Currently browsing category


Anne’s top papers of 2016 + 3 she co-wrote

Yesterday, I posted an epic analysis of my scientific reading habits in 2016, but I didn’t tell you about the papers I read last year that made my heart sing. And I didn’t take much time to brag about my own contributions to the scientific literature. So I’m going to rectify that omission today.

My top 3 papers of 2016 are (in no particular order):

Of rocks and social justice. (unsigned editorial) Nature Geoscience 9, 797 (2016) doi:10.1038/ngeo2836

The whole thing is absolutely worth reading (and it’s not behind a paywall) but here’s where it really starts to hit home:

Two main challenges stand in the way of achieving a diverse geoscience workforce representative of society: we need to attract more people who have not been wearing checkered shirts, walking boots and rucksacks since secondary school, and we need to retain them.

Waters, C. N., Zalasiewicz, J., Summerhayes, C., Barnosky, A. D., Poirier, C., Ga?uszka, A., … & Jeandel, C. (2016). The Anthropocene is functionally and stratigraphically distinct from the Holocene. Science, 351(6269), aad2622.

Want an up-to-date, data-rich, and condensed summary of why many scientists think it is time for a new geologic epoch? This is the paper to read.

Wu, Q., Zhao, Z., Liu, L., Granger, D. E., Wang, H., Cohen, D. J., … & Zhang, J. (2016). Outburst flood at 1920 BCE supports historicity of China’s Great Flood and the Xia dynasty. Science, 353(6299), 579-582.

I am a sucker for a good mega-paleo-flood story, and this one ticks all of the right boxes. An earthquake generates a landslide, which dams a river, and then fails, resulting in one of the largest floods of the last 10,000 years and alters the course of Chinese history. Geology, archaeology, and history combine in this compelling story.

Plus, a bonus paper, that was definitely one of the best papers I read in 2016.

Shields, C., and C. Tague (2015), Ecohydrology in semiarid urban ecosystems: Modeling the relationship between connected impervious area and ecosystem productivity, Water Resour. Res., 51, 302–319, doi:10.1002/2014WR016108.

I’m cheating a little bit here, because this paper came out in 2015. But I read this paper in 2015, and then I read it twice more in 2016. That’s how much I like it. Why? Because it’s a really nice illustration of how physically-based models can reveal the complex and unexpected ways that ecosystems and watersheds respond to urban environments. In a semi-arid environment, deep rooted vegetation can take advantage of the bonus water that gets delivered from rooftop downspouts that drain out onto the land. The additional water use boosts net primary productivity, potentially enough to offset the loss of productivity that occurred when parts of the landscape were paved and built upon. But while deep rooted vegetation, native to the semi-arid landscape, can take advantage of the bonus water, grass can’t. It’s a cool story, with implications for the way we develop and manage urban landscapes – and the way we model them. (This paper is open access as of January 1, 2017!)

I was thrilled to be able to contribute to 3 papers in 2016. 

Turner, V.K., Jarden, K.M., and Jefferson, A.J., 2016. Resident perspectives on green infrastructure in an experimental suburban stormwater management programCities and the Environment, 9(1): art. 4.

In 2015, my team published a paper showing how the installation of bioretention cells, rain gardens, and rain barrels on a residential street in the Cleveland area substantially decreased stormwater runoff. This paper represents the other side of the story – the side that is, just as important (if not more so) – how the people on the street responded to the addition of this green infrastructure. In short, getting residents on board with stormwater management is a big challenge that we’re going to face as we scale-up from demonstration projects to widespread deployment of these technologies. (This paper is open access and free to all.)

Bell, C.D., McMillan, S.K., Clinton, S.M., and Jefferson, A.J., 2016. Hydrologic response to stormwater control measures in urban watershedsJournal of Hydrology. Online ahead of print. doi: 10.1016/j.jhydrol.2016.08.049.

Bell, C.D., McMillan, S.K., Clinton, S.M., and Jefferson, A.J., 2016. Characterizing the Effects of Stormwater Mitigation on Nutrient Export and Stream ConcentrationsEnvironmental Management. doi:10.1007/s00267-016-0801-4

I’m thrilled that first author Colin Bell completed his doctorate in 2016 and got two papers out to boot. These papers are the culmination of 5 years of research in Charlotte, North Carolina. In the Journal of Hydrology, we try to disentangle the effects of stormwater management from the overall signal of urbanization across 16 watersheds. It turns out that for the level of stormwater management we see in the real world, it’s not enough to counter-act the effects of impervious surfaces (pavement and rooftops) as a driver of the hydrologic behavior of urban streams. In Environmental Management, we aim to understand the influence of stormwater ponds and wetlands on water quality in the receiving streams. This turns out to be quite tricky, because the placement of stormwater management structures spatially correlates with changes in land use, but based on differences in concentration between stormwater structure outflow and the stream, we show that it should be possible. This echoes the findings from our 2015 paper using water isotopes to understand stormwater management influences at one of the same sites. Colin will have another paper or two coming out of his modeling work in the next year or so, and we’re still analyzing more data from this project, so keep your eyes out for more work along these lines.

Brock Freyer defends his MS on the Mighty Mississippi

Two people, standing behind a boat, with river and bluffs in the background.

Brock and Anne at the end of field work on the Mississippi River, July 2008.

Today, Brock Freyer will be defending the results of his M.S. research. The title of his research project is: Fluvial Response to River Management and Sediment Supply: Pool 6 of the Upper Mississippi River System, Southeastern Minnesota.

Brock’s committee is composed of Anne Jefferson (advisor), John Diemer and Ross Meentemeyer.

The defense is on Tuesday April 23, 2013, at 1:30 pm in McEniry 307 of UNC Charlotte. As Brock is currently located in Alaska, this will be a Skype defense. All are welcome to attend.


In this age of environmental restorations and rehabilitations, the scale and extent of projects have been getting larger and more expensive. In the Upper Mississippi River System (UMRS) the U.S. Army Corp of Engineers (USACE) has begun the task of restoring the negative effects that over a century of river management has incurred. Due to the scale and cost of such projects, it is essential to understand the natural and human processes that have affected the river system. In the UMRS, erosion and land loss are considered the dominant geomorphological trend, but Pool 6 of the UMRS is an exception to this norm. In Pool 6, deposition and land growth in recent decades have allowed the river morphology to begin reverting to its condition prior to intense river management. Through the application of varied chronological data sets within ArcGIS, spatial variations were measured to better understand where and why changes have occurred. A nested study area approach was applied to Pool 6 by dividing it into three scales: a general Pool wide observation; a smaller more in-depth observation on an area of island emergence and growth in the lower pool; and a subset of that section describing subaqueous conditions utilizing bathymetric data. The results from this study have indicated that site-specific geographic and hydrologic conditions have contributed to island emergence and growth in Pool 6. In Pool 6 land has been emerging at an average rate of 0.08km2/year since 1975.  Within lower Pool 6, land has been emerging on an average rate of 18m2/year since 1940. The bathymetric subset has shown that sediments on average have gained 2.41m in vertical elevation, which translates into just under 828,000 m3 of sediments being deposited in 113 years.  By identifying and describing these conditions river managers will be able to apply such knowledge to locate or reproduce similar characteristics within degraded sections of the UMRS. If the observations hold true in other locations, restoration efforts will be cheaper, more self-sustaining, promote natural fluvial dynamics, and ultimately be much more successful.

We are currently preparing a manuscript for publication.

Looking back at the Upper Mississippi River, moving forward

A student and I are working on finishing a project that has lingered for too many years: a careful analysis of the cumulative effects of river management on islands in the lower part of Pool 6 of the Mississippi River, near my hometown of Winona, Minnesota. There will be a MS thesis soon and hopefully a journal manuscript shortly to follow that, but for now, I’m enjoying discovering new and old research and resources on “the father of waters.”

First, check out this 17-minute silent film on the 1927 Mississippi River flood:

For more information on the film made by the Signal Corps in the 1930s, head here:

Then, check out this 2012 publication from the USGS on “A Brief History and Summary of the Effects of River Engineering and Dams on the Mississippi River System and Delta.”

Finally, there’s a paper just out in Geophysical Research Letters by Frans et al. titled “Are climatic or land cover changes the dominant cause of runoff trends in the Upper Mississippi River Basin?.”

And that’s my afternoon reading sorted.

March Meanderings

Cross-posted at Highly Allochthonous

It all began at the end of February, when I travelled to La Crosse, Wisconsin to the Upper Midwest Stream Restoration Symposium, which was a really stimulating and vital mix of academics, consultants, and government folks all interested in improving the state of the science and practice of stream restoration. I gave a talk on Evaluating the success of urban stream restoration in an ecosystem services context, which was my first time talking about some hot-off-the-presses UNCC graduate student research, and I learned a lot from the other speakers and poster presenters. While the conference was incredibly stimulating, travel delays due to bad weather on both ends of my trip made for a somewhat grumpy Anne (nobody really wants to spend their birthday stuck in a blizzard in O’Hare), so I’ll be thinking carefully about how to plan my travel to the Upper Midwest during future winters. Nonetheless, the view from the conference venue was phenomenal.

icy river and snowy land

View of the Mississippi River from the Upper Midwest Stream Restoration Symposium in La Crosse, WI. Not shown: bald eagles that frequent the open water patches of the river.

March proper saw me give variations of the restoration talk two other times. On the 15th, I gave it as the seminar for Kent State’s Biological Sciences department, and on the 26th, I gave it at the North Dakota State University Department of Geosciences (more about that trip below). In between, I gave a seminar on the co-evolution of hydrology and topography to the Geology Department at Denison University in Granville, Ohio. Students in that department had just returned from a trip to Hawaii, and a very memorable dialogue occured in the midst of me talking about the High Cascades:

“You’ve seen a young lava flow. What would happen if you poured a bottle of water on it?” “It would steam!” “Not that young!”

Closer to home I also hosted a couple of prospective graduate students, helped interview candidates for a faculty position in our department, and went with a colleague to visit an acid mine drainage site about an hour to the south of Kent. In one fairly small watershed, we were able to tour a number of different remediated and unremediated sites, and it certainly lent a whole different perspective to the ideas of stream restoration and constructed wetlands to look at a landscape irrevocably scarred by mining activities.

Orange water flowing from a tube down a hill and into a stream.

Unremediated acid mine drainage flow directly into Huff Run. The orange is iron precipitate.

Wetland plants and a concrete inlet weir.

Constructed wetland as the second stage of acid mine drainage remediation in the Huff Run watershed.

At the end of the month, we finally got our turn for spring break. I ended up with a somewhat epic combination of mounds of work and a big trip to take, possibly the worst combination of the untenured and tenured professor spring break stereotypes (see this PhD comics strip). The first half of the week, I spent in Fargo, North Dakota, home to the famously flood-prone Red River of the North. (I’ve blogged before about why the river so often produces expansive floods.) It was truly fascinating to put my feet on the ground in a place that I’ve read about and watched from afar for years. And my visit was made all the more interesting by my host and guide, Dr. Stephanie Day, a geomorphologist newly at NDSU and who may well unravel some of the Red’s geomorphological peculiarities.

Scientist in foreground, river in midground, background = flat, snowcovered ground.

Stephanie Day, Assistant Professor of Geosciences at North Dakota State University beside the Red River in Moorhead Minnesota. The flat surface in the background is the approximate elevation of the land for miles around.

Looking towards downtown Fargo, ND from the river side of the levee.

Looking towards downtown Fargo, ND from the river side of the levee.

snow and ice covered river, not in much of a valley.

River’s edge view looking towards downtown Fargo. Snow well over knee deep here on 25 March, by my measurements. As all that snow starts to melt, the water will rise.

There’s a pretty good chance we’ll see a major flood on the Red River later this spring, as the >24″ of snow melts out of the watershed, runs off over frozen ground, and enters the northward flowing river. The Fargo Flood page is the place to go to follow the action, and you can count on updates (and more pictures) here as events unfold.

The latter half of my spring break saw me diagonal across the state of Minnesota to my beloved Driftless Area, back across the Mississippi River, and into the state of Wisconsin. I saw my family, finished paper revisions, and wrote part of a grant proposal. Then I flew home, with nary a weather delay in sight.

If March was a tight, recursive meander of talks and trips to the Upper Midwest, then April promises to be a bit anastomosing with lots of different threads woven together to make another month of scientific delight.

The Great Flood of 1913

The 100th anniversary of Ohio’s greatest disaster is just days away. This epic hydro-meteorological event utterly ravaged river towns from Illinois to Ohio and beyond, but it seems like the event has largely been forgotten in history’s annals. Even flood-obsessed me had lived in Ohio for a few months before I even began to piece together the full extent of the disaster. For a crash course in the events of March 23rd-27th, 1913, navigate through this Prezi:

If you want to know more, there’s lots of details at the Silver Jackets’ 1913 Flood website and you can follow along as historian Trudy Bell researches a book on the flood.

After the storm

Cross-posted at Highly Allochthonous

It’s been quite a week. My home in northeastern Ohio got off lightly from “Superstorm” Sandy, compared to places closer to the Atlantic seaboard and in the Caribbean. But still, over 250,000 people lost power due to high wind, especially in Cuyahoga and Lorain counties along the shores of Lake Erie, where huge waves also caused closure of an interstate and damage. Power crews are still working to restore power to tens of out thousands, and most schools and universities were closed for at least one day, if not longer.

NewsChannel5 photo of large waves crashing against shore in foreground, smokestacks in background

Waves from Sandy crashing against the Lake Erie shoreline in Cleveland. Photo from News Channel 5. Click image for link to source.

Large tree fallen in front of house.

A tree down in my neighborhood, which took the branch of another one as it went. This same picture was the one featured on the local paper’s website story about storm damage. Does this mean it was the most dramatic tree to fall in Kent? Whether or not it was, these people got lucky the trees fell away from their house.

There was also some rain. At my house, I got 4.25 inches (108 mm), which is almost exactly what the forecasts predicted. It came as both a drizzle and as heavy rains, but since last Friday afternoon we haven’t seen the sun. Now, northeastern Ohio is supposed to be quite cloudy, but given the local grumbling, this might be a bit of an extraordinary gray and damp cold run. It wasn’t warm rain either, with temperatures neither climbing out of the 40s F (8 C) or dipping below freezing. Isotopic results are pending, but my money is on our moisture source being almost entirely that northern airmass that got itself entangled with the tropical cyclone. Again, any whining about the damp is pretty well offset by everyone acknowledging that we are extremely lucky compared to states to our east.

All that cold rain brought the local river levels way up. There was major flooding on the Cuyahoga River at the downstream end by Wednesday, and the river at its upstream-most gage in Hiram crested on Thursday night. Flow at Hiram peaked around 1900 cubic feet per second (53.8 cubic m/s), which as I eyeball it on the USGS annual peakflow graph appears to be about a 2-year flood. This is actually consistent with my eyeballed estimate of the flow frequency produced by Sandy on Passage Creek, near Callan Bentley’s house in Virginia. I wonder whether that will be consistent for other rivers affected by Sandy.

For me, this was the first chance to the Cuyahoga River in action as it flows through Kent. The river sits in a gorge than separates the two halves of town, and that seems to keep the river from endangering much property in the town. But it did make for a pretty impressive roaring site and sound as I crossed the bridges today. Here are two pictures of Heritage Park in Kent on Friday afternoon about 4 pm. Contrast that with the low water pictures from early June.

Cuyahoga River in Kent Ohio with impressive whitewater as it passes through an old lock.

Cuyahoga River in Kent Ohio with impressive whitewater as it passes through an old lock. Photo at 4:15 pm November 2nd, 2012 by A. Jefferson.

Flooding downstream of an old dam

Note the water level relative to the trees and those vicious rapids downstream of the lock. The dam in the foreground has been taken off-line and turned into a Heritage Park. Photo by A. Jefferson 4:15 pm 2 November 2012.

Lock at low water

The same lock structure as above, except at low water levels. June 2012, photo by A. Jefferson. Note complete absence of rapids downstream of the lock.

Similar view looking downstream past the dam as the picture above. Note how much vegetation is above water here.

Happy New Water Year! For hydrologists, it’s already 2013.

Cross-posted at Highly Allochthonous

Trees with autumn colors on the banks of a river

The Upper Cuyahoga River as it might look about the time of the new water year. Photo by Ohio DNR (click image for source). A 40 km section upstream of Kent is a state scenic river and I really want to canoe it.

There’s nothing particularly deterministic about starting a new year on January 1st. Our wall calendars happen to do so because of the circumstances of history. For hydrologists in the northern hemisphere, January 1st is not a great time to declare one year dead and a new one born. So instead, we transition between years on the 1st day of October. October 1st, 2012 marks day 1 of the 2013 water year. But, why?

Many hydrologic analyses involve calculating statistics on an annual basis. We might want to calculate annual streamflow to determine how much drier 2012 was than 2011. Or we might want to look for trends in the size of floods. For the example of flood statistics, the most common way to get a time series of floods is to identify the largest flood occurring in each year. This “annual peak flow” record is then used within the framework of a probability distribution function to assign a probability of given size flood occurring in any one year. (Problematically, this has been called a “flood frequency distribution”, leading to unnecessary confusion on the part of the public, but I digress…)

We use only the largest flow each year to calculate statistics because it avoids statistical complication. Principally, we want to ensure all of our events are independent of each other. Let’s say there are two days with really high flows in a year. The highest flow is 27.8 m3/s and the second flow is 24.1 m3/s. If the first measurement was made on January 29th and the second one was from March 18th, then these two data are independent (i.e., not the same flood). But if one is from January 29th and the other is January 30th, they are clearly not independent and should not both be used to calculate statistics as if they were. So, we try to avoid this dependence issue by using only the largest flow each year.

Here’s where we come back to the idea of the water year. Let’s use an example from my local stream gage – the Cuyahoga River at Hiram Rapids. There was a big flood on December 31st, 1990 – the biggest flood of the whole year – reaching a peak flow of 71.4 m3/s. On January 1st, 1991 the water was still high – 66.8 m3/s. After that excitement, the rest of 1991 was pretty dry and flow never again gets anywhere near as high, peaking at 25.0 m3/s in early March. How do we calculate the statistics? If we used the calendar year as our basis, we’d end up double counting that late December flood and we’d throw our flood statistics off.

Instead, hydrologists use October 1st as our cut-off date. In many parts of the Northern Hemisphere, summer is a period of low streamflow, driven by strong evapotranspiration and atmospheric circulation patterns. (A prominent exception to the lack of summer flooding is when tropical cyclones make landfall.) In winter, rain-on-snow can produce large floods in some regions, while decreased evapotranspiration and more frontal storms increase the chances of flooding in some southern regions. In the spring, seasonal snowmelt produces flooding in northern and alpine regions. That leaves the autumn as the period least likely to have frequent flooding. Also, because evapotranspiration is subsiding with cooler temperatures, soil moisture and stream flow don’t tend to be recovering from their low points in the summer. So autumn is a time of transition and a time when extremes are unlikely. The graphs below illustrate this for my local stream gage, but similarly shaped distributions would likely exist for many other gages in the US and beyond.

Thus, it’s a perfect time of year to the clear the books and declare a new year for hydrologic statistics. And it’s got as much or more physical rationale than when we change the calendar on the wall. Happy New Water Year Everyone!

Bar graph showing months with the greatest frequency of flooding in December-April

USGS data on annual peak flows for the Cuyahoga River at Hiram Rapids (gage #04202000) illustrates that summer and fall have the least frequent big floods.

March has the highest discharge, July and August have the lowest.

Mean monthly discharge for the Cuyahoga River at Hiram Rapids shows that summer is the period of lowest flow and that by September and October average discharge is starting to increase.

In slow-moving hurricanes, the danger comes from all the water

Cross-posted at Highly Allochthonous When Hurricane Isaac passed over New Orleans as a Category 1 storm on the seventh anniversary of the disastrous Hurricane Katrina, everyone in the US let out a big sigh of relief. A category 1 storm, the lowest level of hurricane intensity on the Saffir-Simpson scale, meant sustained winds in the 74-95 mile per hour (119-153 km/hr) range, which are described as “very dangerous winds [that] will produce some damage.” There were few, if any, mandatory evacuation orders in Louisiana, and the media interviewed people saying that they heard Isaac would be a Category 1 storm so they “didn’t think it would be that bad.” Those people opted to stay in their homes Louisiana’s Plaquemines Parish, along the Mississippi River near New Orleans. Indeed, as the early reports from Louisiana came out, it sounded as if the storm had been relatively low in drama.

Hurricane swirl as Isaac makes landfall in Lousiana contrasts with the bright city lights in the southeastern US

This will be the iconic image of Hurricane Isaac. NASA/NOAA/DoD VIIRS image of the hurricanes clouds superimposed on the city lights on the southeastern US. All those clouds are full of water. Image source:

Only later did reports start to trickle out of levees overtopped, and people stranded on rooftops and in attics, being rescued by neighbors with boats. The flooding this time wasn’t in New Orleans itself, but in nearby Plaquemines Parish, where levee upgrades weren’t scheduled to be completed for a few more years. At least one levee overtopped, flooding the town of Braithwaite and surrounding areas where about 1700 people live, with up to 4.3 m (14 ft) of water. That water ended up trapped between the federal, main Mississippi River levee and more locally managed back levees. State officials have now breached those back levees to more quickly drain the water out of the town, rather than slowly pump the area dry. But several people died inside their flooded homes.

Aerial view of flooding in Louisiana Parish

US Coast Guard photo of floodwaters in Plaquemines Parish, Louisiana

It’s not clear to me from the news reports whether the levee overtopped from a wind- and pressure-driven storm surge or whether it overtopped from the sheer amount of rain that fell on the area, but in either case the slow-moving nature of Hurricane Isaac turned out to make the meager Category 1 hurricane into something much more horrific for some Lousiana communities. A reporter on the scene in Braithwaite described the eyewall, with the most intense winds and rain, stalling out in the area, but throughout its life Isaac was a fairly slow moving tropical cyclone. As it moved across Louisiana, its center was moving north about 9 miles per hour (14.5 km/hr). Typical hurricanes move about 15-20 mph (24-32 km/hr), and some can move up to 60 mph (96.5 km/hr).

The problem with a slow-moving hurricane is that vast amount of precipitation can occur in the affected areas. In some parts of Louisiana, Alabama, Mississippi, and Florida more than 15 inches (380 mm) of rain have fallen in the last week. In New Orleans, the Hydrometeorological Prediction Center reports that 20.08 inches (510 mm). In the image below, you can also see the northward progression of the storm since making landfall.

Colorful image of rain in Louisiana, Florida, Arkansas and Missouri as a result of Isaac.

NOAA's Advanced Hydrologic Prediction Service (AHPS) map of rainfall accumulations for the week leading up to September 1, 2012.

All that water can lead to levee over-topping, like in Plaquemines Parish, and the risk of dam failures. Evacuations were ordered along the Tangipahoa River, which drains into Lake Pontchartrain, because of fears that Percy Quin Dam would fail. More than 50,000 people have been evacuated as the risk of dam failure or the need to intentionally breach the dam is still being evaluated. And, of course, while media attention (and this blog post, guilty as charged) focuses on the dramatic stories, there are many other areas in the Gulf Coast where flooding is on-going. Even as far north as Kansas City and southern Illinois, flood warnings are in effect.

Isaac is a good reminder why the primary cause of death in the US from tropical cyclones is from freshwater flooding. And it suggests that the single-minded focus on hurricane windspeeds may distract us from taking the flooding threat as seriously as we should. Those people who decided to stay in Plaquemines Parish because the Category 1 hurricane wouldn’t be that bad? When the interview was conducted, they were expressing their regret. The president-elect of the American Meteorological Society, J. Marshall Shepherd, wrote a blog post about the Lessons from Isaac, in which he suggested: “Is it time to consider an augmentation of the Saffir Simpson scale to capture the rainfall-flood threat? It is a difficult science problem, but probably one worth investigating. I also argue that our media colleagues must consider their coverage strategy and category “anticipation” or hype carefully.”

Flooding around the world (3 July edition)

Cross-posted at Highly Allochthonous

Here is a brief update on the floods I covered in the last edition of flooding around the world. Note that there has also been flooding in Xiengkoung, Viengtian, Boolikhamxay, and Xayaboury provinces of Laos, as a result of heavy rainfall from a tropical storm; in Russia’s Khabarovsk region (Kiya and Khor rivers), from heavy rainfall; and in the Philippines’ Davao city, from heavy rainfall.

China and the Yangtze River

The U.S. Corps of Engineers increased the output of the Gavins Point Dam spillway to 150, 000 cubic feet per second June 14, 2011. The flow was increased to help regulate the Missouri River due to record snow and rain fall earlier this year. (SDNG photo by Master Sgt. Donald Matthews)

Flow from the Gavins Point Dam spillway was 150, 000 cubic feet per second on June 14, 2011. (SDNG photo by Master Sgt. Donald Matthews, image on Flickr)

Missouri River

The Souris River, continues to flow over Minot, N.D. flood levees June 23, as the water begins to inundate residential neighborhoods. (DoD photo by Senior Master Sgt. David H. Lipp)

The Souris River, continues to flow over Minot, N.D. flood levees June 23, as the water begins to inundate residential neighborhoods. (DoD photo by Senior Master Sgt. David H. Lipp, image from Flickr)

Souris River