Abstract

Persistent wet and dry events over the central United States are examined during summer. Composites based on selected persistent wet and dry events reveal common atmospheric processes and circulation features. During summer, heavy precipitation in the central United States is accompanied by less precipitation to the south, in a band that extends from the tropical eastern Pacific through the Gulf of Mexico into the western Atlantic. Dry conditions also occur along the western coasts of Canada and Mexico during persistent wet episodes in the central United States. This rainfall pattern is supported by an inverse temperature–rainfall relationship over North America. During dry events, high pressure extends throughout a vertical column in a pattern that covers North America from 30° to 60°N. In contrast, during wet events, the high pressure is confined to the eastern half of North America, with low pressure prevailing in the western half. Increased northward meridional winds are found between this cyclonic–anticyclonic dipole, leading to increased moisture flux from the Gulf of Mexico at low levels.

A significant precursor to wet events is the enhancement of westerlies over the eastern Pacific and western North America from 30° to 40°N. Synoptic-scale eddies intensify prior to onset and accelerate this westerly flow as revealed by Eliassen–Palm flux diagnostics. One pentad before onset, rainfall begins in Texas, and the low level jet (LLJ) in the Great Plains strengthens. The intensified LLJ transports moisture into the central United States and the moisture convergence downwind from the LLJ maintains rainfall. For dry events, heating occurs in the tropical eastern Pacific associated with the northward shift of the ITCZ roughly one pentad prior to onset. The prevailing easterly flow over subtropical portions of North America is not conducive to moisture transport into the United States, and without the support of moisture influx from the Gulf of Mexico, dry conditions prevail.

1. Introduction

Extreme precipitation events such as summer floods and severe droughts have a large impact on many segments of the economy, including agriculture, industry, and the water supply. A better understanding of the evolution of these events and their associated circulation features is important for climate monitoring and prediction. Persistent flood and drought episodes are usually accompanied by persistent atmospheric conditions and large-scale dynamical support. Most studies of floods and droughts employ monthly or seasonal mean station data. Madden and Williams (1978) analyzed seasonal mean station data for North America and found negative correlations between temperature and precipitation. Similar inverse relationships have also been observed by Karl and Quayle (1981) using monthly mean data. Chang and Wallace (1987) examined meteorological conditions during summer heat waves and droughts. They found that hot months were characterized by upper-level anticyclones over the central United States and a deep trough along the west coast of the United States. In addition to statistical studies, there have been many case studies of extreme events, such as the 1988 United States drought and the 1993 United States flood, carried out to identify the key physical processes responsible for these events. However, many extreme events, especially floods, last less than a month. Attempts to study these events have been limited by the lack of consistent datasets.

It has been hypothesized that extreme events are influenced by tropical convection. Trenberth et al. (1988) suggested that sea surface temperature anomalies set up a teleconnection pattern with a persistent ridge over the central United States, which caused the 1988 drought. GCM experiments by Palmer and Brankovic (1989) supported this theory. However, a diagnostic analysis by Lyon (1991) did not find observational evidence for this mechanism. An alternative explanation offered by Mo et al. (1991) relied on a normal mode type of instability as an explanation for the unusual persistence of the 1988 pattern. In a later paper, Trenberth and Branstator (1992) suggested that the changes in atmospheric heating in the eastern Pacific associated with the northward shift of the ITCZ might explain the 1988 drought. Lau and Peng (1992) used a vorticity balance model with a specified divergence field to demonstrate that heating in the tropical Pacific (150°E–90°W), just north of the equator, can excite a wave train that appears similar to the observed one of 1988. Their model responses are not sensitive to the exact location of the heating. For the 1993 floods, Mo et al. (1995) suggested that persistent flooding was linked to marked transient eddy activity in the flow in late May or early June. The eddy feedback onto the mean flow caused the strengthening and southward shift of the North American jet, which in turn generated more rain. However, it is still possible that SST anomalies during the warm 1992 ENSO episode were linked to the enhanced storm activity in late May (Bell and Janowiak 1995).

In addition to large-scale dynamical support, the local hydrologic conditions are also important. Namias (1989) emphasized the importance of reduced soil moisture on maintaining warm and dry conditions during summer; reduced soil moisture leads to increased surface heating and reduced local evaporation. Oglesby and Erickson (1989) found in their GCM study that a reduction in summer soil moisture over North America produces atmospheric conditions observed during droughts and leads to a sharp decrease in rainfall. They also stressed the importance of moisture advection from the Gulf of Mexico. The GCM study by Wolfson et al. (1987) confirmed the importance of soil moisture on heat waves. They concluded that there is a positive feedback between extremely dry soil conditions, the persistence of warm temperatures, and droughts. For the 1993 floods, GCM experiments by Beljaars et al. (1996) suggested the importance of a soil moisture reservoir. Increased evaporation and moist soil conditions upstream provide a favorable condition for floods. Paegle et al. (1996) and Higgins et al. (1997a) emphasized the role played by the low-level jet (LLJ), which transports moisture from the Gulf of Mexico. The strong LLJ enhances moisture convergence downwind of the jet core and increases precipitation.

These case studies provide insight into the maintenance and physical mechanisms of selected events. Each event has its own characteristics and it is difficult to assess how representative a given case is. In this paper, our objectives are 1) to identify systematic features in the large-scale flow preceding and during persistent wet and dry episodes and 2) to examine the dynamical and hydrologic processes and the time evolution of these events.

Recently, NCEP (National Centers for Environmental Prediction) in collaboration with NCAR (National Center for Atmospheric Research) has completed a multiyear (1979–95) reanalysis. The reanalysis provides valuable data for studies of multiple flood and drought events since it is produced with a fixed assimilation system and a large input database. The model and procedures used in the NCEP–NCAR reanalysis are documented in Kalnay et al. (1996). Evaluation of large- scale aspects of the hydrological cycle revealed by reanalysis (Mo and Higgins 1996; Higgins et al. 1996b) indicates that the analysis products are generally quite realistic, especially in the data-rich Northern Hemisphere. The reanalysis forecast products have biases, particularly in the 6-h rainfall forecasts that considerably underestimate the diurnal cycle and overestimate daily rainfall over the southeastern United States (Higgins et al. 1996b). These biases, which are largely systematic, can be removed. Together with other independent data sources, such as observed hourly precipitation (Higgins et al. 1996a) and outgoing longwave radiation (OLR), the reanalysis data provide useful information on relationships between rainfall and circulation features.

The independent precipitation dataset used is a set of gridded hourly precipitation analyses over the conterminous United States (Higgins et al. 1996a), based on station observations. Since precipitation does not follow a normal distribution, we apply the square root transformation (Lanzante and Harnack 1982) to rainfall amounts before determining the distribution function. The data and diagnostic procedures are described in section 2. Section 3 examines moisture transport and the linkage between the LLJ and rainfall in the central United States based on composites. Large-scale circulation features during extreme events are discussed in section 4. Section 5 describes the time evolution and linkages between circulation patterns and moisture transport. A summary and discussion are given in sec~tion 6.

2. Data and procedures

a. Data

The primary dataset used in this study is daily averaged global gridded analysis/forecast products from the NCEP–NCAR reanalysis for the period 1979–95. The NCEP data are on a 2.5° × 2.5° latitude–longitude grid and there are 28 levels in the vertical with seven levels below 850 hPa. Rainfall and evaporation were taken from 6-h forecasts. The quality of these forecast products and the atmospheric component of the hydrological cycle simulated by the NCEP reanalysis is discussed by Higgins et al. (1996b) and Mo and Higgins (1996). Daily averages of the National Oceanic and Atmospheric Administration (NOAA) satellite OLR (Liebmann and Smith 1996) for the same period were used to represent tropical convection. We focus on the Northern Hemisphere summer defined as 1 May to 30 September. The seasonal cycle at each grid point is defined as the grand mean plus the first and second harmonics with periods of 12 and 6 months, respectively. Anomalies are defined as deviations from the seasonal cycle. A low-pass filter (Blackmon 1976) is applied to the anomalies at each grid point to remove periods less than about 10 days, and synoptic-scale eddies are obtained with a bandpass filter that retains periods between 2.5 and 6 days.

The second dataset used in this study is the gridded hourly precipitation over the conterminous United States (Higgins et al. 1996a). These are gridded analyses of station observations. The horizontal resolution is 2° latitude × 2.5° longitude.

b. Eliassen–Palm flux diagnostics

To examine the role played by synoptic-scale eddies, we use the extended Eliassen–Palm (E–P) formalism of Hoskins et al. (1983) later modified by Trenberth (1986). The total time change of the zonal mean wind is proportional to the divergence of the E–P flux E, given by

 
formula

with

 
Φz = RT

and the divergence operator given by

 
formula

with

 
ρ0(z) = ρ00ez/H,

where R is the gas constant, f is the Coriolis parameter, S is the static stability, T is the temperature, z/H = ln(p0/p), p is the pressure, p0 = 1000 hPa, ρ00 is the density at the surface, and H is the scale height, which is 8 km. The primes denote bandpass-filtered (periods between about 2.5 to 6 days) eddies.

The static stability is defined as

 
formula

where κ is R/cp. Here, S is calculated using low-pass- filtered (periods greater than about 10 days) global mean temperature instead of the climatological mean temperature.

The horizontal component of the E–P flux vector is vertically coherent and represents the barotropic forcing of the eddies by the mean flow. The vertical component is determined by the poleward transport of heat flux and is large when baroclinicity is pronounced. Here, E can be viewed as an effective momentum flux. When E is divergent (convergent), the eddy forcing tends to increase (decrease) the westerly mean wind.

3. Wet and dry episodes in the central United States

a. Rainfall distribution

The observed mean (1963–95) precipitation (Higgins et al. 1996a) for summer (May–September) (Fig. 1a) shows that the maximum rainfall occurs over the central plains and along the gulf coast. For this study, we focus on a box (34°–46°N, 105°–85°W) centered over the central Mississippi Valley (Fig. 1a); mean precipitation averaged over this box is used to select wet and dry events. The mean daily precipitation (1963–95) for the box (Fig. 1b) shows a rapid increase in rainfall during April and May. The rainfall reaches a maximum in June and is followed by a slow decrease. There is a weak secondary maximum in September. The annual mean for the box is about 2.1 mm day−1. Since the heaviest precipitation in the central United States occurs during the period from May to September, we define these five months as the summer season.

Fig. 1.

(a) Mean (1963–95) observed precipitation for summer (May–September). Contour interval 0.5 mm day−1, (b) daily mean observed precipitation (1963–95) averaged over the box (34°–46°N, 105°–85°W) marked in (a). Units are 1 mm day−1.

Fig. 1.

(a) Mean (1963–95) observed precipitation for summer (May–September). Contour interval 0.5 mm day−1, (b) daily mean observed precipitation (1963–95) averaged over the box (34°–46°N, 105°–85°W) marked in (a). Units are 1 mm day−1.

Richman and Lamb (1985) examined 3–5-day summer rainfall patterns over the central United States. Because of the positive skewness of rainfall distributions, they suggested using the square root or log10 transformation in analyzing rainfall data. The square root transformation (Lanzante and Harnack 1982) is chosen for the present study. Histograms of time-averaged rainfall for the box marked on Fig. 1a during summer (May to September) were examined using observed daily data from 1963 to 1995 (Higgins et al. 1996a). To increase the sample size, the histograms were computed by pooling all grid points in the box together. For each grid point, we formed time series of 3-day total rainfall and obtained the maximum rainfall M of the time series. The rainfall distribution was calculated by dividing rainfall into 20 bins between 0 and M. We then repeated the process to obtain histograms for 5-, 7-, 9-, and 11-day total rainfall (Fig. 2a). Consistent with Epstein (1988) and Richman and Lamb (1985), all rainfall distributions show positive skewness, but the skewness decreases as the number of days included in the mean increases. Calculations were repeated for rainfall after the square root transformation. The resulting histograms (Fig. 2b) are closer to normal distributions.

Fig. 2.

(a) Histogram of 3- (open circles), 5- (dark circles), 7- (open squares), 9- (dark squares), and 11- (crosses) day total rainfall for all grid points in the box (34°–46°N, 105°–85°W). The size of the bin is maximum of rainfall divided by 20, and (b) same as (a) but after the square root transformation of rainfall.

Fig. 2.

(a) Histogram of 3- (open circles), 5- (dark circles), 7- (open squares), 9- (dark squares), and 11- (crosses) day total rainfall for all grid points in the box (34°–46°N, 105°–85°W). The size of the bin is maximum of rainfall divided by 20, and (b) same as (a) but after the square root transformation of rainfall.

To select events, we formed time series of 5-day running mean precipitation summed over the above box during summer. The distribution function obtained after the square root transformation is similar to the histogram obtained by pooling data together (Fig. 2b). Rainfall percentiles were determined using that distribution curve (Fig. 2b, dark circles). We selected extreme rainfall events using the threshold-crossing procedure outlined by Dole and Gordon (1983) with one modification:The magnitude criterion is replaced by a percentile requirement. Wet (dry) events are chosen when the square root of 5-day running mean precipitation is above the 80th percentile (below the 20th percentile) for at least 10 consecutive days. The duration of an event is defined as the period from the first crossing of the threshold (onset) to the next crossing of the threshold (demise). When an event begins in April and ends in May, the April starting date is used. When an event begins in September and ends in October, the October ending date is used. Over the 17-yr period from 1979 to 1995, there are 21 wet and 19 dry events with an average duration for both wet and dry events of 17 days. For wet events, there are six events in May, five in June, four in July, three in August, and three in September. For dry events, there are five events in May, one in June, four in July, four in August, and five in September. Thus, both wet and dry events are fairly evenly distributed throughout the summer. We then obtained composite fields for wet and dry events by averaging over the duration of each event and then averaging over all of the events. The statistical significance was assessed by assuming a normal distribution for all variables except rainfall. For rainfall, the square root transformation was applied before assessing statistical significance. A decorrelation time of 10 days is used when calculating degrees of freedom.

b. Moisture transport during wet and dry events

Composites for extreme events show rainfall concentrated over the central United States during wet events (Fig. 3a). During dry events, rainfall maxima are found over Florida and along the gulf coast and over Arizona and New Mexico (Fig. 3b). Similar rainfall patterns were also observed by Walsh et al. (1982). Consistent with Higgins et al. (1997b), the heavy precipitation in the central United States is accompanied by a weaker North American monsoon. These rainfall features are well captured by the NCEP 6-h forecasts (Fig. 4a). Examination of the mean precipitation for the spring and summer seasons over the United States (Higgins et al. 1996b; Mo and Higgins 1996) indicates that the NCEP reanalysis underestimates the daily rainfall variability in the southern plains and the number of days with rainfall rates less than 1 mm day−1. These biases are systematic since it appears only in the composites of rainfall averaged over extreme events but not in the rainfall difference between wet and dry events. Heavy rainfall events in the central United States are flanked by below- normal rainfall along the southeastern coast of Alaska and British Columbia and in a southwest to northeast band that extends from 120°W to the western Atlantic (Fig. 4a). A similar precipitation pattern also occurs during the mature phase of the dry North American monsoon (Douglas et al. 1993; Higgins et al. 1997).

Fig. 3.

Composite of rainfall for (a) wet events. Contour interval 1, 2, 4, 6, and 8 mm day−1. (b) Same as (a) but for dry events. Contour interval 0.5, 1, 2, and 3 mm day−1 and (c) same as (a) but for the rainfall difference between wet and dry events from observations. Contour intervals are 0.5, 1, 2, 4, and 6 mm day−1. Negative values are shaded.

Fig. 3.

Composite of rainfall for (a) wet events. Contour interval 1, 2, 4, 6, and 8 mm day−1. (b) Same as (a) but for dry events. Contour interval 0.5, 1, 2, and 3 mm day−1 and (c) same as (a) but for the rainfall difference between wet and dry events from observations. Contour intervals are 0.5, 1, 2, 4, and 6 mm day−1. Negative values are shaded.

Fig. 4.

Composite of (a) rainfall, (b) evaporation rate, and (c) 850-hPa temperature difference between wet and dry events. Contour intervals are 1 mm day−1 for (a), 0.3 mm day−1 for (b), and 0.8°C for (c). Area where values are significant at the 95% level are shaded.

Fig. 4.

Composite of (a) rainfall, (b) evaporation rate, and (c) 850-hPa temperature difference between wet and dry events. Contour intervals are 1 mm day−1 for (a), 0.3 mm day−1 for (b), and 0.8°C for (c). Area where values are significant at the 95% level are shaded.

Evaporation differences suggest linkages between surface evaporation and precipitation. Wet events exhibit higher evaporation (Fig. 4b) over the central plains than dry cases with lower evaporation in the Gulf of Mexico and southern Texas near the entrance region of the LLJ (Fig. 5a). Beljaars et al. (1996) demonstrated that enhanced soil moisture in the central United States increases the likelihood of a GCM to simulate the 1993 flood. Consistent with Beljaars et al. (1996), the maximum evaporation difference is located southwest of the maximum precipitation difference. Positive evaporation anomalies in the rainfall region support the positive feedback theory of Mintz (1984). Our results are also relevant to the hypothesis that reduced soil moisture over interior North America causes a reduced local evaporation and may result in drought as suggested in the model experiments by Oglesby and Erickson (1989). Figure 4b also shows the decreased evaporation in the gulf and southern Texas, consistent with the findings of Paegle et al. (1996). McCorcle (1986) suggested that relatively dry conditions in the southern Great Plains may contribute to the dynamical support of floods over Oklahoma. McCorcle (1986) and Paegle et al. (1996) reasoned that drier surface conditions enhance the surface warming. The warmer daytime surface conditions lead to greater low-level buoyancy and resulted in a strengthening of the LLJ and its convergence downwind from the jet core. This in turn generates additional rainfall. The above reasoning is supported by the 850-hPa temperature difference between wet and dry events (Fig. 4c), which shows warmer temperature anomalies over southern Texas for wet events. The temperature difference also indicates an inverse temperature–rainfall relationship in the central United States as suggested by Madden and Williams (1978). It also resembles the correlation map between 500-hPa monthly mean heights and surface air temperature at Kansas (Chang and Wallace 1987) suggesting that persistent heat waves are likely associated with droughts while cold temperatures are linked to floods.

Fig. 5.

Composite of vertically integrated moisture flux (arrows) and flux divergence D(Q) (shading) for (a) wet and (b) dry events. The unit is mm day−1 for D(Q) and 10 kg (m s)−1 for moisture flux.

Fig. 5.

Composite of vertically integrated moisture flux (arrows) and flux divergence D(Q) (shading) for (a) wet and (b) dry events. The unit is mm day−1 for D(Q) and 10 kg (m s)−1 for moisture flux.

The rainfall pattern (Fig. 4a) is at least partly explained by strong moisture transport from the Gulf of Mexico by the LLJ and strong moisture flux convergence downwind from the jet core as previously discussed. Figures 5a and 5b show the vertically integrated moisture flux and the moisture flux divergence D(Q). Here, moisture transport was computed directly from winds and specific humidity every 6 h on the model sigma levels. A low-pass filter, which removes fluctuations with periods less than about 10 days, was applied to the moisture flux and divergence fields before compositing. For wet events (Fig. 5a), the meridional component of moisture flux is strong and a transport of moisture from the Atlantic through the Gulf of Mexico and onto the central plains is evident with outflow across the east coast of North America. The moisture flux convergence, located downwind from the Great Plains LLJ core, enhances rainfall. Moisture flux divergence is found over the subtropical eastern Pacific and over the Gulf of Mexico and the Atlantic, so these regions are relatively dry. Along the west coast, the moisture flow is parallel to the Pacific coast with very little influx of moisture.

For dry events (Fig. 5b), the easterlies transport moisture to Mexico and Central America. The enhanced moisture flux convergence is consistent with excessive rainfall there. There is little moisture transport (or moisture flux convergence) into the central United States, favoring dry conditions. The main moisture “pipeline” is associated with the anticyclone off the east coast of the United States and is located east of the main pipeline for the wet case. The moisture is transported from the Caribbean northeastward along the southeastern United States coast into the Atlantic. The moisture flux convergence supports rainfall in Florida, but the major center of moisture flux convergence is located in eastern Canada and in the Atlantic. Along the west coast, the strongest inflow shifts to western Canada providing this region with abundant moisture.

The moisture balance is given in Fig. 6. The precipitation minus evaporation (Precip − Evap) difference between wet and dry events (Fig. 6a) shows the same pattern as the rainfall difference. Both precipitation and evaporation were taken from the 6-h forecasts in the assimilation cycle. If the model is perfect, the Precip − Evap difference should be balanced by the vertically integrated moisture flux divergence D(Q) difference [Precip − Evap + D(Q) = 0] since the total precipitable water (TPW) tendency [d(TPW)/dt] is small (Mo and Higgins 1996). The D(Q) difference (Fig. 6b) shows a similar pattern to Precip − Evap with opposite sign as expected, but there is an imbalance of roughly 1 mm day−1 near the rainfall maximum. This is largely due to the fact that the assimilation model underestimates daily rainfall variability (Higgins et al. 1996b). Differences in the moisture transport (Fig. 6d) show increased northward transport located between an anomalous cyclonic circulation center to the northwest and an anticyclonic circulation center to the southeast. A third, much weaker, anticyclonic center is located in the Pacific. This three-cell pattern is consistent with the composite difference of 850-hPa temperature (Fig. 4c) and with composites of nocturnal moisture flux anomalies during LLJ events (Higgins et al. 1997a). The composite difference of the total precipitable water (Fig. 6c) is consistent with the moisture inflow (Fig. 6d). Moisture is transported from the Gulf of Mexico to the central United States with the strongest outflow across the east coast of the United States. That leaves more TPW over the central and eastern United States and less TPW over western North America.

Fig. 6.

Composite of (a) Precip − Evap, (b) vertically integrated moisture flux divergence D(Q), (c) total precipitable water, and (d) vertically integrated moisture flux difference between wet and dry events. Contour interval 1 mm day−1 for (a) and (b) and 0.2 cm for (c). The unit for moisture flux is 10 kg (m s)−1.

Fig. 6.

Composite of (a) Precip − Evap, (b) vertically integrated moisture flux divergence D(Q), (c) total precipitable water, and (d) vertically integrated moisture flux difference between wet and dry events. Contour interval 1 mm day−1 for (a) and (b) and 0.2 cm for (c). The unit for moisture flux is 10 kg (m s)−1.

The moisture flux composite (Fig. 5) indicates the importance of the Great Plains LLJ, which transports considerable moisture at low levels. Figure 7 shows vertical profiles of the meridional wind at 30°N for wet and dry events. The LLJ is centered in the boundary layer and has a maximum between 875 and 925 hPa for both cases, but the LLJ for wet periods is considerably stronger and is shifted slightly eastward. There is a large difference in magnitude. The maximum for wet events exceeds 8 m s−1 and it is only 4 m s−1 for dry events. During wet periods, the northerlies along the west coast extend to 500 hPa while they are confined below 700 hPa for dry events. For dry events, there is a sign reversal of meridional winds between upper and lower levels. This sign reversal tilts toward the east with height for wet events.

Fig. 7.

Composite of the vertical profile of meridional wind at 30°N for (a) wet and (b) dry events. Contour interval is 1 m s−1.

Fig. 7.

Composite of the vertical profile of meridional wind at 30°N for (a) wet and (b) dry events. Contour interval is 1 m s−1.

c. Diurnal cycle

Higgins et al. (1997a) among others have shown that there is a strong diurnal cycle of rainfall over the central United States with a nocturnal maximum. We examined the diurnal cycle of rainfall during wet events by compositing observed rainfall every 3 h (Fig. 8). The maximum occurs during 0400–0600 UTC, which is consistent with Paegle et al. (1996) that during persistent wet events, most rainfall is nocturnal.

Fig. 8.

Composite of precipitation from (a) 0100–0300, (b) 0400–0600, (c) 0700–0900, (d) 1000–1200, (e) 1300–1500, (f) 1600–1800, (g) 1900–2100, and (h) 2200–2400 UTC for wet events. Contour interval is 0.5 mm (3 h)−1. Areas where rainfall is greater than 1 mm are shaded.

Fig. 8.

Composite of precipitation from (a) 0100–0300, (b) 0400–0600, (c) 0700–0900, (d) 1000–1200, (e) 1300–1500, (f) 1600–1800, (g) 1900–2100, and (h) 2200–2400 UTC for wet events. Contour interval is 0.5 mm (3 h)−1. Areas where rainfall is greater than 1 mm are shaded.

The rainfall path follows closely the trajectory of the LLJ as indicated by composites of vertically integrated meridional moisture transport anomalies (departure from the daily mean) at 6-h intervals (Fig. 9). The maximum of the LLJ occurs at 0600 UTC, which sustains nocturnal convection [2300 Central Standard Time (CST)]. The connection between this maximum and the nocturnal rainfall is consistent with the findings of Helfand and Schubert (1995), Bonner (1968), Bonner and Paegle (1970), and Higgins et al. (1997a). The LLJ persists until about 1200 UTC (0500 CST) and then diminishes. In contrast, the Baja jet along the west coast shows weaker diurnal variability. The structure of the LLJ and its diurnal oscillation are similar to the LLJ climatology (Higgins et al. 1997a). During wet events, the LLJ extends northward with an increased magnitude and is associated with enhanced moisture convergence.

Fig. 9.

Vertically integrated meridional moisture flux at (a) 0000, (b) 0600, (c) 1200, and (d) 1800 UTC for wet events expressed as the departure from the daily mean, (000 + 0600 + 1200 + 1800 UTC)/4. Contour interval is 10 kg (m s)−1.

Fig. 9.

Vertically integrated meridional moisture flux at (a) 0000, (b) 0600, (c) 1200, and (d) 1800 UTC for wet events expressed as the departure from the daily mean, (000 + 0600 + 1200 + 1800 UTC)/4. Contour interval is 10 kg (m s)−1.

4. Large-scale circulation anomalies during wet and dry events

The 200-hPa streamfunction, zonal wind and wind anomalies averaged over wet and dry events are displayed in Figs. 10 and 11, respectively. Areas where the streamfunction anomalies are significant at the 95% level are shaded. For the wind anomaly composites, only anomalies significant at the 95% level are shown. Large anomalies are located in the Western Hemisphere as indicated by the 200-hPa streamfunction anomalies with zonal means removed (Figs. 10a and 11a). Figure 10a shows a mostly longitudinal wave train extending from about 140°E in the subtropics to midlatitudes through the Gulf of Alaska and North America to the Atlantic for the wet composite. The largest amplitudes are found at about 45°N over the United States, with central values separated by roughly 30° longitude. Heavy rainfall is located to the east of the trough. A weaker “wave-train- like” feature extends from the equator about 140°W northeastward through the subtropical eastern Pacific to the western United States. It merges with the longitudinal wave train over the western half of the United States. A similar type of northeast-oriented wave pattern originating from 170°W near the equator is found for the dry case (Fig. 11a), but it is nearly in quadrature with the wave train for the wet case. The high pressure is found over North America. We do not offer explanations for these patterns, which are a subject of our ongoing investigations.

Fig. 10.

(a) Composite of anomalous 200-hPa streamfunction for wet events. Zonal mean at each latitude is removed. Contour interval is 2 × 106 m2 s−1. Areas where anomalies are statistically significant at the 95% are shaded; (b) same as (a) but for 200-hPa zonal wind. Contour interval is 8 m s−1 and (c) same as (a) but for 200-hPa wind anomalies. The unit is m s−1. Anomalies not statistically significant at the 95% are omitted.

Fig. 10.

(a) Composite of anomalous 200-hPa streamfunction for wet events. Zonal mean at each latitude is removed. Contour interval is 2 × 106 m2 s−1. Areas where anomalies are statistically significant at the 95% are shaded; (b) same as (a) but for 200-hPa zonal wind. Contour interval is 8 m s−1 and (c) same as (a) but for 200-hPa wind anomalies. The unit is m s−1. Anomalies not statistically significant at the 95% are omitted.

Fig. 11.

Same as Fig. 10, but for dry events.

Fig. 11.

Same as Fig. 10, but for dry events.

During wet conditions, the Pacific jet is at its climatological position but there are large anomalies associated with the North American jet (Fig. 10b). The jet strengthens with the jet core located near 45°N, 80°W. The southward shift of the jet stream features a second maximum located over southern California. For dry events (Fig. 11b), the Pacific jet is at its climatological position, but the zonal winds over North America shift northward. The North American jet extends farther into the Atlantic with the jet core near 55°N, 60°W. Composites of wind anomalies (Figs. 10c and 11c) are consistent with the moisture flux differences (Fig. 5d). For wet events, increased southerlies are located between a cyclonic–anticyclonic dipole over the United States. The anticyclone located in the North Pacific is also present. The anticyclonic feature over the eastern Pacific near Central America is consistent with the 200-hPa streamfunction composite (Fig. 10a). For dry conditions, decreased southerlies are located between an anticyclonic–cyclonic dipole over North America with another anticyclonic center located over the North Atlantic.

Vertical profiles of height anomalies for extreme events (Fig. 12) show the largest anomalies in the Western Hemisphere. The height anomalies increase with altitude throughout the troposphere. The profiles exhibit a slight vertical tilt of about 10° longitude within the lowest 400 hPa, but overall an equivalent barotropic structure prevails. Note that these anomalies are embedded within the westerlies. For the equivalent barotropic vertical structure to be maintained during an average event (about 17 days), there needs to be an approximate balance between the time-averaged vorticity advection and divergence fields, the leading terms in the vorticity equation for the scales of interest. We will return to this point in the following section.

Fig. 12.

Composite of the vertical profile of height anomalies averaged over 40°–50°N for (a) wet and (b) dry events. Contour interval is 10 m.

Fig. 12.

Composite of the vertical profile of height anomalies averaged over 40°–50°N for (a) wet and (b) dry events. Contour interval is 10 m.

5. Evolution of wet and dry events

Lagged composites of Outgoing Longwave Radiation Anomaly (OLRA), rainfall, 200-hPa streamfunction, zonal wind, divergence, divergent wind and moisture flux, and flux divergence anomalies were computed from the pentad centered at day −20 to the pentad centered at day +10 for both wet and dry events. Composites were made for bandpass-filtered kinetic energy per unit mass at 200 hPa and bandpass-filtered E–P fluxes and total E–P flux divergence with both horizontal and vertical components included. For bandpass- filtered kinetic energy, E–P fluxes, and rainfall, the climatological mean for the summer months was removed from the composites to obtain anomalies. Statistical significance is assessed by assuming a normal distribution for each variable except rainfall. A decorrelation time of 10 days is used to estimate the degrees of freedom. For rainfall anomalies, the statistical significance is assessed after the square root transformation. To examine whether the 1993 and 1988 events dominate the composites, we also produced composites without these events. We found that the patterns are generally the same and our conclusions do not change.

a. Wet events

Figure 13 plots time–longitude cross sections of composite mean 200-hPa zonal wind (Fig. 13a) and streamfunction (Fig. 13b) anomalies in midlatitudes from day −20 to day +10 to show the evolution of wet events. Areas where values are statistically significant at the 95% level are shaded. The vertically integrated meridional flux (Fig. 13c) and flux divergence D(Q) anomalies (Fig. 13d) over the United States are also given. Prior to the 1993 floods, intensified synoptic-scale eddy activity was observed in the Pacific North American area roughly 15–20 days prior to the onset of flooding. The eddy activity was associated with a southward displacement of the jet (Mo et al. 1995). The synoptic-scale eddy forcing is also important here. The Pacific subtropical jet starts to shift southward and extends to the west coast of the United States at about day −18 (Fig. 13a). The zonal mean wind anomalies reach a maximum at day −13. The composite averaged from day −15 to day −10 (Fig. 14b) shows positive zonal wind anomalies at 30°–35°N extending from 140°W to California and negative anomalies located to the north of the positive anomalies. Bell and Janowiak (1995) suggested that the jet shift observed during the 1993 floods was a response to the 1992 warm ENSO episode. Examination of individual events indicates that only 9 out of 21 wet events occurred during warm ENSO years. During these episodes, tropical convection excites a Rossby wave with enhanced westerlies extending eastward in the subtropics. Nevertheless, floods also occurred during non-ENSO events suggesting that ENSO is not the sole mechanism responsible for shifts of the jet. Another possibility is synoptic-scale eddy forcing. The E–P flux divergence (Fig. 14c) averaged over the same period shows divergence along 35°N and convergence to the north. This north–south dipole is consistent with eddy forcing accelerating westerly winds to the south and weakening them to the north. Here, we do not attempt to establish the cause for the synoptic-scale eddy enhancement, but rather point out that the jet displacement is consistent with the diagnosed eddy activity.

Fig. 13.

(a) Time–longitude cross section for 200-hPa zonal mean wind anomalies averaged from 30° to 40°N from day −20 to day +10 for wet events. Contour interval is 2 m s−1. Contours 1 m s−1 and −1 m s−1 are added. Zero contours are omitted. Areas where anomalies are statistically significant at the 95% level are shaded; (b) same as (a) but for anomalous 200-hPa streamfunction averaged from 40° to 50°N. Contour interval is 2 × 106 m2 s−1; (c) same as (a) but for vertically integrated meridional moisture flux averaged from 30° to 35°N. Contour interval is 30 kg (m s)−1. Positive values are shaded, zero contours are omitted, and (d) same as (c) but for anomalous vertically integrated moisture flux divergence D(Q) averaged from 30° to 40°N. Contour interval is 1 mm day−1 and contours 0.5 and −0.5 mm day−1 are added.

Fig. 13.

(a) Time–longitude cross section for 200-hPa zonal mean wind anomalies averaged from 30° to 40°N from day −20 to day +10 for wet events. Contour interval is 2 m s−1. Contours 1 m s−1 and −1 m s−1 are added. Zero contours are omitted. Areas where anomalies are statistically significant at the 95% level are shaded; (b) same as (a) but for anomalous 200-hPa streamfunction averaged from 40° to 50°N. Contour interval is 2 × 106 m2 s−1; (c) same as (a) but for vertically integrated meridional moisture flux averaged from 30° to 35°N. Contour interval is 30 kg (m s)−1. Positive values are shaded, zero contours are omitted, and (d) same as (c) but for anomalous vertically integrated moisture flux divergence D(Q) averaged from 30° to 40°N. Contour interval is 1 mm day−1 and contours 0.5 and −0.5 mm day−1 are added.

Fig. 14.

(a) Composite of OLRA averaged from day −15 to day −10 for wet events. Contour interval is 3 W m−2. Zero contours are omitted. Areas where anomalies are statistically significant at the 95% level are shaded; (b) same as (a) but for 200-hPa zonal wind anomalies. Contour interval is 1 m s−1; (c) same as (a) but for the E–P flux divergence. Contour is 0.3 m s−1 day−1. Values greater than 0.3 (less that −0.3) are light (dark) shaded, and (d) same as (a) but for 200-hPa streamfunction anomalies. Contour interval 1 × 106 m2 s−1.

Fig. 14.

(a) Composite of OLRA averaged from day −15 to day −10 for wet events. Contour interval is 3 W m−2. Zero contours are omitted. Areas where anomalies are statistically significant at the 95% level are shaded; (b) same as (a) but for 200-hPa zonal wind anomalies. Contour interval is 1 m s−1; (c) same as (a) but for the E–P flux divergence. Contour is 0.3 m s−1 day−1. Values greater than 0.3 (less that −0.3) are light (dark) shaded, and (d) same as (a) but for 200-hPa streamfunction anomalies. Contour interval 1 × 106 m2 s−1.

During this period, there is increased rainfall and enhanced moisture flux convergence over the central United States (Fig. 13d), suggesting that enhanced synoptic- scale eddy also brings rainfall (negative OLRA, Fig. 14a). The negative streamfunction anomalies located in the central United States (Figs. 13b and 14d) are consistent with rainfall activity. Rainfall increases soil moisture in this area, providing positive feedback toward more rainfall (Mintz 1984; Oglesby and Erickson 1989 and others). That may create favorable conditions for persistent wet conditions to occur later.

From day −10 to day −1, the rain belt extends westward to Texas as indicated by moisture flux convergence (Fig. 13d) and OLRA (Figs. 15a and 15c). The zonal wind anomalies are weaker (Fig. 13a), but the jet still extends across the United States and is located south of its climatological position (Fig. 15b). Composites of OLRA (Fig. 15c), moisture flux convergence (Fig. 16a), and precipitation (Fig. 16b) anomalies averaged for the pentad before onset (day −5 to day −1) all show more rainfall over Texas accompanied by dryness over the eastern United States. In the Tropics, negative OLRAs (Fig. 15c) extend from 120°W, 10°N through Central America to the Atlantic. Enhanced convection can also be found at 160°E along the equator. The E–P flux divergence is weak over the continent (not shown).

Fig. 15.

(a) Composite of OLRA averaged from day −10 to day −6 for wet events. Contour interval is 3 W m−2. Zero contours are omitted. Areas where anomalies are statistically significant at the 95% level are shaded; (b) same as (a) but for 200-hPa zonal wind anomalies. Contour interval is 1 m s−1; (c) same as (a) but for OLRA averaged from day −5 to day −1 and (d) same as (b) but for 200-hPa streamfunction anomalies averaged from day −5 to day −1. Contour interval 1 × 106 m2 s−1.

Fig. 15.

(a) Composite of OLRA averaged from day −10 to day −6 for wet events. Contour interval is 3 W m−2. Zero contours are omitted. Areas where anomalies are statistically significant at the 95% level are shaded; (b) same as (a) but for 200-hPa zonal wind anomalies. Contour interval is 1 m s−1; (c) same as (a) but for OLRA averaged from day −5 to day −1 and (d) same as (b) but for 200-hPa streamfunction anomalies averaged from day −5 to day −1. Contour interval 1 × 106 m2 s−1.

Fig. 16.

(a) Composite of D(Q) and vertically integrated moisture flux averaged from day −5 to day −1 for wet events. Contour interval for D(Q) 0.5 mm day−1. Values greater than 0.5 mm day−1 (less than −0.5 mm day−1) are dark (light) shaded. The unit for flux is 10 kg (m s)−1; (b) same as (a) but for precipitation from the NCEP reanalysis 6-h forecasts. Contour interval is 0.5 mm day−1. Zero contours are omitted. Areas where anomalies are statistically significant at the 95% level are shaded, and (c) same as (a) but for 200-hPa divergence and divergent winds. Contour interval is 0.5 × 10−6 s−1 and the unit for wind is 1 m s−1.

Fig. 16.

(a) Composite of D(Q) and vertically integrated moisture flux averaged from day −5 to day −1 for wet events. Contour interval for D(Q) 0.5 mm day−1. Values greater than 0.5 mm day−1 (less than −0.5 mm day−1) are dark (light) shaded. The unit for flux is 10 kg (m s)−1; (b) same as (a) but for precipitation from the NCEP reanalysis 6-h forecasts. Contour interval is 0.5 mm day−1. Zero contours are omitted. Areas where anomalies are statistically significant at the 95% level are shaded, and (c) same as (a) but for 200-hPa divergence and divergent winds. Contour interval is 0.5 × 10−6 s−1 and the unit for wind is 1 m s−1.

At the same time, the 200-hPa streamfunction anomaly composite (Fig. 15d) shows a wave train extending from the area of tropical negative OLRAs (Fig. 15c) through the North Pacific and the Gulf of Alaska to North America similar to the composite over the duration of wet events (Fig. 10a), with stronger zonal wind anomalies centered at 120°W (Fig. 13a). The ridge along the east coast of the United States deepens (Fig. 15d). A trough located over the western United States near 110°W starts to appear at day −3. At day −2, negative anomalies at that location become statistically significant at the 95% level (Fig. 13b). Conservation of potential vorticity requires that a trough be formed in the lee of the mountains in the presence of westerly upstream conditions.

Recall that the vertical profile of height anomalies (Fig. 12a) during persistent wet events shows an equivalent barotropic structure. For the equivalent barotropic vertical structure to be maintained during wet episodes, the time-averaged vorticity advection term should be balanced by the divergence field. At about 100°W, just east of the low pressure centers at upper levels, there is positive vorticity advection (not shown). The balance requires divergence (Fig. 16c) at upper levels for the wet case at this location. The opposite divergence pattern is expected at low levels by mass continuity (Fig. 16c). The Rockies block the low-level flow from the west, and thus the required convergence field for the wet case results in an enhancement of the low-level southerlies (Fig. 16a). This is consistent with D(Q) (Fig. 16a), which shows moisture flux convergence in the central United States since the contributions to the D(Q) are mainly from the low-level winds rather than the specific humidity (Wang and Paegle 1996). Other physical processes are also important at low levels, which are not included in this simple argument. For example, Zhong et al. (1996) and others suggest the importance of inertial oscillations for LLJ dynamics.

After onset, the trough–ridge pair over the United States persists and the LLJ continues to strengthen as indicated by the meridional moisture transport (Fig. 13c). The moisture flux and moisture flux convergence reach maximum values at day +3 (Fig. 13d). The enhanced LLJ provides additional moisture influx to the central United States. The combination of wet soil and increased moisture influx allows wet conditions to persist.

b. Dry events

For dry events, circulation anomalies do not develop until 10 days before onset. Figure 17 shows time–longitude cross sections for OLRA, vertically integrated meridional flux, and 200-hPa streamfunction anomalies from day −10 to day +10 for the evolution of dry events. Two weeks prior to onset of dry events, negative OLRAs appear over the eastern Pacific near Central America and continue to strengthen (Fig. 17a). The negative OLRAs are accompanied by positive OLRAs over the Gulf of Mexico (Fig. 17b). The OLRAs (Fig. 18a) averaged over the pentad before onset (day −5 to day −1) show negative anomalies just north of the equator from the date line through the eastern Pacific to Central America. These anomalies signal a northward shift of the ITCZ, which has been related to the 1988 drought by Trenberth and Branstator (1992). At the same time, positive OLRAs are located over the Gulf of Mexico, the southeastern United States, and northeastern South America.

Fig. 17.

(a) Time–longitude cross section for OLRA composite averaged from the equator to 10°N from day −10 to day +10 for dry events. Contour interval is 3 W m−2. Negative values are shaded and zero contours are omitted; (b) same as (a) but for OLRA averaged from 20° to 25°N; (c) same as (a) but for vertically integrated meridional moisture flux averaged from 30° to 35°N. Contour interval is 20 kg (m s)−1. Positive values are shaded. (d) Same as (a) but for 200-hPa streamfunction anomalies averaged from 40° to 50°N. Contour interval is 2 × 106 m2 s−1. Positive values are shaded.

Fig. 17.

(a) Time–longitude cross section for OLRA composite averaged from the equator to 10°N from day −10 to day +10 for dry events. Contour interval is 3 W m−2. Negative values are shaded and zero contours are omitted; (b) same as (a) but for OLRA averaged from 20° to 25°N; (c) same as (a) but for vertically integrated meridional moisture flux averaged from 30° to 35°N. Contour interval is 20 kg (m s)−1. Positive values are shaded. (d) Same as (a) but for 200-hPa streamfunction anomalies averaged from 40° to 50°N. Contour interval is 2 × 106 m2 s−1. Positive values are shaded.

Fig. 18.

(a) Composite of OLRA averaged from day −5 to day −1 for dry events. Contour interval is 3 W m−2. Zero contours are omitted. Areas where anomalies are statistically significant at the 95% level are shaded. (b) Same as (a) but for 200-hPa streamfunction anomalies. Contour interval is 1 × 106 m2 s−1, (c) Same as (a) but for composite of anomalous 200-hPa divergence and divergent wind anomalies. Contour interval for divergence is 0.5 × 10−6 s−1. Values greater than 0.5 × 10−6 s−1 (less than − 0.5 × 10−6 s−1) are light (dark) shaded. Zero contours are omitted. The unit for divergent wind anomalies is 1 m s−1. (d) Same as (a) but for 200-hPa synoptic-scale eddy kinetic energy per mass. Contour interval is 3 (m s−1)2. Zero contours are omitted. (e) Same as (a) but for vertical velocity. Contour interval 0.008 Pa s−1; (f) same as (a) but for D(Q) and vertically integrated moisture flux. Contour interval for D(Q) is 0.5 mm day−1.

Fig. 18.

(a) Composite of OLRA averaged from day −5 to day −1 for dry events. Contour interval is 3 W m−2. Zero contours are omitted. Areas where anomalies are statistically significant at the 95% level are shaded. (b) Same as (a) but for 200-hPa streamfunction anomalies. Contour interval is 1 × 106 m2 s−1, (c) Same as (a) but for composite of anomalous 200-hPa divergence and divergent wind anomalies. Contour interval for divergence is 0.5 × 10−6 s−1. Values greater than 0.5 × 10−6 s−1 (less than − 0.5 × 10−6 s−1) are light (dark) shaded. Zero contours are omitted. The unit for divergent wind anomalies is 1 m s−1. (d) Same as (a) but for 200-hPa synoptic-scale eddy kinetic energy per mass. Contour interval is 3 (m s−1)2. Zero contours are omitted. (e) Same as (a) but for vertical velocity. Contour interval 0.008 Pa s−1; (f) same as (a) but for D(Q) and vertically integrated moisture flux. Contour interval for D(Q) is 0.5 mm day−1.

Examination of the OLRAs averaged over the duration of each event indicates that the OLRA composite (Fig. 18a) is robust with 16 events out of 19 events showing negative anomalies in the eastern Pacific and positive anomalies over the Gulf of Mexico. The OLRA composite is consistent with the 200-hPa divergence anomaly composite (Fig. 18c). Positive anomalies (enhanced divergence) are collocated with negative OLRAs over the eastern Pacific suggesting heating related to the shift of the ITCZ. This heating produces an enhanced local Hadley circulation as indicated by the divergent wind anomalies. The wind anomalies extend from this center northeastward to the Gulf of Mexico and the eastern United States; these areas lie under the subsiding branch of the Hadley cell as indicated by convergence (positive OLRA).

This picture is consistent with a composite of vertical velocity anomalies (Fig. 18e), which indicates rising motion in the area of negative OLRAs (divergence) and sinking motion in the area of positive OLRAs (convergence). The sinking motion is collocated with moisture flux divergence (Fig. 18f). The fluxes turn southward through the gulf to the eastern Pacific and the moisture convergence further enhances precipitation (negative OLRAs) in that area (Fig. 18f). The LLJ weakens and the moisture transport through the gulf and into the central United States diminishes, promoting dry conditions over the central United States. The synoptic-scale eddies do not play a role before onset since the eddy activity as measured by the 200-hPa kinetic energy per mass is weak over the United States and negative anomalies are located in the Pacific and in the Atlantic (Fig. 18d).

Fig. 18.

(Continued)

Fig. 18.

(Continued)

The ridge over the central United States (Fig. 11a) is not established until after onset. The composite of 200- hPa streamfunction anomalies before onset (Fig. 18b) shows negative anomalies along the west coast of the United States and over Central America. Over the United States, anomalies are weak and statistically insignificant. After onset, a strong ridge becomes established over the central United States (not shown), and a pattern consistent with the composite average of Fig. 11a emerges.

6. Summary and discussion

We examined the evolution of circulation patterns and moisture transport during wet and dry extreme precipitation episodes over the central United States. Persistent wet and dry events were chosen based on observed station precipitation. After determining events, composites averaged over all wet (dry) events were made to study the evolution and circulation features during persistent rainfall patterns. During summer, there is an inverse relationship between rainfall in the central United States and rainfall in the southeastern United States and Mexico as well as along the western Canadian coast. This rainfall pattern is supported by circulation anomalies. For wet (dry) events, increased (decreased) northward meridional wind anomalies are located between a cyclonic–anticyclonic (anticyclonic–cyclonic) dipole over the United States. However, wet and dry events are not mirror images of each other and the physical mechanisms responsible for wet and dry events are very different. The vertical profiles of height anomalies indicate that anomalies are concentrated in the Western Hemisphere and height anomalies increase with altitude with maxima at 200-hPa. Overall, the profile exhibits an equivalent barotropic vertical structure.

During the 1993 floods, the North American jet was located south of its climatological location and unusually strong eddy activity was found prior to onset (Mo et al. 1995). These features appear in the composites of flood episodes. Strong eddy kinetic energy anomalies are found two pentads before onset. The total E–P flux divergence indicates that synoptic-scale eddies act to enhance the westerly flow over the United States consistent with the strengthening of the North American jet across the United States. The eddy activity is also consistent with an increase of soil moisture, which is an important aspect of wet episodes. Among the global- scale anomalies preceding the wet events, the enhanced eddies and accelerated zonal flow appear to be the most significant precursors on a weekly timescale.

One pentad before onset, rainfall begins in Texas and the southerly transport strengthens. Meanwhile, the North American jet shifts southward. Mo et al. (1995) used a simple shallow-water model to demonstrate that a lee trough can be forced by the Rocky Mountains in the presence of the strong westerly flow. The positive feedback between the persistent trough and the LLJ increases the strength of the LLJ. In the Tropics, negative OLRAs are observed near the date line and over Central America. Soon afterward, a wave train from the Tropics to the Gulf of Alaska starts to develop. It is possible that the wave train is excited by enhanced convection in the Tropics. The increased moisture transport by the LLJ, wet soil, and the dynamical support by the wave train work together to sustain the wet events.

In contrast to wet events, synoptic-scale eddy activity is weak before the onset of dry episodes. Before onset, negative OLRAs appear in the eastern Pacific and over Central America as a result of the northward shift of the ITCZ. This pattern appears to be the most robust precursor of dry events. The subsidence of the enhanced Hadley cell generated by heating is located over the Gulf of Mexico and the southeastern United States. The convergence increases the southern branch of moisture transport to Mexico and the eastern Pacific. The LLJ diminishes and less moisture is transported to the central United States.

This study confirms the linkages between the LLJ and precipitation in the central United States. We found that precipitation during floods is largely nocturnal. Our findings are consistent with many studies based on GCM simulations (Helfand and Schubert 1995; Paegle et al 1996) and reanalyses (Higgins et al. 1997a), which show the influence of the LLJ on moisture transport and precipitation over the United States.

This study concentrates on extreme events on the timescales of 10 days to a season. Major drought episodes can last months or years such as the widespread droughts of the 1930s and 1950s. Over the Great Plains, the 1-yr lag correlation for seasonal mean temperature shows year-to-year persistence as demonstrated by Madden (1977). The conditions of soil moisture in spring may also influence the summer temperature or precipitation. These have not been considered in this study.

Acknowledgments

This investigation was supported by NOAA Grants GC95-013 and GC95-833 to the University of Utah, by Interagency Agreement S-41367-F under the authority of NASA/GSFC, and by the NOAA Office of Global Programs under the GEWEX Continental Scale International Project.

REFERENCES

REFERENCES
Beljaars, A. C. M., P. Viterbo, A. Betts, and M. J. Miller, 1996: The anomalous rainfall over the United States during July 1993: Sensitivity to land surface parameterization and soil moisture anomalies. Mon. Wea. Rev., 124, 362–383.
Bell, G. D., and J. E. Janowiak, 1995: Atmospheric circulation associated with the Midwest floods of 1993. Bull. Amer. Meteor. Soc., 76, 681–695.
Blackmon, M. L., 1976: A climatological spectral study of the 500-mb height of the Northern Hemisphere. J. Atmos. Sci., 33, 1607–1623.
Bonner, W. D., 1968: Climatology of the low level jet. Mon. Wea. Rev., 96, 833–850.
——, and J. Paegle, 1970: Diurnal variations in the boundary layer winds over the south central United States in summer. Mon. Wea. Rev., 98, 735–744.
Chang, F. C., and J. M. Wallace, 1987: Meteorological conditions during heat waves and droughts in the United States Great Plains. Mon. Wea. Rev., 115, 1253–1269.
Dole, R. M., and N. D. Gordon, 1983: Persistent anomalies of the extratropical Northern Hemisphere wintertime circulation: Geographical distribution and regional persistent characteristics. Mon. Wea. Rev., 111, 1567–1587.
Douglas, M. W., R. A. Maddox, K. Howard, and S. Reyes, 1993: The Mexican monsoon. J. Climate, 6, 1665–1677.
Epstein, E., 1988: A precipitation climatology of five day sequences. NOAA Tech. Rep. NWS41, National Weather Service, NOAA, 162 pp.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-year reanalysis project. Bull. Amer. Meteor. Soc., 77, 437–471.
Karl, T. R., and R. G. Quayle, 1981: The 1980 summer heat wave and drought in historical perspective. Mon. Wea. Rev., 109, 2055–2073.
Helfand, H. M., and S. D. Schubert, 1995: Climatology of the Great Plains low-level jet and its contribution to the continental moisture budget of the United States. J. Climate, 8, 784–806.
Higgins, R. W., J. E. Janowiak, and Y. Yao, 1996a: A gridded hourly precipitation data base for the United States (1963–1993). NCEP/Climate Prediction Center Atlas 1, NCEP/NWS/NOAA, 47 pp.
——, K. C. Mo, and S. D. Schubert, 1996b: The moisture budget of the central United States in spring as evaluated in the NCEP/NCAR and the NASA/DAO reanalyses. Mon. Wea. Rev., 124, 939–963.
——, Y. Yao, J. E. Janowiak, and K. C. Mo, 1997a: Influence of the Great Plains low-level jet on summertime precipitation and moisture transport over the central United States. J. Climate, 10, 481–507.
——, ——, and J. Wang, 1997b: Influence of the North American monsoon system on the United States summer precipitation regimes. J. Climate, 10, 2600–2622.
Hoskins, B. J., I. N. James, and G. H. White, 1983: The shape, propagation and mean flow interaction of large-scale weather systems. J. Atmos. Sci., 40, 1595–1612.
Lanzante, J. R., and R. P. Harnack, 1982: Specification of United States summer season precipitation. Mon. Wea. Rev., 110, 1843–1850.
Lau, K. M., and L. Peng, 1992: Dynamics of atmospheric teleconnections during the northern summer. J. Climate, 5, 146–158.
Liebmann, B., and C. A. Smith, 1996: Description of a complete interpolated outgoing longwave radiation dataset. Bull. Amer. Meteor. Soc., 77, 1275–1277.
Lyon, B., 1991: An observational study of persistent temperature anomalies over the North American region. Ph.D. thesis, Massachusetts Institute of Technology, 172 pp.
Madden, R. A., 1977: Estimates of the autocorrelations and spectra of seasonal mean temperatures over North America. Mon. Wea. Rev., 105, 9–18.
——, and J. Williams, 1978: The correlation between temperature and precipitation in the United States and Europe. Mon. Wea. Rev., 106, 142–147.
McCorcle, M. D., 1986: The effect of soil moisture and ambient conditions of the Great Plains boundary layer. Ph.D. dissertation, University of Utah, 161 pp.
Mintz, Y., 1984: The sensitivity of numerically simulated climates to land surface boundary conditions. The Global Climate, J. Houghton, Ed., Cambridge University Press, 79–105.
Mo, K. C., and R. W. Higgins, 1996: Large-scale atmospheric water vapor transport as evaluated from the NCEP/NCAR and the NASA/DAO reanalyses. J. Climate, 9, 1531–1545.
——, J. R. Zimmerman, E. Kalnay, and M. Kanamitsu, 1991: A GCM study of the 1988 United States drought. Mon. Wea. Rev., 119, 1512–1532.
——, J. N. Paegle, and J. Paegle 1995: Physical mechanisms of the 1993 summer floods. J. Atmos. Sci., 52, 879–895.
Namias, J., 1989: Cold waters and hot summers. Nature, 338, 15–16.
Oglesby, R. J., and D. J. Erickson, 1989: Soil moisture and the persistence of North American drought. J. Climate, 2, 1362–1380.
Paegle, J., K. C. Mo, and J. N. Paegle, 1996: Dependence of simulated precipitation on surface evaporation during the 1993 United States summer floods. Mon. Wea. Rev., 124, 345–361.
Palmer, T. N., and C. Brankovic, 1989: The 1988 United States drought linked to anomalous sea surface temperature. Nature, 338, 54–57.
Richman, M. B., and P. J. Lamb, 1985: Climate pattern analysis of three and seven day summer rainfall in the central United. States:Some methodological considerations and a regionalization. J. Climate Appl. Meteor., 24, 1325–1343.
Trenberth, K. E., 1986: An assessment of the impact of transient eddies on the zonal flow during a blocking episode using localized Eliassen–Palm diagnostics. J. Atmos. Sci., 43, 2070–2087.
——, and G. W. Branstator, 1992: Issues in establishing causes of the 1988 drought over North America. J. Climate, 5, 159–172.
——, ——, and P. A. Arkin, 1988: Origins of the 1988 North American drought. Science, 242, 1640–1645.
Walsh, J. E., M. B. Richman, and D. W. Allen, 1982: Spatial coherence of monthly precipitation in the United States. Mon. Wea. Rev., 110, 272–285.
Wang, M., and J. Paegle, 1996: Impact of analysis uncertainty upon regional atmospheric moisture flux. J. Geophy. Res., 101, 7291–7303.
Wolfson, N., R. Atlas, and Y. C. Sud, 1987: Numerical experiments related to the summer 1980 heat wave. Mon. Wea. Rev., 115, 1345–1357.
Zhong, S., J. D. Fast, and X Bian, 1996: A case study of the Great Plains low-level jet using wind profiler network data and a high- resolution mesoscale model. Mon. Wea. Rev., 124, 785–806.

Footnotes

Corresponding author address: Kingtse C. Mo, Climate Prediction Center, NCEP/NWS/NOAA, NP52, 4700 Silver Hill Rd., Stop 9910, Washington, DC 20233. E-mail: wd52km@sqi44.wwb.noaa.gov