Abstract

An analysis of extreme daily precipitation events that occurred in the south-central United States during May and June 2010 is carried out using gridded station data and reanalysis products in use at the National Centers for Environmental Prediction (NCEP). Various aspects of the daily extremes are examined from a climate perspective using a 62-yr (1948–2010) period of record, including their historical ranking, common circulation features, moisture plumes, and the possible influence of ENSO. The analysis also considers how the frequency and intensity of daily extremes is changing in the United States. Each of the 2010 flash flood events examined here was associated with historic daily rainfall totals. Several of the events had meteorological conditions in common at upper and lower levels of the atmosphere, and all of the events fit well into an existing classification scheme for heavy precipitation events associated with flash flooding. Each case exhibited characteristics of the “Maya Express” flood events that link tropical moisture plumes from the Caribbean and Gulf of Mexico to midlatitude flooding over the central United States. Consistent with recent assessment reports, it is shown that extreme daily precipitation events in the United States have increased in frequency during the most recent 30-yr period (1980–2009) when compared to the previous 30-yr period (1950–79), though the increases are relatively small during May and June.

1. Introduction

The period May–June of 2010 was characterized by a large number of localized heavy rain events leading to flash flooding across portions of the south-central United States. For example, on 3 May a heavy precipitation event in Nashville, Tennessee, led to flooding that killed 31 people, which was the highest death toll from a nontropical cyclone flooding event in the United States since 1994. On 11 June a heavy precipitation event that occurred in just a few hours over mountainous terrain in western Arkansas led to the disastrous flash flood in Albert Pike Recreation Area, which killed 20 people. The period 14–15 June 2010 was the rainiest 2-day period in history for Oklahoma City, Oklahoma, which experienced heavy flooding and extensive damage, with rainfall totals exceeding 10 in. in some areas. Other historic extreme events were observed during May and June 2010 in locations that included eastern Texas and south-central Kansas. Subsequent to these extreme events, heavy rains continued across portions of the upper Midwest and northern Great Plains during July and August, including Iowa, Illinois, South Dakota, and Minnesota.

With the large number of heavy precipitation events leading to major flooding during the spring of 2010, the National Oceanic and Atmospheric Administration/National Weather Service (NOAA/NWS) received a “flood” of questions asking whether the events were related to climate change. Such questions are consistent with recent assessment reports (e.g., Karl et al. 2009), that have concluded that the amount of precipitation falling in the heaviest downpours in the United States has increased approximately 20% on average in the past century, and that this trend is very likely to continue, with the largest increases in the wettest places. Karl et al. (2009) goes on to say that widespread impacts (e.g., to the water, energy, transportation, agriculture, ecosystems, and health sectors) are due to changes in the frequency and intensity of heavy precipitation events that are occurring now and that are expected to increase.

The NWS is responsible for providing early warning of weather and hydrologic extremes, as well as accurate information with minimal uncertainty on their possible causes. Improved attribution of the causes of these events supports the NWS mission requirements to help the public prepare for and respond to the associated flooding threats.

This paper addresses some of the questions people are asking about the May and June 2010 precipitation and flooding events, but from a climate perspective:

  • How did the May and June 2010 precipitation events rank in the historical record?

  • Were there circulation features in common to these events?

  • Were these events influenced by ENSO?

  • How is the frequency and intensity of precipitation extremes changing?

The questions above are examined using operational analyses produced by the National Centers for Environmental Prediction (NCEP), including the Climate Prediction Center (CPC) unified global daily gauge analysis (1948–2010, available online at http://www.cpc.ncep.noaa.gov/products/precip/CWlink/MJO/enso.shtml), and the NCEP–Department of Energy (DOE) reanalysis (Kanamitsu et al. 2002). This study is focused on the precipitation extremes that occurred during May and June 2010, and not on the very wet conditions and more generalized flooding that occurred in the upper Midwest and northern Great Plains during the summer of 2010.

A brief summary of the datasets and analysis procedures (section 2) is followed by an examination of the historical rankings of the May and June 2010 extreme precipitation events (section 3). Circulation patterns and moisture plumes associated with the events are examined in section 4. Relationships to the ENSO cycle and long-term trends in precipitation extremes are discussed in sections 5 and 6, respectively. A summary and plans for future studies are discussed in section 7.

2. Datasets and analysis procedure

a. Observed precipitation

The observed daily precipitation analysis was obtained from the CPC unified rain gauge database (Chen et al. 2008; Higgins et al. 2000). The database averages roughly 17 000 daily station reports around the globe, with excellent coverage over the United States (roughly 10 000 daily station reports). The database was used to produce a multiyear (1948–2010) daily precipitation analysis over the continental United States. The analysis is on a (latitude, longitude) = (0.125°, 0.125°) grid (approximately 14 km) and was produced using an optimal interpolation (OI) scheme. Precipitation was accumulated over a 24-h period ending at 1200 UTC on the calendar date. Several types of quality control were applied including “duplicate station” and “buddy” checks among others.

Previous assessments of objective techniques for gauge-based analyses of daily precipitation (e.g., Chen et al. 2008) have shown that OI-based schemes are among the best over the complex terrain of the western United States, though it is important to note that the particular choice of analysis scheme is a source of uncertainty. Relationships between the spatial distribution and temporal continuity of the station data and errors in the final gridded OI analysis were examined in Chen et al. (2008). The selection of long time series for individual stations is less important than ensuring sufficient data coverage and application of state-of-the art quality control and analysis procedures. Earlier comparisons of gridded analyses and station observations over the United States (e.g., Higgins et al. 2007) have shown that gridded analyses are reliable for studies of fluctuations in daily precipitation as long as the station coverage is sufficiently dense and state-of-the-art analysis techniques and quality control procedures are applied to the daily data. Nevertheless, it is important to note that the station density and changes in the station density over time are sources of uncertainty in the analysis.

For the nationwide analyses in sections 5 and 6 a mask has been applied to locations where the daily precipitation extremes do not exceed a threshold for the mean (1950–2009) daily precipitation of 0.25 mm. This accounts for months and regions (especially in the western United States) where the number of daily precipitation events is low.

b. NCEP–DOE reanalysis

Circulation features associated with the May and June 2010 precipitation extremes were examined using the NCEP–DOE reanalysis (hereafter referred to as R2, Kanamitsu et al. 2002). The R2 used a very similar analysis system to the NCEP–NCAR reanalysis (Kalnay et al. 1996) and an upgraded version of the same general circulation model, with known errors fixed and assimilation of a more complete stream of observational data after 1993. The R2 was also updated in real time, spanning the period from 1979 to the present.

In this study we examine the following fields: 500-hPa height, 500-hPa vector wind, 500-hPa vertical motion, 925-hPa vector wind, sea level pressure, and precipitable water. (The results shown in Figs. 45 are based on analyses at 0000 UTC on the calendar dates indicated on the figures.)

c. Data processing

Historical rankings for each event were obtained using the observed daily precipitation analysis from the CPC unified rain gauge database (see section 2a). Daily rainfall for the 15-day period centered on the target date (±7 days) were ranked for each (latitude, longitude) = (0.125°, 0.125°) grid box for the 62-yr period (1948–2010). Rankings for the May and June 2010 heavy precipitation events were obtained using the grid box surrounding the target location.

Brooks and Stensrud (2000) proposed a technique to construct a gridded analysis for extreme precipitation event frequency by defining the frequency as a logarithmic function of intensity and by interpolating the statistics at station locations. While a comparison of the performance of this technique to other methods is an important research topic, in this study we examine the extreme precipitation events directly using the analyzed fields of daily precipitation. Although interpolation of station data may damp the magnitude of large precipitation events at individual locations, this smoothing effect is small over regions with a dense gauge network, including the central and eastern United States.

A significance test is applied for the national results in section 5 and 6 (see Figs. 6 and 8). The significance levels are based on 1000 Monte Carlo simulations. For the significance test (used in Fig. 6), the number of random numbers in each set was determined by the number of observed cases with neutral, El Niño, and La Niña conditions during each season. The random number sets were used to create random samples for the number of top 150 events in each case. (A similar test was developed for Fig. 8, but random samples were created for the top 50 events. The contours on Figs. 6 and 8 indicate where the departures from the expected number of events are significant at the 95% level.)

Other tests performed in this study involve counting extremes at various daily precipitation thresholds over particular regions. The details of these procedures are discussed below as they are used.

3. Historical ranking

Daily precipitation totals for five May and June 2010 extreme events, each of which ranked first in the historical record at the location of interest are examined. The rankings for each event were derived from the CPC daily gauge precipitation analysis (1948–2010) and are based on 24-h precipitation totals ending at 1200 UTC on the calendar date of interest. In each case, the daily precipitation pattern is shown (Fig. 1) as well as the percentile rankings for daily precipitation amounts in the grid box surrounding the location of interest (Fig. 2). A summary of the results for all five events is given in Table 1.

Fig. 1.

Daily precipitation (mm) for selected extreme events during May and June 2010: (a) 3 May 2010, (b) 10 Jun 2010, (c) 11 Jun 2010, (d) 13 Jun 2010, and (e) 15 Jun 2010. Precipitation totals are for the 24-h period ending at 1200 UTC on the calendar dates indicated.

Fig. 1.

Daily precipitation (mm) for selected extreme events during May and June 2010: (a) 3 May 2010, (b) 10 Jun 2010, (c) 11 Jun 2010, (d) 13 Jun 2010, and (e) 15 Jun 2010. Precipitation totals are for the 24-h period ending at 1200 UTC on the calendar dates indicated.

Fig. 2.

Daily precipitation percentiles for days corresponding to the daily extremes in Fig. 1. (a) 3 May 2010 for the grid box surrounding Nashville (b) 10 Jun 2010 for the grid box with maximum precipitation over eastern Texas, (c) 11 Jun 2010 for the grid box with maximum precipitation over western Arkansas, (d) 13 Jun 2010 for the grid box with maximum precipitation over south-central Kansas, and (e) 15 Jun 2010 for the grid box with maximum precipitation over central Oklahoma. For reference, precipitation amounts for 99th percentile rankings are indicated by the horizontal dashed lines. Precipitation totals are for the 24-h period ending at 1200 UTC on the calendar dates indicated.

Fig. 2.

Daily precipitation percentiles for days corresponding to the daily extremes in Fig. 1. (a) 3 May 2010 for the grid box surrounding Nashville (b) 10 Jun 2010 for the grid box with maximum precipitation over eastern Texas, (c) 11 Jun 2010 for the grid box with maximum precipitation over western Arkansas, (d) 13 Jun 2010 for the grid box with maximum precipitation over south-central Kansas, and (e) 15 Jun 2010 for the grid box with maximum precipitation over central Oklahoma. For reference, precipitation amounts for 99th percentile rankings are indicated by the horizontal dashed lines. Precipitation totals are for the 24-h period ending at 1200 UTC on the calendar dates indicated.

Table 1.

Dates, daily rainfall amounts, and historical ranking for five extreme daily precipitation events during May and June 2010.

Dates, daily rainfall amounts, and historical ranking for five extreme daily precipitation events during May and June 2010.
Dates, daily rainfall amounts, and historical ranking for five extreme daily precipitation events during May and June 2010.
  • Nashville, Tennessee: The major flooding event at Nashville, Tennessee, occurred on 3 May 2010 (Fig. 1a). Daily rainfall on 3 May, for the grid box surrounding Nashville was 184.0 mm, which ranked first in the historical record (Fig. 2a).

  • Eastern Texas: The major flooding event in eastern Texas occurred on 10 June 2010 (Fig. 1b). Daily rainfall on 10 June, for the grid box with the maximum rainfall in eastern Texas was 216.4 mm, which ranked first in the historical record (Fig. 2b).

  • Western Arkansas: The major flooding event in western Arkansas occurred on 11 June 2010 (Fig. 1c). Daily rainfall on 11 June, for the grid box with the maximum rainfall in western Arkansas was 103.4 mm, which ranked first in the historical record (Fig. 2c).

  • South-central Kansas: The major flooding event in south-central Kansas occurred on 13 June 2010 (Fig. 1d). Daily rainfall on 13 June, for the grid box with the maximum rainfall in south-central Kansas was 167.5 mm, which ranked first in the historical record (Fig. 2d).

  • Central Oklahoma: The major flooding event in central Oklahoma occurred on 15 June 2010 (Fig. 1e). Daily rainfall on 15 June, for the grid box with the maximum rainfall in central Oklahoma was 147.0 mm, which ranked first in the historical record (Fig. 2e).

The five extreme events highlighted here are a subset of the total number that occurred during May and June 2010, but they are representative of the precipitation regime that prevailed over the south-central United States during the period. For reference, if all grid boxes over the continental United States (CONUS) are considered, then the weather systems that occurred on these 5 dates (3 May, 10 June, 11 June, 13 June, and 15 June) plus 2 adjacent dates (2 May and 12 June, on which the same weather systems were present) accounted for roughly half (48%) of the total number of grid boxes at which record daily rainfall was observed during the period May and June 2010.

Each of the selected daily precipitation extremes was associated with historical rainfall totals. Although May and June 2010 was an exceptional period in terms of the number of extreme rainfall and major flood events over that 2-month period, this does not confirm or deny a trend in these types of events (see section 6).

Upper bounds on daily precipitation extremes

Spatial maps of the extreme maximum precipitation values that have occurred at each location are noisy (bull’s-eyes) due to the insufficient length of the historical data record (1948–2010). Although there are a number of events with similar characteristics in the historical record (i.e., precipitation amounts, moisture transport, etc.), they have occurred at different locations, so it becomes a sampling problem to characterize how unusual the 2010 events were. Clearly, a record length that is many times greater than the one currently available is required to ensure adequate sampling of extreme events.

For planning purposes we can estimate the upper bounds on daily precipitation extremes that could reasonably be expected to occur given a much longer historical record. A simple approach is to determine the extreme maximum value that has been observed in a specified region in the actual historical record and to use that value as an estimate of what might reasonably be expected to occur at all locations within the region, given a long enough historical record. Such an estimate will be sensitive to the size of the region, so a key consideration is to select regions of the appropriate size to help ensure that all locations are within roughly the same climate zones. Also, the estimates will depend on whether raw station data or gridded analyses are used to identify extreme events, as the former will yield larger values than the latter.

For the purposes of this exercise, the gridded daily analysis is used, in part because the gridded data have undergone rigorous quality control (see section 2a). After some experimentation with the sensitivity of the upper bounds on daily precipitation extremes to the size of the region, grid boxes at a horizontal resolution of (latitude, longitude) = (1° × 1°) were chosen. Each grid point (at 0.25° resolution) within the 1° grid box was checked, and spatial maps of the extreme maximum daily precipitation amount within the 1° box were generated by month (Fig. 3).

Fig. 3.

Estimates of the extreme maximum daily precipitation (mm) that might be expected to occur given a sufficiently long historical record. See text for the procedure used to estimate the upper bounds on daily precipitation extremes.

Fig. 3.

Estimates of the extreme maximum daily precipitation (mm) that might be expected to occur given a sufficiently long historical record. See text for the procedure used to estimate the upper bounds on daily precipitation extremes.

During the warm season the upper bounds can exceed 300 mm along the Gulf Coast (July–October) and the East Coast (August–September) due to land falling tropical cyclones, and (especially during October) due to other nontropical triggering mechanisms including coastal extratropical cyclones, synoptic-scale fronts, topography, and large-scale ascent (Nielsen-Gammon et al. 2005). The North American monsoon can provide rainfall exceeding 100 mm day−1 over portions of the Southwest during August and September. Synoptic-scale disturbances can provide rainfall exceeding 200 mm day−1 along the Gulf Coast (November–March) and the West Coast (November–March). Moist plumes from the Gulf of Mexico and the Great Plains low-level jet contribute to extreme events exceeding 250 mm day−1 along the Gulf Coast and in portions of the Great Plains during the spring (April–June). Elsewhere, the upper bounds are generally in the range of 100–200 mm day−1 through much of the year, except over the intermountain west where they tend to be less than 75 mm day−1.

While the extreme maximum values in Fig. 3 are a reasonable first cut, it is important to note that changes in the observing network and the insufficient length of the historical record introduce potential sources of uncertainty.

4. Circulation patterns and moisture plumes

The meteorological conditions associated with the May and June 2010 heavy precipitation events are examined, with emphasis on the circulation features and associated moisture plumes. Emphasis is placed on identification of any conditions in common to the 2010 events, including the potential role of moisture plumes from the Gulf of Mexico. The following analyzed fields from R2 are used: 500-hPa height, 500-hPa vector wind, 925-hPa height, 925-hPa vector wind, and precipitable water. The 500-hPa vertical motion and sea level pressure were also examined but are not shown. Results are based on analyses that are valid at 0000 UTC on the calendar date indicated.

In the discussion below the classification scheme of Maddox et al. (1979) is used as an organizing framework for heavy precipitation events associated with flash flooding. Maddox et al. (1979) used analyses of surface and standard level upper-air data to identify and define three basic meteorological patterns associated with flash flooding in the central and eastern United States: synoptic events, frontal events, and mesohigh events. Synoptic events develop in association with a relatively intense synoptic scale cyclone or frontal system. A major trough at 500 hPa is usually moving eastward and the associated surface front is often quasi-stationary. Convective storms repeatedly develop and move over the same general area. Frontal events involve a stationary or very slowly moving synoptic-scale frontal boundary that helps to trigger and focus heavy rain storms. However, unlike synoptic events, the heavy rains occur on the cool side of the surface front as warm unstable air flows over the frontal zone. Mesohigh events (the most common) are associated with a nearly stationary thunderstorm outflow boundary. These events typically occur to the east of slow moving large-scale frontal systems.

  • Nashville, Tennessee (3 May 2010): This case fits under the synoptic pattern for flash floods described by Maddox et al. (1979). It featured a major upper-level trough over the central United States, with strong flow from the Rio Grande Valley northeastward across the Tennessee and Ohio Valleys (Fig. 4, top). Strong low-level flow and moisture transport extended from the central Gulf of Mexico north-northeastward across the Southeast and mid-Atlantic states (Fig. 4, bottom). The origins of this moisture plume extended farther south and east toward the Caribbean Sea in association with an active Caribbean easterly low-level jet (not shown). Moist (high precipitable water) air was clearly evident along the western edge of the subtropical ridge anchored off the southeastern coast of the United States. Large low-level moisture convergence and strong synoptic-scale upward motion were evident over the Tennessee and Ohio Valleys (not shown), and a quasi-stationary cold front slowly moved from the Mississippi Valley to the Tennessee and Ohio Valleys (particularly apparent from comparison of 6-h analyses).

  • Eastern Texas (10 June 2010): This case fits under the mesohigh pattern for flash floods described by Maddox et al. (1979). During this event an upper-level cutoff low over northeast Texas, embedded within a synoptic-scale ridge, moved slowly northeastward (Fig. 4, top). Strong low-level flow and moisture transport from the western Gulf of Mexico progressed northward across eastern Texas (Fig. 4, bottom). Low-level moisture convergence, weak upper-level flow, weak vertical wind shear, and relatively cold air (center of cutoff low) all favored the slow-moving convective storms and nearly stationary thunderstorm outflow boundaries that characterized this event.

  • Western Arkansas (11 June 2010): This case fits under the mesohigh pattern for flash floods described by Maddox et al. (1979). This case, a continuation of the previous, featured an upper-level trough over northwestern Arkansas embedded within a synoptic-scale ridge (Fig. 4, top). Strong low-level flow and moisture transport from the western Gulf of Mexico progressed northward across eastern Texas and Arkansas toward the upper Midwest (Fig. 4, bottom). Low-level moisture convergence, weak upper-level flow, and strong upward motion all favored the slow-moving convective storms that characterized this event.

  • South-central Kansas (13 June 2010): This case fits under the synoptic-scale pattern for flash floods described by Maddox et al. (1979). It featured a major upper-level trough over the western United States with strong southwesterly flow from New Mexico northeastward across eastern Colorado (Fig. 4, top). Strong low-level flow and moisture transport from the western Gulf of Mexico progressed north-northeastward across the southern and central Plains (Fig. 4, bottom). Large low-level moisture convergence occurred over southern Kansas. Low-level northerly flow over the Dakotas and Nebraska and low-level southerly flow over Texas and Oklahoma impinged on a quasi-stationary frontal boundary across Kansas, which provided lift for the warm moist low-level flow and favored organized slow moving convective storms and the training of convection.

  • Central Oklahoma (15 June 2010): This case fits under the mesohigh pattern for flash floods described by Maddox et al. (1979). A weakening upper-level trough over the central Great Plains was embedded in broad west southwesterly flow from New Mexico eastward across the central United States (Fig. 4, top). Low-level flow and moisture transport from the western Gulf of Mexico progressed northward across Texas and into Oklahoma (Fig. 4e, bottom). Low-level moisture convergence and strong upward motion over north Texas and southern Oklahoma favored slow-moving convective storms and nearly stationary thunderstorm outflow boundaries in the region, including Oklahoma City.

Fig. 4.

(top) The 500-hPa height (dam), 500-hPa vector wind (m s−1), and 500-hPa isotachs of wind speed (m s−1) for selected extreme events during May and June 2010. (bottom) 925-hPa vector wind (m s−1) and 925-hPa isotachs of wind speed (m s−1) for selected extreme events during May and June 2010. Dates of the extreme events are indicated across the top of the figure. Isotachs are shaded as indicated by the color bars. Analyses are valid at 0000 UTC on the calendar dates indicated.

Fig. 4.

(top) The 500-hPa height (dam), 500-hPa vector wind (m s−1), and 500-hPa isotachs of wind speed (m s−1) for selected extreme events during May and June 2010. (bottom) 925-hPa vector wind (m s−1) and 925-hPa isotachs of wind speed (m s−1) for selected extreme events during May and June 2010. Dates of the extreme events are indicated across the top of the figure. Isotachs are shaded as indicated by the color bars. Analyses are valid at 0000 UTC on the calendar dates indicated.

Thus, the heavy precipitation and flooding events during May and June 2010 had meteorological conditions in common. Two of the cases fit under the synoptic pattern for flash floods while the remaining three cases fit under the mesohigh pattern (Maddox et al. 1979) as described above. At lower levels (Fig. 4, bottom), the eastern Texas, south-central Kansas, and central Oklahoma events were similar in that the flooding occurred just downstream of the low-level jet (LLJ) wind maximum. As shown by Maddox et al. (1979), and going back to Means (1956), flooding often occurs downstream of the LLJ wind maximum in the region of strong low-level convergence. The Nashville, Tennessee, and western Arkansas events were different in that they featured a flood region well within the low-level jet core region (i.e., the low-level jet region extended north and south of the flood area).

All of the May and June 2010 events were associated with warm moist air masses that brought record-breaking warm temperatures to surrounding regions of the country. During the overnight hours of the 11 June flood in Arkansas, 50 airports in the South and Midwest experienced record-warm minimum temperatures (NOAA/National Climatic Data Center 2010a). Just prior to the Nashville flood on 3 May, over 100 locations in the eastern half of the United States experienced record-warm minimum temperatures on 2 May (NOAA/National Climatic Data Center 2010b). The air mass that spawned the 15 June Oklahoma City floods set record-warm minimum temperatures at more than 2 dozen airports across the central and eastern portions of the United States on 14 June (NOAA/National Climatic Data Center 2010a).

All five cases featured strong low-level flow and moisture transport from the Gulf of Mexico northward into the southern Great Plains or Southeast. Spatial maps of precipitable water (PW) show the presence of these plumes of high PW air in the vicinity of (and to the east) of the major flood events (Fig. 5). In each case the moisture plumes in Fig. 5 extended deep into the Caribbean, and exhibited the characteristics of the “Maya Express” flood events (Dirmeyer and Kinter 2009) that link tropical moisture from the Caribbean and Gulf of Mexico to midlatitude flooding over North America. During the heavy rainfall events the fetch of Caribbean moisture linked into the Great Plains low-level jet, creating a much longer “atmospheric river” of moisture. These cases were also related to strengthening or displacement of the Atlantic subtropical ridge, as was the case with the Maya Express events of Dirmeyer and Kinter (2009). It is worth noting that a link between moisture from the Caribbean Sea and the flooding over the Great Plains during 1993 has also been established in previous work (Dirmeyer and Brubaker 1999).

Fig. 5.

Precipitable water (mm) for selected extreme events during May and June 2010. Dates of the extreme events are indicated across the top. Analyses are valid at 0000 UTC on the calendar dates indicated.

Fig. 5.

Precipitable water (mm) for selected extreme events during May and June 2010. Dates of the extreme events are indicated across the top. Analyses are valid at 0000 UTC on the calendar dates indicated.

The atmospheric PW associated with these events was also examined at specific locations near the extreme precipitation events to see whether the PW values were unprecedented. Moist air masses were present in each case, with PW values close to 50 mm, which is quite common for tropical air advancing northward from the Caribbean and Gulf of Mexico. Although the PW values were high, they were not at record levels when compared to other events in the historical record. Extreme values over the United States may reach 55–60 mm (and on rare occasions near 65 mm), associated with landfalling tropical cyclones. It is also worth noting that in the deep tropics (Amazon basin, Indonesia regions) PW values are often as high as 60–65 mm. Nevertheless, the 2010 events over the United States were textbook examples of the association of heavy rainfall with strong moisture feeds and high PW air.

5. Relationship to the ENSO cycle

Shifts in the frequency of daily precipitation extremes over the central United States during the spring and summer months may be tied to the phase of ENSO. This was examined systematically by ranking daily precipitation events at each grid point by season [January–March (JFM), April–June (AMJ), July–September (JAS), October–December (OND)] for the period 1950–2009. The top 150 events were selected and sorted based on whether they occurred during El Niño, La Niña, or ENSO-neutral conditions. A classification of historical warm (El Niño) and cold (La Niña) conditions developed by the CPC was used to determine the ENSO phase. El Niño and La Niña conditions were identified using the Oceanic Niño Index (ONI; Kousky and Higgins 2007). (The ONI index can be found on the CPC Web site http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ensoyears.shtml.)

The numbers of El Niño events are 14, 15, 16, and 20 for JFM, AMJ, JAS, and OND, respectively; the numbers of La Niña events are 19, 16, 15, and 20 for JFM, AMJ, JAS, OND, respectively. The remaining years are neutral ones.

The results above were used to compute the percentage of top 150 events that occurred during neutral, El Niño, and La Niña conditions by season as well as the percentage of the total number of seasons with neutral, El Niño, and La Niña conditions. Subtracting the 2 percentages yields the percent departure from expected of the number of top 150 events by ENSO phase (Fig. 6). The contours (black lines) indicate where the departures are significant at the 95% level. A mask has been applied to locations where the top 150 daily precipitation events do not exceed a threshold for the mean (1950–2009) daily precipitation of 0.25 mm.

Fig. 6.

Percent departure from expected of the number of top 150 daily precipitation events by ENSO phase. Results are shown by season and are based on the period 1950–2009. The number of El Niño events is 14, 15, 16, and 20 for JFM, AMJ, JAS, and OND, respectively. The number of La Niña events is 19, 16, 15, and 20 for JFM, AMJ, JAS, and OND, respectively. A mask has been applied to locations where the top 150 daily precipitation events do not exceed a threshold for the mean (1950–2009) daily precipitation of 0.25 mm. The contours (black lines) indicate where the departures are significant at the 95% level.

Fig. 6.

Percent departure from expected of the number of top 150 daily precipitation events by ENSO phase. Results are shown by season and are based on the period 1950–2009. The number of El Niño events is 14, 15, 16, and 20 for JFM, AMJ, JAS, and OND, respectively. The number of La Niña events is 19, 16, 15, and 20 for JFM, AMJ, JAS, and OND, respectively. A mask has been applied to locations where the top 150 daily precipitation events do not exceed a threshold for the mean (1950–2009) daily precipitation of 0.25 mm. The contours (black lines) indicate where the departures are significant at the 95% level.

The results in Fig. 6 are consistent with the known ENSO-related patterns and permit an assessment of the extent of the departures (i.e., increases and decreases) in daily extremes during spring as compared to the other seasons. During El Niño, significant increases in extremes occur along the southern tier of states during the fall and winter and over portions of the central Great Plains during the winter months, consistent with the wetter-than-normal conditions often experienced in those regions. Significant decreases in daily precipitation extremes occur in portions of the Ohio and Tennessee Valleys during the winter season. In general the El Niño influence is relatively weak during the spring and summer months, though there are a few small areas with significant increases in the Pacific Northwest during the spring and Intermountain West during the summer. During La Niña, significant increases in daily extremes occur in the Pacific Northwest during the fall and in the Ohio Valley during the winter, where wetter-than-normal conditions typically occur. Significant decreases in extremes are found along the southern tier of states during the fall and winter months and in portions of the central Great Plains during the winter. La Niña influences are relatively weak during the spring, but there are some areas of the west and south with significant decreases in extremes.

During ENSO-neutral conditions there are a few small areas with significant increases or decreases in extremes depending on the time of year, but in general the signal is generally quite weak including during the spring. Although the period May–June 2010 was characterized by neutral conditions (occurring just after the 2009–10 El Niño and just before the 2010–11 La Niña), there is no evidence to implicate neutral conditions as a factor in the extreme events over the south-central United States during the period.

6. Trends

The gridded precipitation analysis was used to examine the number of daily precipitation events at various thresholds (including extreme events) that occurred during the period 1950–2009 over the CONUS. The year 2010 was not included because the analysis was carried out midway through 2010. For each day in the time series, we counted the number of grid points within the CONUS at which the total precipitation exceeded the thresholds 25, 50, 75, 100, and 125 mm. A plot of the time series (365-day running mean) of the number of events for the thresholds of 25, 50, 75, 100, and 125 mm is shown in Fig. 7.

The results in Fig. 7 show several interesting features. Each of the time series has large interannual variability, with good agreement between El Niño (La Niña) and increases (decreases) in the number of daily events especially at the higher thresholds (100 and 125 mm). There is little obvious evidence of trends at lower thresholds, but some indication of increased variability and increases in the number of extremes at the higher thresholds, particularly after 1980. Karl et al. (2009) found a decrease in the frequency of light and moderate precipitation events during the past 30 yr, and a clear increase in the frequency of heavy downpours which accounted for most of the observed increase in overall precipitation during the last 50 yr. An examination of the number of daily precipitation events at various thresholds that occurred during the period 1950–2009 over CONUS, but by season (i.e., JFM, AMJ, JAS, and OND), also did not reveal any obvious trends. Again there was a good correspondence with ENSO, especially during the fall and winter seasons when ENSO influences on storminess are the strongest. The extent to which any changes in variability might be due to changes in station counts versus a stronger connection between ENSO and extremes has not been investigated.

Fig. 7.

Number of daily precipitation events exceeding thresholds of 25, 50, 75, 100, and 125 mm for the continental United States. A 365-day running mean of the results is shown.

Fig. 7.

Number of daily precipitation events exceeding thresholds of 25, 50, 75, 100, and 125 mm for the continental United States. A 365-day running mean of the results is shown.

We also examined counts of the number of very heavy precipitation events in the more recent period (1980–2009) and in the earlier period (1950–79), in particular to see if there were any obvious signals in daily precipitation extremes over the central United States during the spring and summer months in the more recent period. For this we ranked the daily precipitation values (1950–2009) for each month at all grid points in CONUS. We took the top 50 values (roughly the 97th percentile or greater) and counted the number of events that occurred during the period 1950–79 (30 yr) versus during the period 1980–2009 (30 yr). Counts (out of a possible total of 50 events at each location) for 1980–2009 for each month are shown in Fig. 8. Areas where the count is larger than 27 of 50 events are shaded in greens and areas where the count is less than 23 of 50 events are shaded in browns. The contours (black lines) indicate where the departures are significant at the 95% level. A mask has been applied to locations where the top 50 daily precipitation events do not exceed a threshold for the mean (1950–2009) daily precipitation of 0.25 mm.

Some significant increases in extremes have been observed during the recent 30-yr period, particularly over portions of the central and southern United States during the fall and winter months (Fig. 8). The patterns appear to be ENSO-like, especially during the winter months, which may indicate that the strong El Niño episodes (1982/83, 1991/92, and 1997/98) help to explain the increases in heavy precipitation events. Focusing on the central United States, there does not appear to be a spatially coherent change (increase or decrease) in the number of extreme daily events during May and June. There is considerable month-to-month variability in the significant areas on Fig. 8, except for the fall and winter months in the areas mentioned above.

Fig. 8.

Number of top 50 daily precipitation events that occurred during 1980–2009. Areas shaded in greens and blues had more events in the recent period (1980–2009) than in the earlier period (1950–79). See text for details on the ranking procedure. A mask has been applied to locations where the top 50 daily precipitation events do not exceed a threshold for the mean (1950–2009) daily precipitation of 0.25 mm. The contours (black lines) indicate where the departures are significant at the 95% level.

Fig. 8.

Number of top 50 daily precipitation events that occurred during 1980–2009. Areas shaded in greens and blues had more events in the recent period (1980–2009) than in the earlier period (1950–79). See text for details on the ranking procedure. A mask has been applied to locations where the top 50 daily precipitation events do not exceed a threshold for the mean (1950–2009) daily precipitation of 0.25 mm. The contours (black lines) indicate where the departures are significant at the 95% level.

Table 2 shows the percentage of top 50 extreme events that occurred during the period 1950–79 and during the period 1980–2009 for CONUS as well as the percent difference between the 2 periods. For the CONUS, on average there have been 2.6% more daily extreme events during 1980–2009 than during 1950–79. The largest increases are during autumn (October and November) and late winter (March). A 3.8% increase in daily precipitation extremes occurred during May, though these increases appear to have occurred primarily in the Southwest and Ohio Valley, not in the south-central United States (Fig. 8).

Table 2.

Percentage of the number of top 50 daily precipitation events that occurred during the period 1950–79 and during the period 1980–2009 for the continental United States.

Percentage of the number of top 50 daily precipitation events that occurred during the period 1950–79 and during the period 1980–2009 for the continental United States.
Percentage of the number of top 50 daily precipitation events that occurred during the period 1950–79 and during the period 1980–2009 for the continental United States.

Overall, the results in Fig. 8 and Table 2 are qualitatively consistent with recent assessments (e.g., Karl et al. 2009) that have reported increases in the amounts of precipitation falling in very heavy precipitation events and a trend toward more very heavy precipitation for the nation as a whole. No attempt has been made to look at quantitative increases in the number of cases on a regional or local basis, although the data are available.

7. Discussion and future plans

The months of May and June 2010 were characterized by a large number of heavy rain events leading to major flooding across portions of the central and southern United States. Each of the major flood events during this period was associated with historic rainfall totals. Several of the events had meteorological conditions in common, and all of the events fit well into the classification scheme of Maddox et al. (1979) for heavy precipitation events associated with flash flooding. While moist air masses emanating from the Gulf of Mexico were present in each case, the precipitable water values were not at record levels when compared to some recent land falling tropical cyclones. However, all five of the cases examined here were associated with moisture plumes that extended deep into the Caribbean, and exhibited the characteristics of the “Maya Express” flood events (Dirmeyer and Kinter 2009) that link tropical moisture from the Caribbean and Gulf of Mexico to midlatitude flooding over North America.

Changes in the numbers of extremes for each phase of ENSO are consistent with the known ENSO-related shifts in the large-scale circulation, Pacific jet stream, and patterns of storminess. The largest ENSO-related departures (increases and decreases) in daily extremes are during the fall and winter months, with only a few areas with significant departures during the spring depending on the ENSO phase. Although the period May–June 2010 was characterized by neutral conditions (occurring just after the 2009–10 El Niño and just before the 2010–11 La Niña), there is no evidence to implicate neutral conditions as a factor for the extreme events over the south-central United States during the period.

This analysis showed an average increase of 2.6% in extreme events for the CONUS during the period 1980–2009 as compared to the period 1950–79. The largest increases occurred in October, November, and March (i.e., during the transition periods in the annual cycle). There was a 3.8% increase in daily precipitation extremes for May in the more recent 30-yr period, but these increases appear to have occurred primarily in the Southwest and Ohio Valley, not in the south-central United States.

According to Karl et al. (2009), much of the increase in the amount of rain falling in the heaviest downpours has occurred since 1970, during a period in which average temperatures in the United States have increased by approximately 1°F (e.g., Solomon et al. 2007). Solomon et al. (2007) also concludes that water vapor in the global atmosphere has increased by about 4% since 1970. Trenberth et al. (2005) used satellite measurements to document a 1.3% decade−1 increase in water vapor over the global oceans since 1988. Although one cannot attribute a single event (or even series of events over a season) to climate change, it is logical to conclude that a systematic increase in water vapor in the atmosphere could have a systematic influence on extreme precipitation events by invigorating storms and by providing additional moisture for heavy rainfall. While we cannot conclude that the May and June 2010 heavy precipitation events were a consequence of global warming, it is logical to suggest that the events were enhanced by the presence of significant water vapor anomalies in the atmosphere as shown here. Moreover, it is logical to conclude that we can expect an increase in heavy precipitation events and the associated flooding in the United States (and worldwide) if the climate continues to warm.

Additional studies of daily precipitation extremes are planned to support the findings reported here and to improve our understanding of the linkage between precipitation extremes, climate variability, and climate change. This will include a comparison of the precipitation and circulation patterns that prevailed during the spring and summer of 2010 and 1993, the year of the Great Flood in the Midwest, to sort out the extent of any similarities between the two years. This will also include studies of the relative contributions of diurnal, daily, interannual, and decadal variability in daily precipitation to the total variability and of the uncertainties due to changes in instrumentation and changes in the number of observations through the historical record. Ultimately, these studies are aimed at improving the ability of climate models (such as the NCEP climate forecast system) to reproduce the statistics of daily precipitation extremes found in nature.

Acknowledgments

The authors gratefully acknowledge the assistance of Dr. Soo-Hyun Yoo of the Climate Prediction Center, who provided considerable assistance with the analysis for this project, and Dan Collins who helped with the statistical significance tests. The authors also gratefully acknowledge Drs. Wei Shi and Jon Gottschalck, who provided constructive reviews of early versions of this manuscript.

REFERENCES

REFERENCES
Brooks
,
H. E.
, and
D. J.
Stensrud
,
2000
:
Climatology of heavy rain events in the United States from hourly precipitation observations
.
Mon. Wea. Rev.
,
128
,
1194
1201
.
Chen
,
M.
,
W.
Shi
,
P.
Xie
,
V. B. S.
Silva
,
V. E.
Kousky
,
R. W.
Higgins
, and
J. E.
Janowiak
,
2008
:
Assessing objective techniques for gauge-based analyses of global daily precipitation
.
J. Geophys. Res.
,
113
,
D04110
,
doi:10.1029/2007JD009132
.
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
,
19 383
19 397
.
Dirmeyer
,
P. A.
, and
J. L.
Kinter
III
,
2009
:
The Maya Express—Late spring floods in the U.S. Midwest
.
Eos, Trans. Amer. Geophys. Union
,
90
,
101
102
.
Higgins
,
R. W.
,
W.
Shi
,
E.
Yarosh
, and
R.
Joyce
,
2000
:
Improved United States Precipitation Quality Control System and Analysis
.
NCEP/Climate Prediction Center Atlas 7, 40 pp. [Available online at http://www.cpc.ncep.noaa.gov/research_papers/ncep_cpc_atlas/7/index.html]
.
Higgins
,
R. W.
,
V.
Silva
,
J.
Larson
, and
W.
Shi
,
2007
:
Relationships between climate variability and fluctuations in daily precipitation over the United States
.
J. Climate
,
20
,
3561
3579
.
Kalnay
,
E.
, and
Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project
.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
.
Karl
,
T. R.
,
J. M.
Melillo
, and
T. C.
Peterson
, Eds.,
2009
:
Global Climate Change Impacts in the United States
.
Cambridge University Press, 191 pp
.
Kanamitsu
,
M.
,
W.
Ebisuzaki
,
J.
Woollen
,
S.-K.
Yang
,
J. J.
Slingo
,
M.
Fiorino
, and
G. L.
Potter
,
2002
:
NCEP–DOE AMIP-II Reanalysis (R-2)
.
Bull. Amer. Meteor. Soc.
,
83
,
1631
1643
.
Kousky
,
V. E.
, and
R. W.
Higgins
,
2007
:
An alert classifications system for monitoring and assessment of the ENSO cycle
.
Wea. Forecasting
,
22
,
353
371
.
Maddox
,
R. A.
,
C. F.
Chappell
, and
L. R.
Hoxit
,
1979
:
Synoptic and meso-α scale aspects of flash flood events
.
Bull. Amer. Meteor. Soc.
,
60
,
115
123
.
Means
,
L. L.
,
1956
:
Some basic parameters associated with the flood rains at Chicago, October 9–12, 1954
.
Mon. Wea. Rev.
,
84
,
253
260
.
Nielsen-Gammon
,
J. W.
,
F.
Zhang
,
A. M.
Odins
, and
B.
Myoung
,
2005
:
Extreme rainfall in Texas: Patterns and predictability
.
Phys. Geogr.,
26
,
340
364
.
NOAA/National Climatic Data Center
, cited
2010a
:
State of the Climate: National Overview for June 2010
.
NOAA/National Climatic Data Center
, cited
2010b
:
State of the Climate: National Overview for May 2010
.
Solomon
,
S.
,
D.
Qin
,
M.
Manning
,
M.
Marquis
,
K.
Averyt
,
M. M. B.
Tignor
,
H. L.
Miller
Jr.
, and
Z.
Chen
, Eds.,
2007
:
Climate Change 2007: The Physical Sciences Basis
.
Cambridge University Press, 996 pp
.
Trenberth
,
K. E.
,
J.
Fasullo
, and
L.
Smith
,
2005
:
Trends and variability in column-integrated atmospheric water vapor
.
Climate Dyn.
,
24
,
741
758
.