Currently browsing tag


Congratulations to Darren and Aly!

DarrenCongratulations to Darren Reilly who did a wonderful job defending his MS thesis on Tuesday. Darren’s thesis focused on the identification of groundwater pollution and its sources in rural northeastern Pennsylvania residential water wells. Darren will be preparing his thesis for publication in a journal and is looking for a job in the energy or environmental sectors. Check him out on LinkedIn.

Congratulations also to Alison Reynolds who won first place in the Kent State Undergraduate Research Symposium, Geology/Geography category for her poster on “Sensitivity of precipitation isotope meteoric water lines and seasonal signals to sampling frequency and location.” Aly is a junior this year, and will be continuing to be a valuable member of our research group this summer and next year before heading somewhere fabulous for graduate school.

Congrats Darren and Aly. It is a pleasure to work with such passionate and dedicated students.

A nice British video explaining the connection between rivers and groundwater. I can’t get the embed to work, so you’ll have to click through to watch: This is why I say I study rivers AND groundwater – if you want to understand how water moves through a watershed, you’ve got to …

New paper: Seasonal versus transient snow and the elevation dependence of climate sensitivity in maritime mountainous regions

Snowline near Skykomish, Washington (photo on Flickr by RoguePoet, used under Creative Commons)

Snowline near Skykomish, Washington (photo on Flickr by RoguePoet, used under Creative Commons)

Jefferson, A. 2011. Seasonal versus transient snow and the elevation dependence of climate sensitivity in maritime mountainous regions, Geophysical Research Letters, 38, L16402, doi:10.1029/2011GL048346.


In maritime mountainous regions, the phase of winter precipitation is elevation dependent, and in watersheds receiving both rain and snow, hydrologic impacts of climate change are less straightforward than in snowmelt-dominated systems. Here, 29 Pacific Northwest watersheds illustrate how distribution of seasonal snow, transient snow, and winter rain mediates sensitivity to 20th century warming. Watersheds with >50% of their area in the seasonal snow zone had significant (? ? 0.1) trends towards greater winter and lower summer discharge, while lower elevations had no consistent trends. In seasonal snow-dominated watersheds, runoff occurs 22–27 days earlier and minimum flows are 5–9% lower than in 1962, based on Sen’s slope over the period. Trends in peak streamflow depend on whether watershed area susceptible to rain-on-snow events is increasing or decreasing. Delineation of elevation-dependent snow zones identifies climate sensitivity of maritime mountainous watersheds and enables planning for water and ecosystem impacts of climate change.

Floodwaters rising on the Red River

Cross posted at Highly Allochthonous

Fargo, North Dakota is coming out of its 3rd snowiest winter since 1885. Snow continued to fall into late March, and daytime temperatures have only been above freezing for few weeks. At night, it’s still below freezing, though starting tomorrow night the forecast calls for above freezing minimum temperatures. Soils are already saturated, and more rain is possible this weekend.

In short, it is perfect flood weather for the Red River that runs along the Minnesota-North Dakota border and into Canada. This is a place with the perfect geography for extensive flooding, and a long history of big spring floods.

Checking the water level on a bridge between Fargo and Moorhead. Photo from Minnesota Public Radio.

Checking the water level on a bridge between Fargo and Moorhead. Photo from Minnesota Public Radio.

Every town along the Red River has been devastated by a flood more than once. So they’ve all got emergency response plans in place for weather just like this. For example, Moorhead (Minnesota, across the river from Fargo) has a nifty GIS feature that shows how each foot of flood water affects each city block.

Residents are already filling sand-bags to build temporary levees. But with year after year of flooding, and with successful sand bag efforts the last two years, some residents might be taking this year’s flood predictions in a somewhat complacent fashion. But looking at the National Weather Service’s North Central River Forecast Center projections, there’s plenty of reason for concern all along the Red River.

As of 9 am Central time on 7 April 2011, most of the US portion of the Red River is already above flood stage, but water levels will continue to rise almost everywhere for at least the next week.

Flood stages as of 9 am 7 April 2011. Screen grab from NCRFC.

Current flood levels along the Red River and nearby drainages, as of 9 am, Thursday 7 April 2011. Orange circles indicate minor flooding, red indicates moderate flooding, purple indicates major flooding. Screenshot from the North Central River Forecast Center, using data supplied by the USGS.

The flood wave will move downstream – from south to north. In Wahpeton, a crest is expected today, with a second – equally high if not higher – crest next week. There the flood crest is likely to fall a few feet short of record water levels set in 1997.

Between Wahpeton and Fargo, tributaries to the Red River are having major flooding as well – in part because of backwater effects from the main river. If the Red River is flooding, there’s no place for water flowing down the tributaries to go. Instead they back up, causing even more widespread flooding.

In Fargo (ND) and Moorhead (MN) – which have a combined population of 200,000 people – the flood will not crest until late Sunday. Right now, the National Weather Service is predicting a crest of 39.5 feet, which 1.3 feet short of the record flooding of 2009. However, there some chance that the river will crest at 41 feet, or even higher if there is precipitation in the next few days. Currently, 80% of the city is protected by sand bags and levees to a height of 41 feet, but those may need to go even higher.

NWS Flood Forecast for Fargo, North Dakota (7 April 2011)

NWS Flood Forecast for Fargo, North Dakota (7 April 2011)

Two weeks ago, the National Weather Service issued a longer-term flood forecast for the Red River at Fargo. At that time they considered it a 10-50% percent chance that the river would reach 40 to 44.3 feet by mid-April. They provided a probability of exceedence curve for their modeled projections of this year’s flood season against the historical record of flooding, as shown below. To understand this graph, it helps to look at a few specific points. Right now, the river is at 35.32 feet. Based on the outlook from two weeks ago, it was virtually inevitable that the river would reach this level, with a probability greater than 98%, as shown by the black triangles. In contrast, 35.32 feet is reached less than 5% of the years in the historical record for Fargo, as shown by the blue circles. The current projected crest of 39.5 feet was given about a 50% chance of being exceeded as of two weeks ago, yet it has only be reached twice (1997, 2009) in 111 years of record. Two weeks ago, the National Weather Service was saying that there was a 25% chance the river could go above 42 feet, which is higher than the top of the sand bag levees now being prepared.

NWS Chance of exceeding river levels on the Red River at Fargo, conditional simulation based on current conditions as of March 24, 2011

NC River Forecast Center's 90 model showing the Red River at Fargo's chances of exceeding certain water levels, relative to the historical record.

The short term forecasts, like the one two above, have better skill than long term forecasts like the immediately above, but the long term forecasts are vital for emergency managers, city officials, and riverside land-owners in making early plans for the flood. The reason they’ve got all the sand and sand bags on hand in places like Fargo is because they knew there was a good chance a really big flood was coming. They’ve been talking about it since January.

Downstream (north) of Fargo-Moorhead lies Grand Forks, with about 100,000 people in its metropolitan area. Grand Forks was swamped by the flood of 1997, but the current forecasted peak stage this year is about 3.5 feet lower, though the crest won’t reach Grand Forks until late next week. For now, they are watching the water levels and making their preparations. Downstream further, lies Winnipeg, Manitoba. The flood crest won’t reach there until late April, but already the river is 17 feet above normal winter stage, and only 5 feet below the 2009 flood peak. Needless to say, they too are sand-bagging.

But for the next few days, the action focuses on the Fargo-Moorhead area. You can check out the updated data and forecasts or you can watch the flood play out in Moorhead with a live webcam pointed at the downtown waterfront: live video from 702 Flood Cam – Moorhead on

A continental divide that runs through a valley

Now that’s pathological.

Parts of the Upper Midwest are disappearing under spring floods. The Red River of the North is at major flood stage, again, and the Minnesota River flood crest is moving downstream. It’s a pretty frequent occurrence in both of these river systems, and in part, flooding is a legacy of the glacial history of the area. The Red River flows to the north along the lake bed of Glacial Lake Agassiz, which is pathologically flat. The Minnesota River flows to the south along the channel of the Glacial River Warren, which was gouged out of the landscape by water draining from Lake Agassiz.

14,000 years ago there was direct connection between what is now the Red River basin and the Minnesota River basin. Today, there’s a continental divide – with the Red flowing toward Hudson Bay and the Minnesota flowing toward the Mississippi and Gulf of Mexico. But what a strange continental divide it is – for it runs through the former outlet of Lake Agassiz, in what is now known as Brown’s Valley or the Traverse Gap. This divide is not so much a high point in the landscape, but a just-not-quite-as-low area. The little community of Brown’s Valley sits between Lake Traverse (flows to the North, forming the headwaters of the Red) and Big Stone Lake (flows to the south, forming the headwaters of the Minnesota).

Here’s what it looks like on Google Earth. Note that I’ve set the terrain to 3x vertical exaggeration, so that you have some hope of seeing the subtle topography of this area.


And here’s a very, very cool oblique photo from Wikipedia. It shows the divide looking from north to south — mostly covered by floodwaters in 2007. It’s not every day you get to see a continental divide covered in water.


Why does the Red River of the North have so many floods?

Cross-posted at Highly Allochthonous

Communities along the Minnesota-North Dakota border are watching the water levels, listening to the weather forecasts, and preparing for another season of flooding. It must be a disconcertingly familiar routine, as this will be the third year in a row in which the Red River of the North reaches major flooding levels. But this isn’t merely a run of bad luck for residents in the Red River Valley, major floods are to be expected in a place with an unfortunate combination of extremely low relief and a river at the whim of snowmelt and ice jams.

The Red River of the North begins in Minnesota, near the border with North and South Dakota, and it flows northward through Fargo/Moorhead, Grand Forks, and Winnipeg before emptying into Lake Winnipeg, Manitoba. The landscape around the Red River is excruciatingly flat (Figure 1), for the Red River Valley isn’t a stream-formed feature at all, but is the remnant landscape of Glacial Lake Agassiz, which held meltwaters from the Laurentide Ice Sheet for more than 5000 years. The modern Red River has barely managed to incise into this flat, flat surface, because it slopes only very gently to the north (~17 cm/km). Instead, the river tightly meanders across the old lake bed, slowly carrying its water to the north. Topographically, this is a pretty bad setting for a flood, because floodwaters spread out over large areas and take a long time to drain away.

Topography of the US portion of the Red River Valley from SRTM data as displayed by NASA's Earth Observatoryredriver_srtm_palette

Figure 1. Topography of the US portion of the Red River Valley from SRTM data as displayed by NASA's Earth Observatory

The climate of the Red River watershed makes it prone to flooding during the spring, usually peaking in about mid-April. The area receives about 1 m of snow between October and May, and the river freezes over. In late March to early April, the temperatures generally rise above freezing, triggering the start of snowmelt. Temperatures warm soonest in the southern, upstream end of the watershed and they get above freezing the latest near the mouth of the river. This means that snowmelt drains into the river’s upper reaches while downstream the river is still frozen, impeding flow (Figure 2). As the ice goes out, jams can temporarily occur and dam or back up the river, exacerbating local flooding problems.

Red River near Oslo, Minnesota, 3 April 2009, photo by David Willis

Figure 2. Red River near Oslo, Minnesota, 3 April 2009. Here the main river channel is still clogged with ice, while surrounding farmland is underwater. Photo by David Willis of

Together the topography and climate of the Red River watershed are a recipe for large-scale flooding, and the historical record shows that floods are a frequent occurrence on the river. Usually, hydrologists talk about rivers in terms of their flow, or discharge, which is the volume of water per second that passes a point. But, when talking about floods like those on the Red River, it’s not so much volume that matters as how high the water rises (“stage”). The National Weather Service is responsible for flood prediction in the US, and they define flood stage as “the stage at which overflow of the natural streambanks begins to cause damage in the reach in which the elevation is measured.” If the water level continues to rise, “moderate flooding” occurs when “some inundation of structures and roads near streams. Some evacuations of people and/or transfer of property to higher elevations are necessary.” Further increases in water levels can bring a river to “major flooding“, when “extensive inundation of structures and roads. Significant evacuations of people and/or transfer of property to higher elevations.” That’s the sort of flooding that will happen in places along the Red River this spring, as it has many springs in the historical record (Figure 3).

Annual peak stage on the Red River at Grand Forks, North Dakota

Figure 3. Annual peak stage on the Red River at Grand Forks, North Dakota. Data replotted from the USGS, with local NWS flood stages shown.

Already, flood warnings are being issued for the Red River and its tributaries. As I’ll discuss in my next post, the long-range forecast for this spring’s floods on the Red is looking pretty grim. But as the communities along the river brace for the on-coming flood, it is important to remember that the geology and climate of the region make repeated major floods inevitable.

Flooding in Pakistan

A post by Anne Jefferson For the past two weeks, unusually heavy monsoon rains have deluged Pakistan, resulting in flooding and landslides. Pakistan is heavily populated all along the Indus River valley, so this is a slow-moving disaster of epic proportions. The latest news reports estimate that flooding has displaced 14 million people – more than the number of people affected by the 2010 Haiti earthquake, 2005 Kashmir earthquake, and 2004 Indian Ocean tsunami combined.

In the first 10 days of August, parts of Pakistan received almost 24 mm per day more rainfall than usual – during what is usually the wettest part of the year, when the monsoon rains fall. During June-September, the relatively cool Indian Ocean has high atmospheric pressure system, while the intense summer sun heats up the Indian subcontinent and forms a low pressure system. Much like a river flows downhill, air in the atmosphere flows from high pressure to low pressure areas. So moisture rich air from the ocean flows toward India and Pakistan – bringing months of cloudy weather and intense rain. The images below are from true color images from NASA’s MODIS satellite showing the Indus and Chenab rivers in the area around Islamabad. The top image shows the clearest day from recent weeks (9 August), while the bottom image shows a more typical day (6 August).

NASA MODIS image on Northern Pakistan from 9 August 2010

NASA MODIS image on Northern Pakistan from 9 August 2010

NASA MODIS image of northern Pakistan from 6 August 2010

NASA MODIS image of northern Pakistan from 6 August 2010

This year, the monsoon precipitation has been especially intense in Pakistan, because the jet stream is experiencing a blocking event – when the normal eastward progression of weather patterns in the midlatitudes gets stalled out and you get the same weather for weeks on end. This created an additional low pressure zone over Pakistan’s northern mountains, bringing even more moisture to the headwaters of the Indus River. The rain also seems to be exacerbating the landslide and landslide dam problems in the region of the Hunza River, a tributary to the Indus in the northern mountains. (Update: Jeff Masters has a nice explanation of this jet stream blocking event and how it links the Russian heat wave and Pakistan floods.)

Flooding on the Indus is not an instantaneous disaster – it is one that will continue to occur for weeks, with consequences that last years. Because the flooding is being caused by prolonged intense precipitation, there can be multiple flood peaks – where the water level crests, starts to fall, and then rises again. There are already two flood peaks moving downstream, and if further rain falls, there may be three peaks to the flooding, and it could last through the end of August. Also, since much of the rain has fallen in the north, closer to the headwaters of the Indus, the flooding began in the north, and the flood waves are transmitted downstream over a matter of days to weeks. While river levels are now slowly declining in the north, the first flood peak is just reaching Hyderabad – the largest city along the river, with a population of 1.5 million. How fast the flood moves downstream depends on the storage properties of the channel and floodplain. There is some possibility that if the first flood peak stalls out in an area with lots of floodplain storage or obstruction of flow by debris-choked bridges, the second flood peak could catch up, creating an even larger disaster. NASA has a series of wonderful images showing the flood progressing downstream from Sukkur, north of Hyderabad, from 8 -12 August. The latest image is shown below.

Flooding on lower Indus River, 12 August 2010 (NASA MODIS image, combination of infrared and visible light)

Flooding on lower Indus River, 12 August 2010 (NASA MODIS image, combination of infrared and visible light)

Even when the final flood peak reaches the ocean and the Indus River returns to its banks all along its course, the human disaster will continue to unfold. 14 million people have been displaced by this flood – and those people may have lost everything they own. Beyond the personal losses, there has been devastation to infrastructure such as roads and bridges, complicating relief efforts and even making access to some areas nearly impossible. The Indus is the source of water for irrigation canals throughout Pakistan and damage to them is likely to be intense, especially near the river, so agricultural productivity will suffer even in areas that escaped inundation. Village wells will have been contaminated by floodwaters, so access to safe drinking water will be an issue for months.

That’s a lot for any nation to handle – 1 in 12 residents directly affected by the flood – but for Pakistan’s already fragile national government it will be an especially difficult challenge. As the flooding has unfolded, Pakistan’s government has appeared less equipped to provide immediate relief to flood victims than Islamist charities, which will probably increase their support as they fill empty stomachs and provide shelter. There are other aid groups working to ameliorate the suffering. Two of my favorites are MercyCorps, providing clean water, food, and clean up tools in the Swat Valley, and Medecins Sans Frontiers/Doctors Without Borders, which is providing sanitation kits and basic supplies in Kyber Pakhtunkhwa and Baluchistan. Dave Petley, who has worked in Pakistan, recommends FOCUS Humanitarian Assistance.

Anne's picks of the literature: river and floodplain sediments

ResearchBlogging.orgIn July, four geomorphology papers particularly piqued my interest, and, as I started to summarize them, I realized they were loosely connected by a common theme. These four papers all attempt to understand what controls the sediments that make up the streambed and floodplain and that get preserved in the geologic record. White et al. look at how riffle positions are governed by valley width variations, while Jerolmack and Brzinski find striking similarities in grain size transitions observed in rivers and dune fields. Hart et al. examine the relationship between glacial advances and downstream sediment deposition, while Sambrook Smith et al. investigate the sedimentological record of floods.

White, J., Pasternack, G., & Moir, H. (2010). Valley width variation influences riffle–pool location and persistence on a rapidly incising gravel-bed river Geomorphology, 121 (3-4), 206-221 DOI: 10.1016/j.geomorph.2010.04.012

In gravel-bed rivers, channels commonly take the form of alternating pools and riffles. During low flows, pools have deep, slow flow, while riffles are shallow and fast. During floods, pools scour deeper, while riffles may get sediment deposited. This counter-intuitive behavior is explained by channel width variations during high flow – riffles tend to be wider than pools. As the water level rises, valley width may come into play. If the river is confined by valley walls, it will be deeper and faster and able to carry more sediment than where the river is unconfined by the valley walls. By what is known as flow convergence routing, deposition occurs where the river is least width-confined and has the lowest transport capacity. In this paper, White et al. examine the location and persistence of riffles in relation to oscillations in valley width for one reach of California’s Yuba River. Using repeat aerial photography, they show that many riffle crests are located in the widest portions of the valley, and that these riffle crests were persistent for decades. Despite being downstream of several dams, the study reach was geomorphically active – with frequent overbank flows, planform change, and rapid incision (0.16 m/yr), and yet riffles located in the widest parts of the valley remained stationary. Conversely, where riffles were created by large mid-channel gravel bars, and were not in sync with valley width oscillations, they tended to be destroyed by large floods. These results support the idea that flow convergence routing is an important control on pool-riffle channel form and stability and that the common assumption of uniform flow is invalid. Many river restoration practices are based on uniform flow assumptions, and the authors assert that without considering the implications of flow convergence routing, restoration practices are fundamentally misguided.

Jerolmack, D., & Brzinski, T. (2010). Equivalence of abrupt grain-size transitions in alluvial rivers and eolian sand seas: A hypothesis Geology, 38 (8), 719-722 DOI: 10.1130/G30922.1

Rivers generally exhibit downstream fining of sediments – in which the coarsest sediments are found near the headwaters and the finest sediments are found near the mouth. This fining trend occurs exponentially downstream – rapidly in steep rivers as boulders and cobbles give way to gravel-bed streams and much more slowly in low-relief settings where sand and silt can form the streambed for hundreds of kilometers. One interesting phenomenon is that the transition from gravel-beds to sand-beds occurs much more abruptly than might be expected, and many rivers have a deficit of sediment in the coarse sand/fine gravel size ranges (1-10 mm). Downstream fining in rivers has been attributed both to abrasion and to selective deposition of the coarse particles, but laboratory abrasion studies often show much lower rates of downstream fining than are observed in real rivers. Like rivers, wind-blown eolian sediments also exhibit an abrupt grain size transition – between sand and silty loess. Jerolmack and Brzinski (2010) examine the transport and abrasion dynamics of dunes and gravel-bed rivers to understand what mechanisms might be creating the abrupt grain size transition in both systems. Maximum geomorphic work (sediment transport times frequency of the event) in both gravel-bed streams and sand dune fields occurs when Shields’ shear stress is only less than two times greater than the stress required to mobilize the sediment. Similarly, the two systems are comparable in terms of abrasion collision dynamics, as estimated by the collision Stokes number. Abrasion produces smaller particles as big ones collide into each other and chip small pieces off. Abrasion efficiency decreases rapidly as grain size decreases, resulting in a minimum sediment size, which for rivers is in the range of ~10 mm gravel. Meanwhile, the small chipped-off pieces are sand-size (less than 2 mm) and continue to be transported downstream in suspension when the gravel settles out. As Jerolmack and Brzinski conclude “abrasion produces a bimodal grain-size distribution while sorting acts to segregate these grains to produce an abrupt transition.” While the authors acknowledge that the sediment transport in rivers and air are each subject to different constraints, they maintain that their abrasion/sorting hypothesis may explain the longitudinal sediment distribution in both environments. They also propose several additional areas where work is needed to test their hypothesis – including studies of gravel and sand source regions in multiple river systems.

Hart, S., Clague, J., & Smith, D. (2010). Dendrogeomorphic reconstruction of Little Ice Age paraglacial activity in the vicinity of the Homathko Icefield, British Columbia Coast Mountains, Canada Geomorphology, 121 (3-4), 197-205 DOI: 10.1016/j.geomorph.2010.04.011

Paraglacial geomorphology refers to landscape forms and processes that occur in areas adjacent to glaciers and the movement of large amounts of sediment from valley slopes to river systems that accompanies glacial advances and retreats. How rapidly this sediment is transferred from glacial areas to paraglacial areas is of interest to geomorphologists working in alpine and polar landscapes. This paper uses dendrochronology and geomorphic mapping to investigate paraglacial geomorphology and the time lags between glacier activity and downstream sediment deposition in the southern British Columbia coastal mountains. In 1997, a moraine dam overtopped and breached, draining a proglacial lake, and flooding the rivers downstream. The floodwaters eroded through 4 m of paraglacial valley-fill units with in-situ tree stumps and woody detritus. Tree rings from the stumps indicate that they died because of rapid burial by overlying sediment (i.e., from flood deposits). Multiple valley-fill deposits indicate and provide dates for six aggradation events between 718 and 1794, and correlation of these dates with independent regional glacial chronologies suggest that all of the aggradation events occurring during periods of glacier advance. This suggests that river valleys downstream of glacier limits were affected by synchronous redistribution of sand and silt from glacial forefields, moraines, and valley slopes when climates were cold and wet and glaciers were active, and that there was little lag between glacial erosion and advance and sediment delivery to downstream areas. This is paper is cool because it provides data that speak to the rapid delivery of sediment from glacial to paraglacial areas and because it uses trees in the paraglacial deposits themselves to give a much longer dendrochronology than can usually be obtained in glaciated areas.

Sambrook Smith, G., Best, J., Ashworth, P., Lane, S., Parker, N., Lunt, I., Thomas, R., & Simpson, C. (2010). Can we distinguish flood frequency and magnitude in the sedimentological record of rivers? Geology, 38 (7), 579-582 DOI: 10.1130/G30861.1

Over time, the flows that commit the most geomorphic work are those moderately high flows that occur moderately frequently – generally every 1-2 years. But large floods – like those that occur every 50 to 500 years on average – can dramatically reshape the form of the river and floodplain. These generalizations are based on observations of modern river systems, but how do they apply to the sedimentological record that will preserve the river’s legacy for future eons? Using detailed digital elevation models (DEMs) and ground penetrating radar (GPR) surveys of the South Saskatchewan River, Sambrook Smith et al. investigated the legacy of a flood with a 1 in 40 year recurrence interval (i.e., 0.31% probability). What they found is that while there were significant erosion and deposition across the braided river, the depth of sediment scoured or deposited was not substantially greater than that observed following much smaller floods. The larger high bars steered flow around them, even during the flood, forcing channel erosion, but only facilitating less than 0.5 m of deposition on bar surfaces. While erosion and deposition was spatially extensive in the large flood, the style and scale of the deposits was similar to small floods. Thus, there was no distinct legacy of this flood event that would differentiate it from smaller events in the geologic record. More generally, the authors conclude that in rivers which can widen during floods (i.e.,are not valley confined), there may be little preserved evidence in the sedimentological record to identify low-frequency high magnitude events from run-of-the-mill annual floods. The present is the key to the past, but the past that is preserved in the geologic record loses some of the sweet details of the dynamic events that act on the earth’s surface.