Missouri and other parts of the Midwest are experiencing severe flooding, perhaps historic flooding. The record flood in this part of the country occurred in 1993. According to Chris Boerm, transportation manager for Archer Daniels Midland, the 1993 flood was concentrated in Iowa and the upper Midwest. This one is more expansive, affecting the entire Mississippi River, the Arkansas River, the Illinois River, the Ohio River, and the Missouri River. (quoted in Sullivan, Singh, and Bloomberg, 2019). Some 203 river gages along U.S. rivers are at or above flood stage.
With every flood, it seems, we hear a chorus complaining that flooding is getting more severe, and that our efforts to manage our major rivers have actually made things worse. Flood plains upstream act as sponges, absorbing flood water and then releasing it slowly over time, thus reducing the severity of flooding downstream. But levees along the river prevent this from happening, funneling all of the water downstream, worsening flooding there.
I thought I would look and see if, indeed, there is a trend towards increased flooding, and if so, how severe it was.
First I decided that I would focus only on Missouri. Then I decided that I would focus only on select rivers that represented diverse geographical areas of the state. Then I decided I would focus only on rivers that were relatively major rivers. And finally, I decided that I would eliminate rivers that I felt were almost entirely controlled by dams. The White River, for instance, is one our longer rivers, though it only flows through Missouri for part of its length. While in Missouri, it is impounded by 3 reservoirs: Table Rock Lake, Lake Tanneycomo, and Bull Shoals Lake. So, I eliminated it, and other similar rivers. I did not, however, eliminate the Missouri and the Mississippi. Though those rivers are regulated by dams and impounded into reservoirs, their many floods indicate that they are not almost entirely controlled by anything.
But how to measure flooding? I decided to use two measurements routinely made by the United States Geological Survey at thousands of river gages, which cover every major river in the country: peak streamflow, and peak gage height. Peak streamflow is the highest amount of water flowing down the river at any given time during a water year (water years begin in the summer). Peak gage height represents the highest the river is during a water year. These two measurements are not specific indicators of flooding. However, high readings go along with flooding, and if these two measurements are increasing, it would provide support for the idea that floods are getting worse.
Figure 1 shows a map of the river gages I selected for my study. They included gages on the Mississippi River at Grafton and at Thebes, a gage on the Missouri River at Kansas City, gages on the Meramec River near Steelville and near Eureka, a gage on the Gasconade River at Jerome, a gage on the Grand River at Sumner, a gage on the Pomme de Terre River at Polk, and a gage on the Current River at Van Buren.
Each gage has historical data for peak streamflow and peak gage height for each water year. How far back the data goes varies between gages. I turned this data into graphs, shown as Figures 2-10. For each graph, streamflow is shown in orange, and should be read against the left vertical axis. Gage hight is shown in blue, and should be read against the right vertical axis. I had Excel drop linear regressions on each of the lines, to show the trend over time. They are shown as dotted lines. I will discuss the results after sharing the charts.
(To view a chart, click on it. Once a chart is open, you may cycle through the charts by using the buttons below the charts. To return to this post from the charts, click on the name of the post under the chart.)
As one considers the charts as a group, the most obvious thing that jumps out is the large variation in streamflow from year-to-year. This is particularly evident on smaller streams that don’t gather precipitation from large drainage areas. The Grand River, for instance, had a minimum streamflow of 6,320 cfs in 2003, but a maximum streamflow of 180,000 cfs in 1947. The maximum streamflow was more than 28 times the minimum. However, even on the big rivers the yearly variation was large: on the Mississippi River at Thebes, the minimum was 140,000 cfs in 1934, while the maximum was 1,050,000 in 2016 (7.6 times the minimum).
There are 18 trend lines: 2 lines for each of 9 gage locations. All but 1 show an increasing trend over time. The only trend that isn’t upward is streamflow on the Meramec River near Steelville. I’m not sure what this means, as the gage height there does trend up, and both streamflow and gage height on the Meramec near Eureka also trend up. Eureka is downriver from Steelville. This one finding notwithstanding, with 17 out of 18 trending upward, I think it is safe to say that both streamflow and gage height have been increasing over time in Missouri.
Don’t read too much into the steepness of the different trendlines, they are determined by the scales Excel chose for the vertical axes.
At each location peak streamflow and peak gage height tend to vary within a limited range, but this range is broken in some years by extremes. Even high values in the normal range may go along with flooding in some locations, but the extremes probably indicate more severe flooding. If there is an upward trend in the normal range, it may indicate a trend toward increased minor flooding. But if there is an increase in the extremes, it may indicate that extreme flooding is getting even more extreme. And that is what we find. On most of the charts, the extreme peaks on the right are taller than the extremes on the left.
Put this together with increased development in flood plains, and yikes! The levees better hold!
The trend is not universal, however, and one of the locations that turned out to be more complex was the Missouri River at Kansas City. The highest streamflow there occurred in 1951, and streamflows since then (even in 1993) were lower. Gage height, however, peaked in 1993. The series of dams on the Missouri River were completed in 1962, and they may have moderated streamflow since then. (Although when flooding is extreme, the dams have to dump water to prevent themselves from being overtopped, and that can make things worse. See my posts on Oroville Dam.)
(Added note 6/27/19: This may actually be an effect of levee building. Levees constrict the width of the river during high water. If the river width is sufficiently narrowed, the gage level might be considerably higher, but the river might still be carrying less water.)
So, it was a lot of work to find this data and put these charts together. But they do tend to support the notion that the peak streamflow and the peak level of Missouri’s rivers are increasing over time, and that the severity of especially severe events is, too. I have heard this trend attributed to both levee building and climate change, but this data does not speak to causation.
Sullivan, Brian K., Shruti Date Singh, and Mario Parker Bloomberg. 2019. “Hundreds of Barges Stalled as Floods Hider Midwest Supplies.” St. Louis Post Dispatch, 6/10/2019. Viewed online 6/10/2019 at https://www.stltoday.com/news/local/metro/hundreds-of-barges-stalled-as-floods-hinder-midwest-supplies/article_5a0355ea-3c03-584e-b3df-7a669205176d.html#tracking-source=home-top-story-2.
United States Geological Survey. National Water Information System: Mapper. I used the map to select the river gages for this article 6/10/2019 at https://maps.waterdata.usgs.gov/mapper/index.html.
United States Geological Survey. Peak Streamflow for the Nation. This is a data portal. I downloaded the data for the 9 river gages in this article on 6/10/2019 from https://nwis.waterdata.usgs.gov/usa/nwis/peak.