Stream Restoration after the 2013 Floods
by Mo Ewing

On September 16, 2015 the Denver Post carried an article entitled, “Big Thompson restoration to start soon” (Johnson 2015). It has been two years since the historic floods of 2013. Roads and infrastructure projects are almost complete and now work is beginning to “bring the river back”. The Big Thompson River Restoration Coalition is beginning to raise $69 to $225 million to fund restoration projects on the Big and Little Thompson Rivers. The initial project will create places for the river to overflow its banks, widen the river in spots to handle bigger flows, create pools and eddies, add rocks and trees to portions of the river for fish habitat, restore native plants and native habitats along the river.

Major floods along the Front Range have occurred quite regularly in the past, the latest major ones being the Plum Creek Flood of 1965 and the Big Thompson Flood of 1976. With the forecast of more extreme weather events due to global climate change, the question is, how should we manage our watersheds in the future? Should we conduct massive, costly restoration projects? Or, should we let the rivers alone to determine their natural course?

Even without 100 year flood events, rivers are dynamic systems, constantly changing due to different stream flows created by rain events and snow-melt. A good case could be made for leaving the rivers alone and letting natural events determine what plants grow there.

To understand the complexity of the problem, I found a really interesting article written by Jonathan Friedman (Friedman et al. 1996) which studied how the Plum Creek changed over a period of 35 years after its 1965 flood. Most of that creek was left alone after the flood to revegetate itself, and Friedman’s findings shed some light on the issues that county officials face when deciding how (or whether to) revegetate the riparian areas of their rivers.

In February 2014 I visited some of the creeks to see what happened. I climbed up the Enchanted Mesa Trail in Boulder which follows a tiny ephemeral creek that isn’t even big enough to have a name. Up until the flood, it would run through thick vegetation for a month or two in the spring and then it would dry up in the summer. But that small amount of water had created a thick band of vegetation along its banks. When I visited in February 2014, it looked like a bulldozer had run down the middle of the creek and scoured out all of the vegetation. So much water had run down hundreds of gullies like this one, and so much rubble, rocks, soil and plants were scoured out of the foothills, that when the mass of water hit the major creeks below (South Boulder, Goose, St. Vrain, 2-Mile Canyon, James and Gregory Creeks) it acted like a giant bulldozer. The waters pushed the rocks and rubble into the stream beds of the lower creeks, filling them, making the water overflow the banks, and then pushing the rubble up into the riparian areas on either side.

Even though quite a lot of work was done before I got there in February, there were plenty of examples of what had happened to the shape of the creeks and riparian vegetation along their

Enchanted Mesa Trail, by Mo Ewing

sides. Many of the creeks I saw, like James Creek, had simply changed course. Their old stream beds were filled by rubble and the creek jumped into new lower areas. This occurred in some places, like Boulder Creek, because of the mining that had been done in the past. When the creek had been mined, the creek bed was moved to gain better access to the gravel, and pits were dug in the old watercourse, and the creek was never returned its original location. The floods simply returned the creek to these old watercourses.

In other places huge new gravel bars were created along the stream beds and also into much higher ground into old floodplain areas (like neighborhoods) that hadn’t been flooded in decades. And most stream beds had been filled so that they were as much as 48″ shallower than they had been befor the

floods. That was one of the reasons that Boulder County staff were so worried about new flooding during the spring-melts in 2014 and 2015. Would the shallower creeks over-flood their banks? Would the streams change their course again?

The field research that Friedman did on Plum Creek in 1991 illustrates the difficulty of the revegetation process in a dynamic system. Everything depends on future weather patterns, heavy rains, thunderstorms, snow-pack, snow-melt, and temperature changes.

Plum Creek is different than most of the gravel-based creeks in the Boulder Flood because it is a sand-bed stream. It runs north 14 km along the edge of the Front Range into the South Platte River. Friedman studied the river along 7km from Sedalia to Louvers. Normally the highest river-flows occur during snow-melt in the spring with occasional flash floods occurring during the summer (Friedman et al. 1996). On June 16, 1965, 14.17 inches of rain (81% of the total annual average) fell in four hours, and the valley floor was inundated for 2½ hours (Matthai 1969).

Before the flood, Plum Creek was a relatively sinuous, single-channel stream with steep wooded banks. The flood filled many of the bends in the former channel with sediment and the creek was changed into a straighter, wider, shallower but steeper braided channel. About 50% of the trees were removed and the overall decrease in vegetation resulted in destabilized banks and even with relatively minor high flows, caused further widening of the channels (Ostercamp and Costa, 1987).

Boulder Creek, Photo by Mo Ewing

The river, since the flood of 1965, has created five different plant communities along its floodplain. These communities occur on five relatively flat surfaces separated by increases in elevation (somewhat like a set of stairs). The five surfaces, starting from the creek bed and moving out into the higher levels of

the floodplain consist of:

  • The channel bed at the time the creek was being studied in 1991
    • table bars formed during a period of low creek-bed levels from 1987-1990
    • Stable bars formed by channel narrowing during relatively high creek-bed levels from 1973-1986
      Terraces of coarse sand deposited by the 1965 flood
    • Terraces of fine sand formed before the 1965 flood

The plant community in the channel bed had low litter and vegetative cover because it was usually inundated in the spring. Only six species occurred in more than 10% of the plots: cottonwood species seedlings (which occurred in 63% of the plots), willow species seedlings (34%); two annual graminoids: Cyperus aristatus (Flatsedge) (30%) and Eragrostis pectinacea (Tufted Lovegrass) (23%); an annual herb: Polygonum persicaria (Lady’s Thumb) (20%); and a perennial herb: Veronica anagallis-aquatica (Water Speedwell) (52%). Of the six species, only the cottonwood, willow and sedge species were native.

The plant community on stable bars formed from 1987-1990 were only 13 cm higher than the channel bed and consisted of channel sediment covered by a few centimeters of silt or clay. It had low litter cover and the highest species richness of all the groups. Fifty four species occurred in greater than 10% of the plots; 26 species were exotic. Populus deltoides ssp. monilifera (Cottonwoods) occurred in 40% of the plots, and cottonwood seedlings in 69%. Salix lutea (Strap-leaf Willow) occurred in 42% of the plots, S. x rubens and S. alba var. vitellina (an exotic) in 20%, S. exidua in 35% and Salix seedlings in 48%. The most common native species were Juncus bufonius (Toad Rush) (58%) and J. dudleyi (58%), Lycopus americanus (Water Horehound) (49%), Agalinis tenuifolia (Foxglove) (48%), Eleocharis macrostachya (Spikerush) (42%) and Cyperus aristatus (Flatsedge) (40%).

The plant community on stable bars formed from 1973-1986 averaged 31cm higher than the 1987-1990 stable bars and had higher litter cover, lower vegetative cover, a higher proportion of perennials, and a lower number of species. Only 23 species occurred in more than 10% of the plots, 13 of which were native species. Populus deltoides occurred in 11% of the plots, but Salix exidua occurred in 73% of the plots. The most common native species were Carex emoryi and C. lanuginosa (Sedges) (64%), Aster hesperius (White Aster) (45%), Juncus balticus var montanus (Mountain Rush) (42%), Equisetum arvense (Horsetail) (31%) and Poa compressa (Canada Bluegrass) (22%)

The plant community on terraces deposited by the 1965 flood had low litter and vegetative cover and a low number of species. They were formed when the flood cut off the bends in the channel and filled them with coarse sand. There were no cottonwoods or willows present. The most common species were mostly native taprooted ore caespitose species, including Sporobolus cryptandrus (Sand Dropseed Grass) (79%), Chrysopsis villosa (Golden Aster) (54%), Artemisia campestris ssp. caudata (Sagewort) (43%), Ambrosia psilostachya (Ragweed) (39%), Onosmodium molle var. occidentale (Marbleseed) and Gilia pinnatifida (both 14%).

The plant community on terraces formed before the 1965 flood consisted primarily of medium sand and finer sediments, with lots of organic matter near the surface. Their elevation was similar to 1965 terraces, but they had more trees, litter and vegetation and more exotic and rhizomatous species. 12 species occurred in more than 10% of the plots and eight of these were exotic. Populus deltoides occurred in 2% of the plots and Salix exidua occurred in 4%. The most common native plants were Symphoricarpos occidentalis (Western Snowberry) (21%) and Prunus virginiana (Chokecherry) (14%).

Some of the variation in plant communities are due to channel narrowing plant succession. The edges of the stream bed, are scoured by spring floods, followed by the establishment of willow and cottonwood seedlings with a few fast-growing herbs. If this area is not scoured immediately again new species become established including tap rooted, caespitose and rhizomatous perennials. The shoots and exposed roots of these plants trap sediments more efficiently, raising the elevation of the surface and decreasing the amount of disturbance. Litter and shade increase and water decreases as plants compete more intensively for light and water. This favors rhizomatous perennials. This is the point in succession where the greatest species richness occurs.

However some of the differences in plant communities cannot be explained by creek-bed plant succession. Both Populus and Salix are the first plants to colonize areas scoured by spring or summer floods and indeed they show a high percentage of occurrences on the surfaces that developed from 1987-1990 when the creek bed levels were low. A high discharge of snow-melt


Fourmile Creek Trailhead, Photo by Mo Ewing

in 1973 created the 1973-1986 stable bars but photographs taken in 1976 showed large areas of the bars were still unvegetated and establishment of woody vegetation did not start until 1979 when the creek bed levels were high. Because the bars were high in relation to the stream bed, the vegetation succession differed from the 1987-1990 surfaces because there was less moisture available during the initial colonization. On the highest of the 1973-1986 surfaces

cottonwoods are common, but Salix is scarce, probably because willows are more sensitive to lack of moisture.

From their creation, the gravel terraces deposited by the 1965 flood were high above the stream bed and although they were scoured, no cottonwoods or willows were able to colonize the terraces and none of those species occur there now. Plant communities which grow in riparian areas will vary depending on their elevation from the stream bed and and the resulting availability of moisture.

But elevation does not tell the whole story either. The pre-1965 and 1965 flood terraces are the same elevation (2.45 to 2.40 respectively), however, their plant communities are very different. It is clear that other environmental conditions must be influencing this difference. The mean litter cover in pre-1965 terraces is 69%, and 37% in 1965 terraces. Mean vegetative cover is 57% in pre-1965 and 33% in the 1965 terraces. Occurrence of sand and gravel is 74% in pre-1965, 93% in 1965. Mean percentage of native species is 22% in pre-1965 and 71% in 1965.

In a presentation made last spring at a Boulder monthly meeting Susan Sharrod and Laura Backus described the first vegetation to appear after the floods in the summer of 2014. They described the new vegetation as a mish-mash of native and exotic plants. Where the water table was high, native cottonwoods and willows grew in thick lawns on sandy banks but not on the cobble banks. Probably because the flood happened in September after plant seeds were released, the seeds of green ash, Russian olive, tamarisk, Siberian elm and locust, were probably washed away, and there was no regeneration of these plants. Some “weird” plants showed up, Nuttallia nuda (Blazingstar), Elatine triandra (critically imperiled in Colorado), Hibiscus trionum (an exotic) and Malus (apple) species. The noxious weeds that one would have expected such as Scotch thistle, purple loosestrife and teasel hadn’t shown up yet, but there were lots of thriving non-natives: alfalfa, mullen, kocha, reed canary grass, bindweed and smooth brome. Native species seemed to be doing better in the now damp side streams.

With such an uneven start it is understandable that revegetation plans are in full swing. But depending on river flows and weather events, the rivers will probably do what Plum Creek did, create completely different plant communities on different terraces along their banks, perhaps with a totally different suite of plants from the surrounding area. It will be interesting to see what happens over the next 35 years.


Pamela Johnson, “Big Thompson Restoration to Start Soon”, Denver Post, September 16, 2015

Friedman, Jonathan M., Osterkamp, W.R., Lewis, W.M., “Channel Narrowing and Vegetation Development Following a Great Plains Flood” Ecology, 77(7), 1996, pp 2167-2181

Matthai, H. F. 1969, Floods of June 1965 in South Platte River Basin, Colorado. United States Geological Survey Water-supply Paper 1850-B.

Ostercamp, W. R. and J. E. Costa. 1987 Changes accompanying an extraordinary flood on a sand bed stream. Pages 201-224 in L. Mayer and D. Nash, editors. Catastrophic Flooding. Allen and Unwin, Boston Massachusetts, USA.