Hydroclimatology of the 2008 Midwest floods



[1] The late spring/early summer flooding that occurred in the American Midwest between May and June 2008 resulted from a combination of large-scale atmospheric circulation patterns that supported a steady influx of moisture into the area. A low pressure system centered over the central-western United States steered a strong jet and associated storms along its eastern edge from the west to southwest and an anomalously strong Great Plains Low Level Jet brought continuous warm and moist air into the area from the Gulf of Mexico into the area. We examine and quantify here the impact these circulation patterns had on the hydroclimatology of the Midwest highlighting the magnitude, frequency, geographic distribution, and temporal evolution of precipitation that ultimately magnified the flooding. Historical precipitation records were used to assess the regional rainfall characteristics at various geographic and time scales. Five distinct hydroclimatic characteristics contributed to the definition of the 2008 flood including persistent high surface soil moisture conditions prior to flooding exasperated by anomalously high rainfall, extreme rainfall totals covering extensive areas, increased frequency of shorter-term, smaller-magnitude events, persistent multiday heavy precipitation events, and extreme flood-producing rain storms. The major flooding lasted for approximately 24 days and most greatly impacted the state of Iowa, southern Wisconsin, and central Indiana. Its occurrence during the May–June period makes the event especially unusual for this region.

1. Introduction

[2] Between the end of May and the middle of June 2008, the American Midwest experienced widespread flooding. The event was preceded by prolonged wet surface conditions and coincided with atmospheric circulation patterns that guided moist air into the region [Coleman and Budikova, 2010]. The American Midwest receives over half of its moisture from the Gulf of Mexico by the Great Plains Low Level Jet (GPLLJ) and its strength, shear, and relative divergent circulations influence the timing, location, and intensity of precipitation across the central United States [Stensrud, 1996; Higgins et al., 1997; Dominguez and Kumar, 2005]. The 2008 flooding and unseasonably wet conditions over the study area coincided with a strong and active GPLLJ that continuously brought moist air into the American Midwest between May and early June. This flow further interacted with the strong and southwardly displaced North American jet that amplified the moisture influx into the study area. During the peak period of the 2008 flooding, a midtropospheric trough was positioned over the central United States also enhancing the influx of moist air into the area from the Gulf of Mexico [Coleman and Budikova, 2010]. The strength of the GPLLJ and resultant precipitation changes across the central United States have been tied to the extratropical Atlantic, specifically to the behavior of the North Atlantic Oscillation (NAO) [Weaver and Nigam, 2008]. Weaver and Nigam [2008] found negative phases of the NAO to coincide with stronger than expected influx of Gulf air into the U.S. interior and unseasonably high (low) precipitation in the Midwest (Southeast). Hu et al. [1998] and Ruiz-Barradas and Nigam [2005] argued that the GPLLJ strength and activity are related to the southerly branch of the Bermuda High centered over the North Atlantic, typically impacted by the state of the NAO. Coleman and Budikova [2010] confirmed this connection for the 2008 flood season when NAO was persistently in a strong negative phase.

[3] Ensuing effects of the anomalous rainfall on local hydrology and economy were most significant to areas in central Indiana, southern Wisconsin, and eastern Iowa. During the first week of June, the White River at Newberry, IN (Figure 1), rose 4.5 m above flood stage breaking the previous record set in 1913 [Callahan, 2008]. During the month of June, the White River at Newberry recorded a mean discharge rate of 767 m3/s or 573% of the normal. On 10 June, the banks of Lake Delton in southern Wisconsin overflowed releasing 2635 million L of water from a 267 acre lake washing away five homes and a 120 m section of a local highway [Hesselberg and Seely, 2008]. On 21 June, the Rock River, near Afton, WI, topped the historical crest record by 1.38 m and breaking a 92 year record; the mean discharge rate was 311 m3/s or 571% of the normal, for the month of June. The Cedar River at Cedar Rapids broke a historical river crest record by 3.39 m and experienced a mean June discharge rate of 1315 m3/s or 775% of the normal.

Figure 1.

Place names across the study area mentioned in the text.

[4] Overall losses resulting from the flooding were significant, estimated at approximately $15 billion including about $8 billion in agricultural losses and 24 deaths [National Climatic Data Center (NCDC), 2008, 2009; Facts on File, 2008]. By the middle of June, Iowa alone experienced nearly $10 billion in damages to infrastructure, residential, and commercial properties [USDOC, 2009]. Iowa agricultural losses have been estimated at $4 billion, with 2–3 million acres of farmland inundated [USDOC, 2009]. Impacts on infrastructure and local economies were costly and extensive. Transportation systems on the main channel of the Upper Mississippi River came to a standstill by 12 June, when the Army Corps of Engineers closed a 250 mile stretch of the river from Fulton, IL, to Clarksville, MO, due to flooding [Crumb, 2008]. Failing levee systems contributed to flooding in small towns and agricultural lands along the main channel of the Mississippi River and across much of eastern Iowa [Boshart, 2008; NCDC, 2008]. By the end of the summer, the Federal Emergency Management Agency (FEMA) declared 50 of 92 counties in Indiana, all 31 counties in southern Wisconsin, and 86 of the 99 counties in Iowa eligible for federal and state disaster assistance related to flooding [NCDC, 2008; APa, 2008; Federal Emergency Management Agency (FEMA), 2008; States News Service, 2008]. Approximately 38,000 residents were evacuated from 18 counties in Iowa during the flood event [States News Service, 2008]. Flooding in Iowa closed 747 km of primary roads, 300 bridges, and over 640 km of train tracks were damaged or completely destroyed [United States Department of Commerce, 2009].

[5] The objective of this work is to identify and document the hydroclimatic conditions that helped produce and ultimately exacerbated the flooding across the Midwestern United States between May and June 2008. We place the hydroclimatology in a historical context by examining the characteristics of local, regional, and basin-wide precipitation from daily to seasonal timescales. Such an examination will contribute to our understanding of the natural causes of flooding and the frequency at which this flooding may be expected to occur in the future across the American Midwest.

2. Data Sources, Methods, and Study Area

[6] The general geographic area impacted by the 2008 flood event is shown in Figure 2 and is here referred to as the American Midwest. It encompasses a vast region of approximately 950,000 km2 and includes portions of 10 U.S. states and 54 climate divisions. The study area captures the Greater Upper Mississippi River Basin (GUMRB) as defined by Kunkel et al. [1994] and the Wabash River basin located in Indiana. The GUMRB includes the Mississippi River drainage area north of Cairo, IL, and the southwest portion of the Missouri River. Included within the study area are climate divisions with more than 50% of their respective areas within each of the basins. Divisions outlined in dark bold are used to define precipitation regions used in case studies of areas that experienced the highest magnitudes of flooding during the 2008 late spring/early summer months (section 5).

Figure 2.

Boundaries of the study area in the American Midwest. Included are the Greater Upper Mississippi River Basin (GUMRB) (dark shading) and Wabash River Basin (light shading). Also shown are the climate divisions contained within the study area and the regions used to assess the hydroclimatic conditions for the three case studies (see text for criteria of precise inclusions of climate divisions and justifications used for defining the case-study regions). The GUMRB study area has been outlined after the study by Kunkel et al. [1994].

[7] The data used to analyze the hydroclimatology came from five principal sources. First, total monthly precipitation records estimated at the climate division level were used to examine the hydroclimatology at regional and basin-wide scales. The data were complete and available for the period of record between 1895 and 2008. They were obtained from the National Climatic Data Center, Asheville, NC [NCDC, 1994]. Guttman and Quayle [1996] and Keim et al. [2005] described the construction of the divisions, the computation of data values for each month and climate division on record, the data treatment procedures, and the strengths and weaknesses of the data set. Division values typically comprise between 10 and 20 station records [Kunkel et al., 1994].

[8] Second, historical records of daily precipitation collected at station level were used to examine the hydroclimatology at local scales. The data came from the U.S. Historical Climatology Network (U.S. HCN) [Easterling et al., 1999] and was obtained from the National Climatic Data Center (NCDC) at ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/daily. At present, the precipitation data set contains historical values of daily station precipitation compiled from 14 data sources including the U.S. Cooperative Summary of Day (NCDC DSI-3200), U.S. First Order Summary of Day (NCDC DSI-3210), and the Climate Data Modernization Program Cooperative Summary of Day (NCDC DSI-3206), extend the historical records at many stations across the study area to the beginning of the 20th century. The data set contains the most geographically dense network of stations across the country with many stations having more than 50 years of record. It was our goal to utilize as much of the available data as possible but realizing that considerable gaps in these records exist. Only stations with complete records needed for a given analysis were used here, and as a result, the number of locations suitable for each analysis varied.

[9] Third, detailed local-level analyses of precipitation in the case studies used quality-controlled daily precipitation estimates from radar and rain gauges across the United States as the HCN data set was found to be inadequate to accurately capture the geographic complexity of the precipitation events of interest. The data were obtained from the National Weather Service (NWS) River Forecast Centers at http://water.weather.gov/precip/download.php. The data set is gridded with each value representing average precipitation over areas of roughly 16 km2 for a 24 h period. Seo [1999], Seo et al. [1999], and Seo and Breidenbach [2002] addressed the estimation procedures and quality control issues associated with the estimates. For each date of interest, the valid data were re-gridded using Nearest Neighbor Interpolation Method [Chang, 2008], contoured for the entire United States and subsequently clipped by the various regions of interest to avoid negative edge effects associated with the boundaries as defined by the climate divisions that were chosen to capture the upstream hydrology in each case study region (for additional details see section 5).

[10] Fourth, the NCDC/NOAA U.S. Records database at http://www.ncdc.noaa.gov/extremes/records.php, was used to identify daily precipitation records for May and June across various recoding stations in the region. The data are based on historical daily observations archived in NCDC's Cooperative Summary of the Day data set and preliminary reports from Cooperative Observers and First Order National Weather Service stations. The period of record represents the number of years with a minimum of 50% data completeness. All stations have a period of record of at least 30 years.

[11] Fifth, daily archive satellite imagery in the visible and infrared (IR) spectrum, as well as hourly composite radar overlays, was used to assess mesoscale conditions associated with specific flood events. The data were acquired through a Web-accessible repository maintained by the Precipitation Diagnostic Group in the Mesoscale and Microscale Division of the National Center for Atmospheric Research at http://www.mmm.ucar.edu/imagearchive/.

[12] Records of daily discharge rates collected at three USGS gauging stations were used to assess the local hydrology associated with the hydroclimatic conditions through hydrographs. The data were obtained from the USGS National Water Information System at http://waterdata.usgs.gov/nwis. The gauging stations and corresponding station ID numbers used were Cedar Rapids, IA (05464500), Wisconsin Dells, WI (05404000), and Columbus, IN (03364000).

[13] Climatological values of precipitation were computed based on the 1971–2000 period of record were used to compute anomalies. The precipitation amounts also were placed into historical context by estimating the event's probability of occurrence using the maximum likelihood method and the Generalized Extreme Value (GEV) distribution [Farago and Katz, 1990] for stations with more than 80 years of record as recommended by Kunkel et al. [1994]. The fitted curves were used to estimate the probability of occurrence of the 2008 precipitation totals at basin-wide, regional, and local levels assuming the GEV distribution. The resulting values were then converted into a return period equivalent.

3. Flooding Precursor Conditions

[14] Using the Palmer Hydrological Drought Index, Coleman and Budikova [2010] found anomalously wet surface conditions across a significant portion of the Midwest for the entire 12 month period prior to the flooding. Beginning in summer 2007, above normal precipitation amounts were recorded across the area; between December 2007 and May 2008 precipitation rates continuously exceeded normal values (Figure 3) with record wetness documented at many sites throughout the region [NCDC, 2009]. Four months, August 2007, October 2007, December 2007, and February 2008, had precipitation of at least 150% of the normal; only one month, November 2007, was considerably drier during this period. Excess precipitation from August 2007 to April 2008 amounted to nearly 140 mm throughout the study area. Coleman and Budikova [2010] reported cooler than expected conditions for five consecutive months leading to the flooding, conditions that kept soils wet by reducing potential evapotranspiration rates from the surface. Together, the conditions contributed to river levels that often exceeded flood stages and produced localized early flooding throughout the region in the early spring. The saturated soils along with the elevated stream levels ultimately positioned the region for increased probability of flooding during the following season.

Figure 3.

Monthly precipitation characteristics across the Midwest between May 2007 and August 2008. Shown is the total weighted mean (by climate division) areal precipitation expressed as percent normal (bars), the expected 1971–2000 cumulative precipitation (mm) (dashed line), and estimated cumulative precipitation (mm) (solid line).

4. Hydroclimatic Conditions Associated With the 2008 Flooding

[15] Flooding in the American Midwest is known to be caused and exacerbated by particular precipitation characteristics that manifest at various spatial and temporal time scales. These factors include conditions such as antecedent soil moisture conditions, the presence of heavy rain events, the number of extreme flash-flood-producing rainstorms, and the presence of large-sized rainfall areas [e.g., Kunkel et al., 1994]. The discussions below highlight the various hydroclimatic characteristics that played a role in the development and maintenance of the flooding that occurred in the late spring/early summer season in 2008.

4.1. Seasonal and Monthly Conditions

[16] Between May and June 2008, the American Midwest received anomalous amounts of rainfall that peaked in June at about 157 mm or 150% of its expected value (Figure 3). The magnitude and corresponding date of the study area's largest 10 monthly precipitation events that occurred in the historical record between the months of April and August are shown in Table 1. The rankings were computed using area-weighted (by climate division), precipitation totals from 1895 to 2008 for 1, 2, 3, 4, and 5 month consecutive intervals. For the 1 and 2 month periods, the 2008 totals rank as only the 11th and 10th largest for the entire basin. For the 3, 4, and 5 month periods, the 2008 event had the 8th and 10th, 5th, and 9th highest precipitation totals, respectively. Additional significant observations were made pertaining to the 2008 event not reported in Table 1. When only the historical records for the month of April were considered, 2008 ranked as the 8th wettest on record; when only the historical records for the month of June were considered 2008 ranked as the 6th wettest on record for the entire basin. For the 2 month period, 2008 was the 8th wettest for the April–May as well as May–June period, and it was the 5th wettest for the June–July period. The 2008 year ranks second wettest for the April–June and April–July periods. For April–June, record precipitation amount of 396 mm was recorded in 1947, 2008 falling close behind at 391 mm; for April–July 1993 ranked highest at 580 mm with 2008 coming in second at 508 mm. The relatively low 2008 rankings largely reflect the high localized concentrations of precipitation throughout the region during only a few weeks in late spring/early summer as shown in the subsequent sections. It is evident here also that the 2008 precipitation was especially unusual for the late spring/early summer season during April, May, and June. Estimated return periods for each of the five area-averaged monthly 2008 precipitation periods shown in Table 1 were calculated to demonstrate the likelihood of similar precipitation events occurring. Return periods were determined from all possible consecutive 1 to 5 month combinations between the months of April and August and are shown in Table 2. The 1, 2, and 3 month combinations for the basin estimated return periods at 50 years; the 4 month combination estimate was slightly higher at 64 years; and the 5 month precipitation total was estimated at a mere 15 year return period for 2008.

Table 1. Magnitude (mm) and Date (Month, Year) of the Study Area's Largest 10 Monthly Precipitation Periods Between April and Augusta
Rank1 Month2 Months3 Months4 Months5 Months
  • a

    Precipitation magnitudes are based on area-weighted climate division data for the period 1895–2008. Months occurring during 2008 are in bold. For date, read 7/1993 and 6–7/1993 as July 1993 and June–July 1993.

1196 (7/1993)363 (6–7/1993)485 (6–8/1993)603 (5–8/1993)702 (4–8/1993)
2180 (6/1947)318 (7–8/1993)481 (5–7/1993)580 (4–7/1993)595 (4–8/1915)
3173 (6/1998)298 (5–6/1990)439 (5–7/1915)551 (5–8/1915)578 (4–8/1981)
4171 (7/1958)290 (5–6/1908)408 (5–7/1902)523 (5–8/1902)577 (4–8/1951)
5170 (5/1995)289 (6–7/1958) and 289 (5–6/1943)405 (5–7/1990)508 (4–7/2008)571 (4-8/1902)
6167 (6/1993) and 167 (6/1967)287 (6–7/1915)401 (6–8/1902)503 (5–8/1990)568 (4–8/1990)
7164 (5/2004)286 (6–7/1902)399 (6–8/1915)495 (5–8/1981) and 495 (5–8/1951)567 (4–8/1957)
8163 (5/1908)285 (5–6/1993), 285 (5–6/1935) and 285 (5–6/1915)396 (57/2008) and 396 (4–6/1947)491 (5–8/1905)562 (4–8/1944)
9162 (7/1992)281 (5–6/1957)395 (6–8/1951)483 (4–7/1957) and 483 (4–7/1915)561 (48/2008)
10160 (6/1928)279 (56/2008)391 (49/2008)482 (4–7/1896)559 (4–8/1995)
11157 (6/2008)    
Table 2. Return Periods for the 2008 Study Area's Total Precipitation Events Listed in Table 1
PeriodReturn Period (Years)
1 Month50
2 Months50
3 Months50
4 Months64
5 Months15

[17] The spatial patterns of precipitation received between May and June 2008 across the area reveals significant variation from one place to another (Figures 4a4c). During May, over 200 mm of precipitation was received over large portions of Nebraska, Kansas, southern Missouri, Illinois, and Indiana, totals that locally exceeded more than twice the normal (Figure 4a). The precipitation conditions peaked in June with large portions of southern Wisconsin, Iowa, Missouri, and Indiana receiving more than twice the normal amount of rainfall; southern Wisconsin received in excess of 300 mm, or 3 times the expected amount (Figure 4b). For the entire May–June period, over 300 mm of rainfall was received throughout the region, with the exception of portions of Illinois that received lesser amounts. Greatest totals, in excess of 400 mm, were received in southern Wisconsin, Iowa, southern Missouri, Kansas, and central Indiana (Figure 4c).

Figure 4.

Precipitation characteristics across the American Midwest during May and June 2008. (a) Total precipitation received in May 2008 expressed as percentage of 1971–2000 normal received in May 2008. (b) Total precipitation received in June 2008 expressed as percentage of 1971–2000 normal. (c) Total precipitation (mm) received between May and June 2008.

[18] Increased frequency of shorter-term, smaller-magnitude precipitation events can coincide with localized flooding and are typically depicted in the Midwest by the occurrence of precipitation events greater than 25 mm recorded over a 1 day period [Kunkel et al., 1994]. Between May and June 2008, a large number of locations experienced anomalously high frequency of such events. At this time of the year about one such event would be observed during any one given year. During the month of May 2008, between three and four such days were observed across Nebraska, eastern Kansas, southern Illinois, and Indiana (Figure 5a). During the month of June, the incidence of such events further increased across the region with greatest increases of up to 6 days recorded in Iowa northern Missouri, southern Wisconsin, central Illinois, and Indiana (Figure 5b).

Figure 5.

Number of days with 1 day precipitation totals exceeding 25 mm expressed as (a) total for May 2008 and (b) total for June 2008.

4.2. Weekly and Daily Precipitation Conditions

[19] Weighted area-averaged weekly precipitation amounts (by climate division) experienced throughout the Midwest region as specified by Figure 2 are shown in Figure 6. Total basin-wide areal rainfall amounts in excess of 25 mm over a 1 week period are considered significant and strongly correlated with regional flooding in the area [Changnon, 1996]. Such rates were observed throughout the region for 5 consecutive weeks between the last week in May and the end of June in 2008, a period that coincided with peak flooding events.

Figure 6.

Weekly mean areal precipitation (mm) (weighted by climate division), estimated across the American Midwest between 30 December 2007 and 9 November 2008. Solid line depicts precipitation of 25 mm per week.

[20] Coleman and Budikova [2010] identified specifically the period 21 May to 13 June 2008 as the 24 days when atmospheric conditions were conducive to producing the most significant amounts of precipitation over the region. During this time period, rain fell somewhere throughout the Midwest each day, and on each day, some area of the region received in excess of 25 mm of precipitation. Intense precipitation events, those exceeding 100 mm/d also were not, uncommon across the region during the period. For instance, such events were observed on 30 May throughout northern Iowa, on 3 June in the region bordering Illinois, Missouri, and Iowa; on 5 June, in east-central Illinois central Indiana and in southwestern Iowa; between 8–9 June and on 13 June over southern Wisconsin; and on 7 June more than 200 mm of rainfall fell over southern Indiana. Figure 7a shows the detailed spatial variation of the amount of rainfall that was received across the region during the peak 24 day period 21 May to 13 June. Greatest totals in excess of 300 mm were recorded throughout portions of Wisconsin, Iowa, Kansas, Missouri, and central Indiana. Precipitation totals for this time period with estimated return periods in excess of 250 years were highly concentrated at locations along the Illinois-Indiana boarder, southern Wisconsin, and northeastern Iowa (Figure 7b).

Figure 7.

Precipitation characteristics across the American Midwest during the major flooding period in 2008 between 21 May and 13 June: (a) total precipitation (mm); (b) total precipitation received, expressed as return periods (years); and (c) locations that received in excess of 100 mm of precipitation during a 7 day period. Circles represent 1 instance; stars represent 2 instances.

Figure 7.


[21] At the local level Kunkel et al. [1994] found that heavy rainfall events in excess of 100 mm over a 7 day period often coincide with flooding in small streams; single station heavy 7 day events are expected to occur only once per year. Figure 7c shows the 121 stations that recorded such events during the 24 day period between 21 May and 13 June. Several stations in eastern Nebraska and western Iowa recorded such events twice during this time. Six stations reported rain events that produced in excess of 250 mm in a 7 day period across southern Wisconsin and central Indiana.

[22] It is not surprising that numerous record-breaking rainfall totals were recorded throughout the region during the May–June flooding period in 2008. Iowa recorded the greatest amount of total rainfall for the May–June period on record since 1895, at 370 mm; the estimated return period of this value is estimated to be over 200 years. A total of 66 precipitation records were broken in May and 70 in June across the region (Figure 8). June record-breaking locations encompassed a significantly larger portion of the study area than that of May, typically before the middle of the month. Several historical all-time precipitation records were broken, 10 during May and 13 during June.

Figure 8.

Monthly record precipitation observed between May and June 2008 across the American Midwest. Filled circles denote records reached in May; squares denote records reached in June.

5. Significant Event Case Studies

[23] Several high-precipitation events that occurred in May and June 2008 were linked to intense local and regional convection cells, including mesoscale convective complexes (MCCs) and mesoscale convective systems (MCSs) that are known as significant precipitation producers in this region during the warm season [Kunkel, 1996]. The associated high spatial and temporal concentration of rainfall produced record flooding in several areas, events that carried significant social and economic consequences. Three such events are noted here including the historical flooding of Cedar Rapids in Iowa, the spilling of Lake Delton in southern Wisconsin, and record flooding in central Indiana with a special focus on Columbus, IN. We explore in detail the hydroclimatology, basic hydrology, and the mesoscale systems associated with the flooding in these regions. For each of the three case studies, the synoptic and mesoscale features responsible for the peak flooding were examined by inspecting hourly and daily synoptic surface maps, satellite imagery, and composite radar overlays for frontal type and position, mesoscale convective complexes (MCCs), mesoscale convective systems (MCS), squall lines, and more localized convection cells (see Bader et al. [1995] for a detailed discussion of weather interpretation from satellite and radar imagery). A timeline of significant features producing high localized precipitation amounts is presented. The hydroclimatological conditions associated with the flooding were examined using radar precipitation estimates in three distinct areas. These areas were defined in such a way that the estimation of recurrence intervals for the precipitation would be possible from Huff and Angel [1992] who calculate them for each climate division in the Midwest at various durations. As a result, the case study regions were defined using the boundaries of specific climate divisions (Figure 2). In each case, included were climate divisions that (1) contain the location of interest, (2) are located immediately to the east and west of the location to capture the general atmospheric circulation flow, and (3) are located immediately upstream from the location to best capture precipitation that would likely impact the local hydrology. The total amount of precipitation received during a particular time period was mapped and the mean areal total daily precipitation was estimated for each region and compared to daily discharge rates to understand the relationship between the hydroclimatology and local hydrology. Also, calculated and displayed in the graphs of Figure 9 are the upper 95% confidence limit for the mean areal total daily precipitation (equation image) to express the amount of geographic variability in the precipitation estimates as follows:

equation image

where σAP represents the standard deviation of the areal mean precipitation for a particular day. The value of σAP is used to represent population standard deviation, as the number of grids used to compute the estimates within each region is above 30 with a mean of 184, 417, and 577, across Indiana, Wisconsin, and Iowa, respectively [Burt et al., 2009].

Figure 9.

Hydroclimatic and hydrology conditions associated with significant flood-producing precipitation events in June 2008. Contour maps depict total precipitation received between (a) 3–12 June period across Iowa, (b) 5–9 June period across southern Wisconsin, and (c) 4–9 June period across central Indiana. The graphs display mean total areal daily precipitation (mm) shown as bars with the upper 95% confidence limit (whiskers) and mean daily discharge rates (m3/s) shown as dotted lines at USGS gauging stations: Figure 9a is for Cedar Rapids, IA; Figure 9b is for Wisconsin Dells, WI; and Figure 9c is for Columbus, IN. The locations of Cedar Rapids, Lake Delton, and Columbus are shown as stars on the maps. See Figure 1 for the situation of each region within the greater study area.

Figure 9.


Figure 9.


5.1. Cedar Rapids, Iowa

[24] Between 3 and 12 June, Iowa received several high-intensity precipitation events as a result of frequent storm development along a stationary front [USDOC, 2009]. On 3 June, the upper Midwest was experiencing cloudiness and lighter precipitation from a stationary front situated in southern Minnesota and southern Wisconsin. By 0800 UT (Zulu time), thunderstorms moved into western Iowa, increasing in intensity as they traversed eastward such that between 1200 UT and 1500 UT supercells and multicell thunderstorms were prevalent in central-eastern Iowa. Convective activity remained largely south of Iowa on 4 June until 2300 UT when the stationary front propagated slightly northward, bringing increasing cloudiness and sporadic thunderstorms into the early morning hours of 5 June. The stationary front largely persisted along the Iowa-Minnesota border for the next few days. On 6 June, a strong squall line ahead of a cold front moved in the Cedar Rapids area between 0600 UT and 0700 UT with localized convection remaining throughout the state until the following day. On 8 June at 0200 UT, a new line of thunderstorms was initiated along the stationary front in northern Iowa that moved southward, impacting central-eastern Iowa between 1100 UT and 1500 UT; a secondary thunderstorm wave followed at 2200 UT. Between 9 and 10 June, persistent cloudiness remained with some localized storm cells associated with a trough axis occurring between 0300 UT and 0700 UT on 10 June. New storms developed along the Nebraska-Iowa border at 1100 UT on 11 June, building into a possible MCS structure by 1200 UT. A strong line of thunderstorms ahead of an approaching cold front moved into western Iowa around 0000 UT on 12 June and impacted the Cedar Rapids area between 0400 UT and 0700 UT. The cold front slowed, stalling in eastern Iowa and produced new thunderstorms around 0200–0300 UT on 13 June before moving out of the area by 0600 UT. This last frontal wave was responsible for the record-breaking crest of the Cedar Rapids River.

[25] The mesoscale systems brought large amounts of precipitation into the area between 3 and 12 June (Figure 9a). Total precipitation amounts for the period reached in excess of 200 mm throughout most of Iowa and reached 300 mm upstream of Cedar Rapids along the Cedar River, where water accumulated in the local creeks and rivers and was transported south. Huff and Angel [1992] estimate the return period for a 10 day event in excess of 300 mm to be over 100 years in this region of Iowa; the long-term expected amount of precipitation (i.e., totals with a threshold of a 1 year return period) to be received for a 10 day duration is about 100 mm. Greatest amounts of precipitation were received across the region between 8 and 9 June (Figure 9a). A total of 12 precipitation records for the month of June were reached during this time period. Cedar Rapids was one of the most devastated areas within the Midwest region by the resulting floods; flood waters in the city affected over 7700 properties covering more than 25 km2 of downtown Cedar Rapids [City of Cedar Rapids, 2009]. On 13 June, stage levels at the Cedar Rapids gauging station reached in excess of 9.5 m, breaking the previous record set in 1851 by more than 3 m [NWS, 2010]. The daily discharge rate hydrograph for the Cedar River at Cedar Rapids displays a sharp increase in discharge rates beginning on 8 June that peak on June 13 when the historical stage record was broken (Figure 9a). Discharge rate anomalies at the Cedar Rapids station along the Cedar River were observed to be 775% of the normal for the month of June. This peak discharge rate exceeded the June normal by 2592 m3/s; the estimated return period for the discharge rate was 2000 years for the month of June.

5.2. Lake Delton, Wisconsin

[26] Lake Delton is a man-made, fresh water lake constructed in 1927 in southern Wisconsin near Wisconsin Dells [Goc, 1999]. On 9 June, the northeast bank of the lake failed as a result of heavy precipitation received since 5 June. The precipitation was generated by similar synoptic and mesoscale conditions that were occurring around Cedar Rapids. A stationary front along the Wisconsin-Illinois border on 5 June initiated a line of thunderstorm cells in northern Iowa around 0300 UT and propagated eastward. Strong multicell thunderstorms were prevalent in the Lake Delton region between 0900 and 1100 UT, but lighter precipitation from dissipating storms remained through 1400 UT. On 6 June, storms persisted along the stationary front that stretched from Nebraska to western New York but were largely confined to northern Wisconsin until 1000 UT. At that time a strong squall line ahead of an approaching cold front moved into southern Wisconsin and brought heavy precipitation. Some localized thunderstorm cells remained in the area until 2300 UT when the cold front finally left Wisconsin. On 7 June, the stationary front reemerged and new wave of thunderstorms in Minnesota and Wisconsin developed over the next 2 days. A large supercell moved into southern Wisconsin at 1800 UT, impacting the Lake Delton region between 1900 and 2100 UT. Numerous squall lines of varying precipitation intensity moved across southern Wisconsin on 8 June, the strongest between 0300 and 0400 UT with two weaker storm lines occurring between 1400–1500 UT and 2200–2300 UT. Precipitation did not completely cease in the Lake Delton region until approximately 0600 UT on 9 June.

[27] During the 5 day period between 5 and 9 June, over 225 mm of rain fell around Lake Delton, a precipitation amount with an estimated return period of almost 100 years; the long-term expected precipitation (i.e., totals with a threshold of a 1-year return period) for this duration period is about 75 mm [Huff and Angel, 1992]. The majority of rainfall events arrived between the 8 and 9 June (Figure 9b). A total of 19 daily precipitation records were reached throughout southern Wisconsin during this time period. Wisconsin Dells surpassed their previous records on both days. The resulting impact on the local hydrology is evident through the records of daily discharge rates for the Wisconsin River at the Wisconsin Dells station located directly south of the location where Lake Delton overflowed. The rates show a sharp increase on 9 June (Figure 9b) when they reached over 500 m3/s, twice the normal values typically expected for the month of June; the value ranked at the 92nd percentile of all daily June records. The failure of the northeast bank of Lake Delton released 2635 million L of water from a 267 acre lake flooding out a 120 m portion of County Hwy A and five homes discharging the debris into the Wisconsin River [Foley, 2008; Hesselberg and Seely, 2008; Romell and Marley, 2008]. The water overtopped the narrow strip of land located between the lake and the Wisconsin River located approximately 240 m down-gradient. Once the northeast bank of Lake Delton collapsed, a surge of water created a 120 m wide channel that quickly drained the lake.

5.3. Columbus, Indiana

[28] The 6 day period between 4 and 9 June 2008 produced a majority of the rainfall that resulted in flooding throughout central and southern Indiana. As in Iowa and Wisconsin, Indiana precipitation was largely influenced by storm development along a persistent stationary front. High-precipitation events in central Indiana occurring 7–9 June produced the most intensely localized rainfall that resulted from unique mesoscale conditions. On 7 June at 0000 UT, an approaching cold front from the west stalled along an axis from southern Missouri to northern Illinois. The thunderstorms ahead of the cold front moved from a more linear to cluster formation between 0300 and 0400 UT, such that by 0500 UT, an MCC structure centered over Indiana was readily apparent on satellite IR imagery. Storm clusters languished and strengthened within the MCC, particularly around the Illinois-Indiana boarder between 0600 and 1200 UT. Convective activity did not leave the central Indiana region until 1500 UT, and nearly all storms in Illinois and Indiana dissipated by 2000 UT. The following day was dominated by dry conditions in Indiana as the southeastern high moved into the area pushing the weakened low-pressure system northwestward into southern Canada. A second cold front moved into central Illinois on 9 June. By 1900 UT, supercell and multicell thunderstorms were situated along the central Illinois-Indiana border and by 2000–2100 UT moved out of Indiana into northwestern Ohio. A secondary frontal wave developed around 0200 UT on 10 June, and precipitation across central Indiana was generally lighter with some small local thunderstorms. Extensive cloudiness dominated the region until 1500 UT when conditions began to clear with the high pressure behind the front.

[29] The total amount of precipitation received 4–9 June across central Indiana shows a highly concentrated pattern of precipitation (Figure 9c). The 6 day totals reached in excess of 300 mm east of Columbus and over 150 mm upstream. Huff and Angel [1992] estimate the total rainfall for a 5 day duration even with a 100 year return period at around 220 mm for this region. From 6–7 June, an excess of 254 mm fell across central Indiana, with estimated return intervals of 1000 years [Morlock et al., 2008]. Six June precipitation records were set between 7 and 8 June across the region. The resulting impact on local hydrology was evident between 7 and 9 June as severe flooding occurred throughout central Indiana. The communities most impacted included Columbus, Edinburgh, Franklin, Martinsville, Newberry, Paragon, Seymour, Spencer, and Worthington. Columbus was the hardest hit as all roads into the town were closed isolating the community; about 15% of all structures within the town were flooded [Morlock et al., 2008]. On 7 June, the East Fork of the White River at Columbus rose to near-record flooding within a 6 h period [Morlock et al., 2008]. The discharge rates at the Columbus, IN, gauging station peaked at 1654 m3/s on 8 June reaching an all-time daily discharge record; the expected June historical mean discharge rate is typically 55 m3/s (Figure 9c).

6. Discussions and Conclusions

[30] This study explores the hydroclimatic aspects that played a part in the development and persistence of the floods that occurred in the American Midwest between May and June 2008. Synoptic and mesoscale atmospheric circulation patterns and processes funneled moist and warm air efficiently from the west and the Gulf of Mexico into the Midwest producing large amounts of precipitation over the GUMRB and Wabash River Basin. In combination, these conditions produced extreme precipitation that lead to flooding on tributaries and major river systems during a 24 day period between 21 May and 13 June across the entire state of Iowa, southern Wisconsin, and central Indiana.

[31] The hydroclimatic analysis of the precipitation characteristics reveals five (5) distinct conditions that joined to create the unique nature of the 2008 flooding. The occurrence of anomalous hydroclimatic conditions over the American Midwest during the late spring/early summer season contributed to the 2008 flood event's unusual nature. The 2008 flooding was relatively short-lived, geographically concentrated, and especially unusual for the May–June period. Knox [1988] pointed out that the climatological probability of flooding across the Upper Mississippi River Valley is largest between March and July with the smallest number of events within this time period occurring during the month of May marked by a seasonal transition position. Snowmelt runoff plays a significant role in large-scale flooding in early spring along the main channel of the Mississippi River and its tributaries, peaking in March. Excessive and/or high-intensity rainfall of sufficient duration cause flooding, including localized flash floods, in the summer peaking in June. Summer flooding most often peaks in June as the influx of warm and moist air from the Gulf of Mexico to the area is sufficiently large and the surface is yet to be completely covered by vegetation.

[32] The specific factors important to the flooding in 2008 include (1) persistent high surface soil moisture conditions prior to flooding exacerbated by anomalously high rainfall; (2) extreme rainfall totals covering extensive areas; (3) increased frequency of shorter-term, smaller-magnitude events covering large areas; (4) persistent multiday heavy precipitation events; and (5) extreme flood-producing rain storms. The details as well as the timing of incidence of each characteristic can be summarized as follows:

[33] 1. August 2007 to April 2008: Wet antecedent conditions: Antecedent precipitation, air temperature, and soil moisture conditions played a significant role fostering a unique combination of anomalous surface and atmospheric conditions that ultimately lead to the flooding. Intermittent flooding was recorded in the region starting in March and April. Coleman and Budikova [2010] found unseasonably wet soil moisture conditions and cool temperatures throughout the region during the winter and early spring leading up to the flooding; the wet conditions coupled with cooler-than-normal temperatures acted to minimize surface evapotranspiration rates keeping soils wet or near saturation. The wet surface conditions were caused by anomalously high rainfall observed during 7 of the 10 months preceding the flood outbreak; for 4 months precipitation, totals exceeded 150% above normal throughout the American Midwest.

[34] 2. May–June 2008: Extreme rainfall totals: Rainfall totals reached over 400 mm or more than twice the expected amount between May and June across majority of the study area with largest and often record-breaking amounts of precipitation observed in southern Wisconsin, throughout the state of Iowa, in southern Missouri, and in central Indiana at the peak phase of flooding between the end of May and middle of June.

[35] 3. May–June 2008: Increased frequency of shorter-term, smaller-magnitude rainfall events: Between May and June, the American Midwest area recorded a high incidence of moderate rainfall events as measured by the number of days with 1 day precipitation totals exceeding 25 mm. Two to 6 times the expected number of occurrences was recorded in the region between May and June.

[36] 4. 21 May to 13 June 2008: Persistent and heavy/intense local rainfall: The region received an average of more than 30 mm of precipitation each week for five consecutive weeks between the last week in May and middle of June. Between 21 May and 13 June, rain fell somewhere in the region each day. Over 100 stations in the region reported total precipitation in excess of 100 mm during a 7 day period at least once during this 24 day period; there were eight instances where two such events were recorded. Six stations reported rain events that produced in excess of 250 mm in a 7 day period in southern Wisconsin and southern-central Indiana. Intense precipitation as demonstrated by events exceeding 100 mm/d also was not uncommon across the region during the period.

[37] 5. 21 May to 13 June 2008: Extreme flood-producing regional rainstorms: Several high-precipitation events that occurred in early June 2008 were linked to intense local and regional convection cells that formed along a stationary front, including mesoscale convective complexes (MCCs) and mesoscale convective systems (MCSs) as demonstrated by the case studies of Cedar Rapids in Iowa, Lake Delton in southern Wisconsin, and Columbus in Indiana.

[38] The 2008 flood event can be tied to a unique combination of anomalous antecedent and concurrent conditions, some of which may be tied to global warming. The United States has seen a nationwide 7% increase in annual precipitation over the past century with largest changes recorded over the past 30 years [Groisman et al., 2004]. Significant increasing trends of almost 20% have been detected in heavy and flood-prone precipitation since the early 1900s across the country, especially during the warm season in the American Midwest [Kunkel et al., 2003, 2008]. Scientists have attributed such changes to the observed increases in atmospheric water vapor and the general intensification of the water cycle [Kunkel et al., 2008] associated with anthropogenically induced climate change [Groisman et al., 2005]. The occurrence of the 2008 flood event raises the question of whether its occurrence provides further evidence for a changing character of Midwestern hydroclimatology due to anthropogenic influences. If indeed it does and the frequency of floods may be on the rise, the consequences for engineering and hydrologic planning will be considerable.


[39] The authors would like to thank the reviewers for their comments that significantly improved our work.