Abstract
From mid-June through early August 1993, flooding was severe in the Upper Mississippi River Basin following a wet-weather pattern that persisited over the area for at least 6 months before the flood. The magnitude and timing of several intense rainstorms in late June and July, combined with wet antecedent climatic conditions, were the principle causes of the flooding.
Flood-peak discharges that exceeded the 10-year recurrence interval were recorded at 154 streamflow-gaging stations in the upper Mississippi River Basin. At 42 streamflow-gaging stations, the peak discharge was greater than the previous maximum known discharge. At 14 additional gaging stations, peak discharges exceeded the previous maximum regulated peak discharge. At 46 gaging stations, peak discharges exceeded 100-year recurrence intervals.
INTRODUCTION
From mid-June through early August 1993, severe flooding in the upper Mississippi River Basin (fig. 1) followed heavy and persistent rainfall from January through July. The flooding was unusual because it came so late in the spring-summer runoff season and because of the large number of streamflow-gaging stations that had record or near-record peak discharges. Record peak discharges were recorded from mid-June through early August at U.S. Geological Survey (USGS) streamflow-gaging stations in the Minnesota River Basin in Minnesota; in the Iowa, the Skunk, the Des Moines, the Little Sioux, and the Nishnabotna River Basins in Iowa; on the Mississippi River at Keokuk, Iowa; in the James River Basin in North and South Dakota; in the Platte River Basin in Nebraska; in the Kansas River Basin in Kansas; in the Grand River Basin in Missouri; and along the Missouri from St. Joseph to Booneville, Missouri. Unussually high flood discharges were recorded at other locations throughout the area of flooding. The flooding was also unusual for its long duration and widespread and severe damage. At St. Louis, Missouri, the Mississippi reached flood stage on June 26 and was still above flood stage in mid-August. Millions of acres of agricultural and urban lands in the upper Mississippi Basin were inundated for weeks, and unofficial damage estimates exceeded $10 billion.
Figure 1: The Mississippi River Basin and general area of flood streams, June to August 1993.
Purpose and Scope
As the principal Federal agency responsible for the collection of streamflow data, the USGS operates a network of about 7,300 continuous-reocrd streamflow- gaging stations throughout the Nation. These streamflow records, some of which extend back to the 19th century, form the basis for the discussion of the 1993 flood in this report. This report describes the flooding in the upper Mississippi River Basin from mid-June through early August 1993 and the peak discharges at selected sites. Peak discharges for the 1993 flood are compared with previous maximum discharges and, for selected sites with long periods of record, are graphically compared with all previously known annual peak dischrages.
Flood Recurrance Interval
For comparitive purposes, flood-peak discharges are referenced to a specific recurrence interval or probability of occurance. The recurrence interval is the average number of years between occurences of annual peak discharges that equal or exceed a specified discharge. For example, a discharge that has a 100-year recurrance interval is so large that an equal or greater annual peak discharge is expected, on average, only once in any 100-year period. Because of the random nature of flood events, the times between annual peak discharges of a certain magnitude are far from uniform; a large flood in one year does not preclude the occurence of an even larger flood the next year. In any given year the annual peak dischrage has 1 chance in 100 of equalingg or exceeding the 100-year flood (U.S. Interagency Advisory Committee on Water Data, 1982).
Recurrence intervals for the 1993 flood peaks presented in this report are genreally determined by using the most current published USGS flood-frequency reports for the States in the area of flooding. For Minnesota, South Dakota, Nebraska, and Kansas, however, published flood-frequency reports fo not include recent data; consequently, the recurrence intervals for the 1993 peak discharges are based on unpublished flood-frequency analyses archived in the USGS District offices in these States. Recurrence intervals for the 1993 peak discharges on the Kansas River, the Missouri River, and the Mississippi River are based on unpublished flood-frequency analyses completed by the U.S. Army Corps of Engineers (Gary Dyhouse, St. Louis District, U.S. Army Corps of Engineers, 1993, written commun,; Jerry Buehre, Kansas City District, U.S. Army Corps of Engineers, 1993, written commun.).
For this report, recurrence intervals have been categorized in ranges as follows: from 10 to 50 years, greater than 50 to 100 years, and greater than 100 years. Flood frequency relations for streamflow-gaging stations in the area of flooding are expected to be updated on the basis of 1993 peak-discharge data.
CAUSES AND CHRONOLOGY OF FLOODING
The are of significant flooding in the upper Mississippi River Basin from mid-June to early August 1993 includes southern Minnesota, southwestern Wisconsin, Iowa, western Illinios, Northern Missouri, southern North Dakota, and eastern parts of South Dakota, Nebraska, and Kansas.
Climatic Conditions
The areal extant and magnitude of the 1993 Mississippi River flood was due to the persistent wet-weather pattern that was throughout the upper Midwestern United States for at least 6 months preceding the flood. This persistent pattern resulted from an eastward flowing jetstream that extended from central Colorado northeastward across Kansas to northern Wisconsin (National Weather Service, 1993).
Because of this jetstream, a weather-front convergence zone across the upper Midwestern United States during the spring and summer months that preceded the flood. Moist, warm air from the Gulf of Mexico was drawn northward along this jetstream where it collided with cooler air masses drawn out of central Canada. The resultant unstable air masses persisted throughout the flooded area during spring and summer 1993 (National Weather Service, 1993). The climatic conditions that preceded this flood were remarkably similar to those that preceded the last large flood on the upper Mississippi River in April 1973 (Chinn and others, 1975).
Antecedent Hydrologic Conditions
In many parts of the flooded area, rainfall totals for January through July 1993 were one and one-half to two times the January-through-July normal precipitation for 1961 to 1990 (David Miskus, National Weather Service, 1993, written commun.; fig. 2). For Iowa, which is centrally located in the area of flooding, November 1992 to April 1993 was the second wettest November-to-April period in 121 years of record. Intense rainfall in late March and early April caused serious, localized flooding in the Cedar River Basin in Iowa, and the Big Sioux River Basin in South Dakota.
Figure 2:Areal distribution of total precipitation in the area of flooding in the upper Mississippi River Basin, January to July 1993.
Rainfall and runoff were above average throughout the flooded area from April toJune 1993, and streams were generally bankfull at the end of June. The monthly discharge volumes for the Mississippi River at Keokuk for April through June were more than 200 percent of normal (April-June 1961-90). By late June, most flood-control reservoirs in the upper Mississippi were at or near capacity, and soils throughout the flooded area were saturated from excessive precipitation.
Chronology of the June to August Flooding
The magnitude and timing of several rainstorms in late June and July, combined with wet antecedent climatic conditions, were the principal causes of the severe flooding in the upper Mississippi River Basin. To illustrate the influence of the timing of runoff from these storms on the peak discharges and their dates of occurrence for selected streamflow-gaging stations in the general area of flooding are shown in figure 3.
Figure 3: Peak discharges (Q) and dates of occurence for the 1993 flood at selected streamflow-gaging stations in the upper Mississippi River Basin.
During June 17-18, 2 to 7 inches of rain fell throughout southern Minnesota, northern Iowa, and southwestern Wisconsin. Runoff from this storm caused flooding on the Minnesota on the Mississippi Rivers in Minnesota and the Chippewa and the Black Rivers in Wisconsin. As a result of these floodwaters, the discharge of the Mississippi River at Clinton, Iowa peaked on July 8, 1993 (fig. 3).
Two separate storms in early June cause large-scale flooding in Iowa. During the first storm on July 5, 2 to 5 inches of rain fell in central Iowa and caused lowland flooding on the Iowa, the Skunk, and the Des Moines Rivers. During the second storm on July 8-9, 2 to 8 inches of rain fell in central Iowa. Rivers throughout central Iowa had not receded from the July 5 storm, and the three major reservoirs in this part of the State were at capacity. The runoff from this storm, combined with the runoff from the July 5 storm, caused record or near-record peak discharges at streamflow-gaging stations throughout the Iowa, the Skunk, the Raccoon, and the Des Moines River Basins. The floodwater from these rivers entered the Mississippi River at about the same time as the flood peak from the late June storm in northern basins reached Keokuk (fig. 3). The coincident timing of the flood peaks from these tributary rivers increased the peak discharge on the Mississippi River and aggravated flooding on the Mississipi On July 15-16, 2 to 7 inches of rain fell in eastern North Dakota and western Minnesota and caused flooding in the upper reaches of the Minnesota River Basin in Minnesota and the James River Basin in North Dakota. Although peak discharges from this storm were not as large in the lower reaches of these basins as the peak discharges of late June, the floodwaters from the james River added to the flooding of late July on the Missouri River.
From July 22-24, 2 to 13 inches of rain fell in parts of Nebraska, Kansas, Missouri, Iowa and Illinois. The runoff from this storm caused record peak discharges on the Platte River in Nebraska and contributed large flows to previously filled reservoirs in the Kansas River Basin in Kansas. Discharges also were near-record on the Nishnabotna River in Iowa and the Illinois River in Illinois. Peak discharges on the Kansas River were the largest since 1951, which is before significant river regulation began.
Before the July 22 to 24 storm, the Missouri River was at or near flood stage as a result of large tributary inflows earlier in the month from the James River in North and South Dakota, the Big Sioux River in South Dakota, and the Little Sioux River in Iowa. As a result, floodwaters from the Platte and the Kansas Rivers caused record to near-record peak discharges on the Missouri River at streamflow-gaging stations downstream from the confluence of the Platte River. The flood peak on the Missouri River reached Hermann, Missouri, on July 31 (fig. 3). The peak discharge from the Missouri River caused a second and greater peak discharge at the streamflow-gaging station on the Mississippi River at St. Louis on August 1.
Flood conditions on the Mississippi River differed above and below the confluence of the Ohio River. At Thebes, Illinois, just upstream from the confluence, severe flooding on the Mississippi River peaked on August 7. Downstream from the confluence, flooding on the Mississippi River was not severe because of the less-than-average discharge contributed by the Ohio River and a substantially larger channel capacity in this reach of the Mississippi River. The discharge of the Ohio River was less than average during July and August as a result of generally dry conditions and low reservoir outflows throughout the Ohio River Basin.
PEAK DISCHARGES FOR THE 1993 FLOOD
The locations of 154 streamflow-gaging stations and the ranges of computed recurrence intervals for the 1993 peak discharge are shown in figure 4.
Figure 4: Locations of selected streamflow- gaging stations and ranges in recurrence interval for the 1993 peak discharges in the upper Mississippi River Basin.
Comparison With Historic Floods
Forty-two streamflow-gaging stations in the area of flooding had record peak discharges in 1993. In addition, 14 other stations had peak discharges in 1993 that exceeded the previous maximum regulated discharge. At 46 stations, peak discharges in 1993 had recurrence intervals of greater than 100 years.
Historic annual peak discharges and 1993 peak discharges at four streamflow-gaging stations that have long record of discharge are shown in figure 5. As shown in the figure, the 1993 peak discharge for the Raccoon River at Van Meter, Iowa is remarkable because it was almost twice as large as any previous peak discharge in almost 80 years of record.
Figure 5: Historic peak discharges and peak discharges for the 1993 flood at selected streamflow-gaging stations in the upper Mississippi River Basin.
At two sites on the Mississippi River, the 1993 peak discharges were record or near-record discharges. The 1993 peak discharge for the Mississippi River at Keokuk was substantially greater than previous record peak discharges in 1973 and 1851. Although the 1993 peak discharge for the Mississippi River at St. Louis was not a record discharge, it was substantially larger than that for the large flood of 1973. The 1993 peak discharge for the Missouri River at St. Louis was also slightly greater than that is 1903 and only slightly less than the estimated record peak discharge in 1844.
The 1993 peak discharge for the Missouri River at Booneville, Missouri, also was a record discharge, which exceeded previous historic flood-peak discharges in 1951, 1903, and 1844. Because floodwaters upstream from Lewis and Clark Lake were contained in 1993 by the Missouri River dam system, the large peak discharge at Booneville is particularly notable.
At all four sites shown in figure 5, the 1993 peak discharge was greater than the discharge with a 100-year recurrence interval. At the Mississippi River at St. Louis, the 1993 peak discharge was slightly greater than the discharge with the 100-year recurrence interval. At the Raccoon River at Van Meter, however, the 1993 peak discharge was about 40 percent greater than the discharge with a 100-year recurrence interval.