological analysis because it places a single, large continental precipitation maximum over the Rockies.

Despite the misplacement of the continental precipitation maximum in the models, magnitudes are reasonable and each model captures some representation of strong zonal precipitation gradients near 95°–105°W. Overall, however, the GCMs capture the GPLLJ more accurately than the springtime rainfall climatology, as is expected.

Springtime (AMJ) precipitation differences between the last two decades of the twentieth and twenty-first centuries from the CNRM-CM3, GFDL CM2.0, and HadCM3 simulations are shown in Figs. 10a–c, respectively. Each model projects an increase in springtime rainfall over the upper Midwest, and a decrease in the lower Midwest and Great Plains. Precipitation decreases in the lower Midwest are greater by a factor of 2 or more in GFDL CM2.0 and cover a larger area. [Seager et al. (2007) analyze this drying in the GFDL model, which is extreme but consistent in sign with results from other models. GFDL CM2.0 also exhibits extremely strong drying over northern Africa in twenty-first-century simulations (Cook and Vizy 2006).] The precipitation enhancement is located farther east in the HadCM3 simulation than in the other two. Despite such differences, the general patterns of rainfall anomalies are similar in the three models.

Fig. 10.

Differences in AMJ precipitation, vertically integrated moisture convergence, and evaporation for 2079–99 minus 1979–99, simulated, respectively, using (a), (d), (g) CNRM-CM3, (b), (e), (h) GFDL CM2.0, and (c), (f), (i) HadCM3. The contour interval is 0.3 mm day−1 in all panels, and positive values are shaded.

Fig. 10.

Differences in AMJ precipitation, vertically integrated moisture convergence, and evaporation for 2079–99 minus 1979–99, simulated, respectively, using (a), (d), (g) CNRM-CM3, (b), (e), (h) GFDL CM2.0, and (c), (f), (i) HadCM3. The contour interval is 0.3 mm day−1 in all panels, and positive values are shaded.

The vertically integrated, climatological atmospheric column moisture balance,

 
formula

is used to relate the precipitation differences of Figs. 10a–c with differences in evaporation and moisture convergence within the atmosphere. In Eq. (1), q is specific humidity and ps is surface pressure, so the second term on the rhs of Eq. (1) is the vertically integrated moisture convergence. If the precipitation differences shown in Figs. 10a–c are directly associated with the intensification of the GPLLJ, then the precipitation must be supported by increases in the vertically integrated moisture convergence without (necessarily) being accompanied by increases in evaporation.

Differences in the AMJ vertically integrated moisture convergence for the CNRM-CM3, GFDL CM2.0, and HadCM3 simulations are displayed in Figs. 10d–f, respectively. In all three simulations, positive anomalies in the vertically integrated moisture convergence are essentially coincident with the positive precipitation anomalies in the upper Midwest, as are the zero lines separating positive and negative values. The agreement does not extend into the lower Great Plains and southwest, where the negative precipitation anomalies do not mirror the negative moisture convergence anomalies.

Note that the westerly upper-level flow shifts to the north in these simulations, moving the baroclinic zone and the influence of transient moisture convergence in the storm tracks to the north as well. An examination of the moisture convergence by transients (not shown) indicates that the influence of this shift is apparent in precipitation fields in eastern Canada and off the coast of Maine, but does not contribute to the precipitation response over the central United States during the spring.

Differences in AMJ evaporation from the three model simulations are shown in Figs. 10g–i. The positive precipitation anomalies of Figs. 10a–c are generally supported by positive evaporation anomalies, but evaporation maxima do not coincide with positive precipitation anomalies in the upper Midwest. In contrast, the negative precipitation and evaporation anomaly maxima in the southern plains and Southwest do coincide, implicating reduced evaporation as an important component in the drying in these regions.

Twenty-first-century differences in the springtime moisture balance [Eq. (1)] from the ECHAM5–MPI-OM and MRI CGCM2.3.2a simulations are displayed in Fig. 11. Similar to the other three models (Figs. 10a–c), twenty-first-century rainfall is greater in the upper Midwest and reduced in the lower Midwest (Figs. 11a,b), although the response is subdued in the MRI CGCM2.3.2a simulation (Fig. 11b). Maxima in the vertically integrated moisture convergence (Figs. 11c,d) coincide with maximum in precipitation differences as for the CNRM-CM3, GFDL CM2.0, and HadCM3 simulations (Figs. 10d–f), but they are significantly smaller. In both the ECHAM5–MPI-OM and MRI CGCM2.3.2a simulations, differences in evaporation (Figs. 11e,f) provide significant support for the rainfall increases in the upper Midwest, but do not determine the structure. In the lower Midwest and Southwest, decreases in evaporation tend to offset increases in column moisture convergence, and there is little change in precipitation.

Fig. 11.

Differences in AMJ precipitation, vertically integrated moisture convergence, and evaporation for 2079–99 minus 1979–99 simulated, respectively using (a), (c), (e) ECHAM5/MPI-OM and (b), (d), (f) MRI CGCM2.3.2a. The contour interval is 0.3 mm day−1 in all panels, and positive values are shaded.

Fig. 11.

Differences in AMJ precipitation, vertically integrated moisture convergence, and evaporation for 2079–99 minus 1979–99 simulated, respectively using (a), (c), (e) ECHAM5/MPI-OM and (b), (d), (f) MRI CGCM2.3.2a. The contour interval is 0.3 mm day−1 in all panels, and positive values are shaded.

To relate the vertically integrated moisture convergence differences (Figs. 10d–f and Figs. 11c,d) to changes in the GPLLJ dynamics, Figs. 12a–d display differences in the 850-hPa moisture transport and moisture convergence fields from the CNRM-CM3, GFDL CM2.0, HadCM3, and ECHAM5–MPI-OM simulations. (The results from the MRI CGCM2.3.2a simulation are similar, but weaker.) Each simulation places an increase in 850-hPa moisture convergence over the upper Midwest (north of about 38°N) in a location similar to the rainfall increases, and a decrease in moisture convergence to the south. The increased moisture convergence is north of the region of maximum jet intensification, and the anomalous moisture divergence is to its south.

Fig. 12.

Differences in the AMJ 850-hPa moisture transport (vectors) and moisture convergence (contours) at 850 hPa for 2079–99 minus 1979–99 from (a) CNRM-CM3, (b) GFDL CM2.0, (c) HadCM3, and (d) ECHAM5–MPI-OM simulations. Positive values are shaded. The contour intervals are 10−9 kg-H20/(kg-air s), and the vector scale is indicated at bottom[kg-H2O/(kg-air) (m s−1)].

Fig. 12.

Differences in the AMJ 850-hPa moisture transport (vectors) and moisture convergence (contours) at 850 hPa for 2079–99 minus 1979–99 from (a) CNRM-CM3, (b) GFDL CM2.0, (c) HadCM3, and (d) ECHAM5–MPI-OM simulations. Positive values are shaded. The contour intervals are 10−9 kg-H20/(kg-air s), and the vector scale is indicated at bottom[kg-H2O/(kg-air) (m s−1)].

As discussed above, the relatively coarse resolution and monthly mean perspective of the IPCC AR4 GCM output presents challenges in relating the GPLLJ dynamics with precipitation anomalies. Even though the moisture balance discussed above associates changes in the low-level southerly flow with changes in precipitation, the exact physical processes of the connection are not apparent and, of course, the generation of precipitation in the GCMs depends on physical parameterizations that are different in each model. But to relate differences in the precipitation and dynamical fields more closely, moist static energy (MSE) profiles were examined. MSE is defined as

 
formula

When MSE decreases away from the surface (a negative profile) the atmosphere is unstable and the potential for convection increases. Because the quantities that are used to calculate the MSE (T, q, and z) are dependent variables in GCMs, resolved on the models’ grids and constrained by governing equations rather than calculated through physical parameterizations, an MSE analysis provides some insight into how the large-scale dynamics connects with the precipitation.

The hypothesis is that the simulated twenty-first-century precipitation increases in the upper Midwest are associated with increased low-level moisture advection into the region by the GPLLJ. This advection could lead to increases in low-level MSE by increasing Lq in Eq. (2), which would contribute to a negative MSE profile. This analysis has the potential for relating precipitation to the dynamics of the low-level flow more directly, and also allows for distinguishing between moisture and temperature effects in the stability of the atmospheric column. For example, the simulated surface warming over the continental United States (Fig. 8) is potentially important in changing the low-level MSE and destabilizing the vertical column.

Figure 13a displays differences (twenty-first- minus twentieth-century simulations) in the AMJ MSE profile from the CNRM-CM3 simulation, averaged over the region of the greatest precipitation increases in that model, namely, 39°–45°N, 88°–78°W. The negative slope of the MSE difference (solid line) indicates that the large-scale environment is more unstable in the future simulation. This increased instability is related to increases in low-level moisture (Lq; dotted line). While future simulated moisture levels are greater throughout the vertical column, the increases below 850 hPa are much larger than the increases aloft. As shown in Fig. 10, these increases in low-level moisture are related to the intensification of the jet.

Fig. 13.

Differences in AMJ MSE and its components for 2079–99 minus 1979–99 in the region of precipitation enhancement for the (a) CNRM-CM3 and (b) HadCM3 simulations. Sold lines denote MSE, dashed lines denote its thermal component (cpT), and dotted lines denote its moisture component (Lq).

Fig. 13.

Differences in AMJ MSE and its components for 2079–99 minus 1979–99 in the region of precipitation enhancement for the (a) CNRM-CM3 and (b) HadCM3 simulations. Sold lines denote MSE, dashed lines denote its thermal component (cpT), and dotted lines denote its moisture component (Lq).

The changed profile of the thermal component (cpT; dashed line) stands in sharp contrast. While there is warming throughout the column, this warming is slightly greater in the middle troposphere than at the surface, which stabilizes the vertical column.

The MSE differences displayed for the CNRM-CM3 simulations in Fig. 13 are nearly identical to those in the ECHAM5–MPI-OM simulations (not shown). Changes in the MSE profile in the HadCM3 and GFDL CM2.0 models are somewhat different, but produce the same general result. For example, in the HadCM3 simulations, MSE averaged over the region between 37°–42°N and 88°–76°W (where the AMJ precipitation intensification is most pronounced) produces the difference profile plotted in Fig. 13b (solid line). Here, the profile below 950 hPa is more unstable, but between 950 and 850 hPa the profile is slightly more stable. Above 850 hPa, the MSE profile is more unstable, with a stronger slope than in the case of CNRM-CM3 and ECHAM5–MPI-OM (Fig. 13a). Again, the changes in the MSE profile are due to the atmosphere’s moisture content (dotted line) rather than its thermal characteristics (dashed line). However, in contrast to the case of CNRM-CM3 and ECHAM5–MPI-OM, the thermal profile is more unstable. The MSE and Lq profiles in Fig. 13b suggest two added sources of moisture—one at the surface (enhanced evaporation) and one at 850 hPa (the GPLLJ).

7. Summary and conclusions

Changes in the Great Plains low-level jet (GPLLJ) by the end of this century (2079–99) resulting from greenhouse gas forcing under the IPCC’s A2 forcing scenario were examined in monthly mean output from 18 coupled GCM simulations, and the implications of these changes for Midwest hydrology were evaluated.

To build confidence in the models’ projections, the accuracy of the simulation of the jet and rainfall in twentieth-century simulations was examined to select those models with more accurate representations. While all of the models produce a GPLLJ, in many it is too weak and in some it is too strong. However, a number of models are able to capture the jet’s position and magnitude reasonably accurately. The simulation of precipitation is not as accurate. While rainfall magnitudes in the spring, for example, are reasonable, the position of the precipitation maximum is often misplaced.

Confidence in model projections is also bolstered when the models agree with each other in a so-called multimodel ensemble. In models that validated well in terms of their representation of the GPLLJ, there is agreement that the GPLLJ will become more intense by the end of the century during the spring, specifically, April, May, and June, with meridional wind speeds increasing by approximately 25%. This signifies jet formation earlier in the year, and also an intensification of the jet at its peak, which occurs in June and July at present.

Not only do the models agree that the springtime GPLLJ will be more intense in the future, but they also agree on why the intensification occurs. The circulation anomaly is largely geostrophic and occurs in response to stronger zonal geopotential height gradients across the central United States. The westward extension of the North Atlantic subtropical high (the Bermuda high) is stronger and shifted westward in the simulations for 2079–99, and this is the main reason for the stronger zonal geopotential height gradients in each simulation and, thereby, the strong jet. While the subtropical Atlantic Ocean warms by 1–2 K in each simulation, warming over the continental United States is greater (3–8 K), resulting in a shift of atmospheric mass off the continent and higher 850-hPa geopotential heights over the subtropical Atlantic. Decreased geopotential heights over the western Great Plains and the foothills of the Rockies also play a role in enhancing zonal gradients.

The implications of the springtime GPLLJ intensification for Midwest hydrology were investigated in five models that produce more accurate precipitation distributions in their simulations of present-day climate. In each of these models, springtime precipitation increases significantly (by 20%–40%) in the upper Midwest (the upper Mississippi Valley), and decreases to the south. An examination of the atmospheric column moisture budget indicates that the precipitation increases in the north are more closely associated with changes in moisture convergence by the atmosphere than with evaporation changes, but that evaporation plays a strong role in decreasing rainfall rates over the southern Great Plains and Mississippi Valley. Much of the enhanced convergence that supports the rainfall increases in the north is occurring as a result of meridional convergence at 850 hPa, connecting the rainfall changes with the GPLLJ intensification. While there is surface warming, an evaluation of moist static energy profiles shows that the destabilization of the vertical column that enhances convection is primarily due to changes in the moisture content of the lower atmosphere.

The simulated moisture balance and circulation differences for the twenty-first century are similar to the observed moisture balance and circulation anomalies for May and June 1993, both of which are months of strong flooding throughout the Mississippi Valley. For example, precipitation anomalies for June 1993, shown in Fig. 14a, featured increased rainfall in the upper Midwest that was supported primarily by atmospheric moisture convergence and not enhanced evaporation, according to the NARR (not shown). The high rainfall rates along the gulf coast in 1993 are not reproduced in the future simulations, just as the simulations of the twentieth century miss this rainfall maximum (Fig. 9). The positive precipitation anomalies are associated with enhanced meridional flow and moisture advection across the central United States, and an enhancement of the Bermuda high, as displayed in Fig. 14b. Such analogies between present-day observed climate variability and climate predictions help build confidence in the simulations of the future.

Fig. 14.

(a) Monthly mean precipitation anomalies are shown in mm day−1 from the NARR for June 1993. Contour interval is 1 mm day−1 and values greater than +1 mm day−1 are shaded. (b) 850-hPa wind and geopotential height anomalies (gpm) for June 1993 from the NARR, with positive values shaded. Vector scale (m s−1).

Fig. 14.

(a) Monthly mean precipitation anomalies are shown in mm day−1 from the NARR for June 1993. Contour interval is 1 mm day−1 and values greater than +1 mm day−1 are shaded. (b) 850-hPa wind and geopotential height anomalies (gpm) for June 1993 from the NARR, with positive values shaded. Vector scale (m s−1).

Acknowledgments

This work was supported by NSF Award ATM-0701129. Some of the research presented contributed to the Undergraduate Honors Research project in Economics at Cornell of one of the authors (ZSL). The authors thank S. Nigam, S. Weaver, Editor A. Weaver, and two anonymous reviewers for their insightful comments.

REFERENCES

REFERENCES
Anderson
,
C. J.
, and
Coauthors
,
2003
:
Hydrological processes in regional climate model simulations of the central United States flood of June–July 1993.
J. Hydrometeor.
,
4
,
584
598
.
Arritt
,
R. W.
,
T. D.
Rink
,
M.
Segal
,
D. P.
Todey
,
C. A.
Clark
,
M. J.
Mitchell
, and
K. M.
Labas
,
1997
:
The Great Plains low-level jet during the warm season of 1993.
Mon. Wea. Rev.
,
125
,
2176
2192
.
Bell
,
G. D.
, and
J. E.
Janowiak
,
1995
:
Atmospheric circulation associated with the Midwest floods of 1993.
Bull. Amer. Meteor. Soc.
,
76
,
681
695
.
Blackadar
,
A. K.
,
1957
:
Boundary layer wind maxima and their significance for their growth of nocturnal inversions.
Bull. Amer. Meteor. Soc.
,
38
,
283
290
.
Bosilovich
,
M. G.
, and
W. Y.
Sun
,
1999
:
Numerical simulation of the 1993 Midwestern flood: Land–atmosphere interactions.
J. Climate
,
12
,
1490
1505
.
Byerle
,
L. A.
, and
J.
Paegle
,
2003
:
Modulation of the Great Plains low-level jet and moisture transports by orography and large-scale circulations.
J. Geophys. Res.
,
108
.
8611, doi:10.1029/2002JD003005
.
Cook
,
K. H.
, and
E. K.
Vizy
,
2006
:
Coupled model simulations of the West Arican monsoon system: Twentieth- and twenty-first century simulations.
J. Climate
,
19
,
3681
3703
.
Dirmeyer
,
P. A.
, and
K. L.
Brubaker
,
1999
:
Contrasting evaporative moisture sources during the drought of 1988 and the flood of 1993.
J. Geophys. Res.
,
104
,
D16
.
19383
19397
.
Gray
,
S. T.
,
J. L.
Betancourt
,
C. L.
Fastie
, and
S. T.
Jackson
,
2003
:
Patterns and sources of multidecadal oscillations in drought-sensitive tree-ring records from the central and southern Rocky Mountains.
Geophys. Res. Lett.
,
30
.
1316, doi:10.1029/2002GL016154
.
Gregory
,
J. M.
, and
Coauthors
,
2005
:
A model intercomparison of changes in the Atlantic thermohaline circulation in response to increasing CO2 concentration.
Geophys. Res. Lett.
,
32
.
L12703, doi:10.1029/2005GL023209
.
Helfand
,
H. M.
, and
S. D.
Schubert
,
1995
:
Climatology of the simulated Great Plains low-level jet and its contribution to the continental moisture budget of the United States.
J. Climate
,
8
,
784
806
.
Holton
,
J. R.
,
1967
:
The diurnal boundary layer wind oscillation above sloping terrain.
Tellus
,
19
,
199
205
.
Hoxit
,
L. R.
,
1975
:
Diurnal variations in planetary boundary-layer winds over land.
Bound.-Layer Meteor.
,
8
,
21
38
.
Kanamitsu
,
M.
,
W.
Ebisuzaki
,
J.
Woollen
,
S. K.
Yang
,
J. J.
Hnilo
,
M.
Fiorino
, and
G. L.
Potter
,
2002
:
NCEP-DOE AMIP-II Reanalysis (R2).
Bull. Amer. Meteor. Soc.
,
83
,
1631
1643
.
Lee
,
M. I.
, and
Coauthors
,
2007
:
An analysis of the warm-season diurnal cycle over the continental United States and northern Mexico in general circulation models.
J. Hydrometeor.
,
8
,
344
366
.
McCorcle
,
M. D.
,
1988
:
Simulation of surface-moisture effects on the Great Plains low-level jet.
Mon. Wea. Rev.
,
116
,
1705
1720
.
Means
,
L. L.
,
1952
:
On thunderstorm forecasting in the central United States.
Mon. Wea. Rev.
,
80
,
165
189
.
Mesinger
,
F.
, and
Coauthors
,
2006
:
North American regional reanalysis.
Bull. Amer. Meteor. Soc.
,
87
,
343
360
.
Miguez-Macho
,
G.
,
G. L.
Stenchikov
, and
A.
Robock
,
2005
:
Regional climate simulations over North America: Interaction of local processes with improved large-scale flow.
J. Climate
,
18
,
1227
1246
.
Mitchell
,
M. J.
,
W.
Arritt
, and
K.
Labas
,
1995
:
A climatology of the warm season Great Plains low-level jet using wind profiler observations.
Wea. Forecasting
,
10
,
576
591
.
Mo
,
K. C.
, and
H-M. H.
Juang
,
2003
:
Relationships between soil moisture and summer precipitation over the Great Plains and the Southwest.
J. Geophys. Res.
,
108
.
8610, doi:10.1029/2002JD002952
.
Nigam
,
S.
, and
A.
Ruiz-Barradas
,
2006
:
Seasonal hydroclimate variability over North America in global and regional reanalyses and AMIP simulations: Varied representations.
J. Climate
,
19
,
4041
4058
.
Ringler
,
T. D.
, and
K. H.
Cook
,
1995
:
Orographically induced stationary waves: Dependence on latitude.
J. Atmos. Sci.
,
52
,
2548
2560
.
Ringler
,
T. D.
, and
K. H.
Cook
,
1999
:
Understanding the seasonality of orographically forced stationary waves: Interaction between mechanical and thermal forcing.
J. Atmos. Sci.
,
56
,
1154
1174
.
Ruiz-Barradas
,
A.
, and
S.
Nigam
,
2005
:
Warm-season precipitation variability over the U.S. Great Plains in observations, NCEP and ERA-40 reanalyses, and NCAR and NASA atmospheric simulations.
J. Climate
,
18
,
1808
1830
.
Ruiz-Barradas
,
A.
, and
S.
Nigam
,
2006
:
IPCC’s twentieth-century climate simulations: Varied representations of North American hydroclimate variability.
J. Climate
,
19
,
4041
4058
.
Schneider
,
B.
,
M.
Latif
, and
A.
Schmittner
,
2007
:
Evaluation of different methods to assess model projections of the future evolution of the Atlantic meridional overturning circulation.
J. Climate
,
20
,
2121
2132
.
Seager
,
R.
, and
Coauthors
,
2007
:
Model projections of an imminent transition to a more arid climate in southwestern North America.
Science
,
316
,
1181
1184
.
Song
,
J.
,
K.
Liao
,
R. L.
Coulter
, and
B. M.
Lesht
,
2005
:
Climatology of the low-level jet at the southern Great Plains atmospheric boundary layer experiments site.
J. Appl. Meteor.
,
44
,
1593
1606
.
Stahle
,
D. W.
,
F. K.
Fye
,
E. R.
Cook
, and
R. D.
Griffin
,
2007
:
Tree-ring reconstructed megadroughts over North America since 1300AD.
Climatic Change
,
83
,
133
149
.
Stensrud
,
D. J.
,
1996
:
Importance of low-level jets to climate: A review.
J. Climate
,
9
,
1698
1711
.
Sud
,
Y. C.
,
M.
Mocko
,
K. M.
Lau
, and
R.
Atlas
,
2003
:
Simulating the Midwestern U.S. drought of 1988 with a GCM.
J. Climate
,
16
,
3946
3965
.
Sun
,
W. Y.
,
J. D.
Chern
, and
M.
Bosilovich
,
2004
:
Numerical study of the 1988 drought in the United States.
J. Meteor. Soc. Japan
,
82
,
1667
1678
.
Ting
,
M. F.
, and
H. L.
Wang
,
2006
:
The role of the North American topography on the maintenance of the Great Plains low-level jet.
J. Atmos. Sci.
,
63
,
1056
1068
.
Trenberth
,
K. E.
, and
C. J.
Guillemot
,
1996
:
Physical processes involved in the 1988 drought and 1993 floods in North America.
J. Climate
,
9
,
1288
1298
.
Tuttle
,
J. D.
, and
C. A.
Davis
,
2006
:
Corridors of warm season precipitation in the central United States.
Mon. Wea. Rev.
,
134
,
2297
2317
.
Uccellini
,
L.
,
1980
:
On the role of upper tropospheric jet streaks and leeside cyclogenesis in the development of low-level jets in the Great Plains.
Mon. Wea. Rev.
,
108
,
1689
1696
.
Walters
,
C. K.
,
2001
:
Airflow configurations of warm season southerly low-level wind maxima in the Great Plains. Part II: The synoptic and subsynoptic-scale environment.
Wea. Forecasting
,
16
,
531
551
.
Walters
,
C. K.
, and
J. A.
Winkler
,
2001
:
Airflow configurations of warm season southerly low-level wind maxima in the Great Plains. Part I: Spatial and temporal characteristics and relationship to convection.
Wea. Forecasting
,
16
,
513
530
.
Weaver
,
S. J.
, and
S.
Nigam
,
2008
:
Variability of the Great Plains low-level jet: Large-scale circulation context and hydroclimate impacts.
J. Climate
,
21
,
1532
1551
.
Wexler
,
H.
,
1961
:
A boundary layer interpretation of the low level jet.
Tellus
,
13
,
368
378
.

Footnotes

* Current affiliation: The University of Texas at Austin, Austin, Texas

Corresponding author address: Dr. Kerry H. Cook, Jackson School of Geosciences, C1100, The University of Texas at Austin, Austin, TX 78712. Email: khc6@cornell.edu