September
Lawn Lake and Fall River Valley: Flood Rhythms
By the end of the first week in September, summer is clearly over. In most years, the first hints of golden and orange appear among the vegetation. Individual branches or whole aspen trees start to turn yellow, as do the willows in the valley bottoms. The leaves of wild strawberries become scarlet and steel-blue berries cluster along the branches of the ground junipers. Where elk congregate, the bulls are stripping velvet from their antlers and regularly emitting the high-pitched squeals optimistically described as bugling. The days have become noticeably shorter and are likely to start with frost. On some mornings the rising sun reveals a dusting of new snow on the highest peaks. Normally, this is the time of year that nobody thinks about rivers in flood. The summer thunderstorms are largely over and snowmelt is a long way off.
I think mostly about floods as I start up the trail toward Lawn Lake. I have hiked this trail before to see the traces of a dam-burst flood that occurred in 1982 and the splendid scenery of this valley. Now I return to see the effects of the latest flood. The flood of September 2013 disrupted the normal seasonal rhythms of precipitation making its way into streams. The rhythm of floods in the national park can be a regular seasonal pulse that nourishes river ecosystems. Snowmelt floods flush silt and clay clogging the spawning gravels on riffles, erode the streambed and banks to make deeper pools, and carry down a new load of nutrients from the surrounding forest. Floods can also occur as an irregular beat that seems as dangerous as a heart arrhythmia but is just as important to rivers as the annual snowmelt floods. I sometimes think of rivers as the veins and arteries of a landscape, circulating water that sustains plants and animals by carrying water, sediment, and nutrients. The analogy is not perfect, but the atmosphere would be the heart, delivering the life-giving water that circulates down the rivers. I doubt, however, that many people who were in the park during the flood think of the rain that fell over a few days in September 2013 as life-giving.
I visit Lawn Lake and the Fall River valley a year after the big flood to take stock and to think about why some streams in Rocky Mountain National Park responded to the rain very differently than others. Much more erosion and deposition occurred along the Fall River than along other rivers in the park. This responsiveness to the flood reflects both the river’s glacial history and the more recent history of water engineering.
The Legacy of Infrastructure
Like many valleys in the park, the Fall River and its tributary the Roaring River step down in alternating steep and gentle sections. The upper Roaring River starts at Lawn Lake and flows gently down a broad trough in which a Pleistocene glacier left huge masses of sediment. Just before entering the Fall River valley, the Roaring River cuts steeply through the lateral moraine left by a glacier along the Fall River. Here the Roaring River lives up to its name, crashing down whitewater cascades lined by tall banks of boulders mixed with sand. Entering the broad, gentle main valley, the water then winds down a meandering channel over fist-sized cobbles.
This is a lovely morning to be out, clear and calm, although the clouds start to roll in by mid-morning, hiding the top of Longs Peak within an hour. I labor up the switchbacks on the lateral moraine and then proceed more easily up the rest of the Roaring River valley. Masses of boulders lie piled along the channel. The newly deposited rocks are nearly white, not yet stained darker gray by lichens and weathering. I pass the remnants of the dam that failed in 1982 and reach the lakeshore.
A glacial moraine dams the water in Lawn Lake. Early settlers in the region built on this dam, adding a 26-foot-high earthen mound across the natural dam in order to store more water for irrigation. An outlet pipe built into the dam was used to release water and regulate lake level. Built in 1903, the dam served its purpose for many years, despite the lake’s subsequent inclusion in Rocky Mountain National Park. Things unraveled in 1982.
At 5:30 a.m. on the morning of July 15, 1982, the Lawn Lake Dam abruptly failed. Lead caulking around the connection between the outlet pipe and the gate valve had deteriorated, allowing water to leak from the pipe into the earthen dam. At first the water probably seeped through slowly, a diffuse wetting front that would deserve sinister music if watched as a movie. As more water leaked from the metal pipe, the force of the flowing water began to dislodge sediment and create a path that then concentrated additional water flowing with more force. The process, known as piping, creates subsurface conduits analogous to human-made pipes. When the pipe grew wider and enough sediment was eroded, the Lawn Lake Dam collapsed. The failure released 674 acre-feet of water into the Roaring River. An acre-foot is a volume of water well described by its name: the amount of water that equates to a 1-foot-depth across an entire acre—325,851 gallons. (A football field covers 1.32 acres.)
I pause for a while at the lake before continuing up the valley past Crystal Lake to the saddle between Hagues Peak and Fairchild Mountain. Once I am past Lawn Lake, the region appears more remote and natural, without the long swath of pale boulders along the river that the 1982 flood created and the 2013 flood renewed. The 1982 dam burst created a peak flow of 18,000 cubic feet per second along a river in which the normal snowmelt peak flow is about 200 cubic feet per second. The enormous pulse of water released by the dam ripped up the forested river banks and swept along house-sized boulders left by the Pleistocene glacier. Where the Roaring River is slightly less steep, the flood left sediment. Where the river course steepens, the flood scoured from 5 to 50 feet into the streambed and widened the channel by tens of feet.
Although plenty of glacial sediment remains in the valley today, the floodwater of 1982 effectively carried much of the sediment nearly 5 miles downstream to the junction with the Fall River, where the flood spread across the wider, flatter Fall River valley. When water loses momentum, it drops sediment. As the 1982 flood peak declined, the water dropped 364,600 cubic yards of sediment, creating the Lawn Lake alluvial fan. This fan was up to 44 feet thick and spread across more than 42 acres, a pile of sediment waiting for the next flood.
Views from 2008 at the alluvial fan, looking upstream (left) and from the side of the Fall River valley, looking down onto the fan (right; the Fall River flows from upper right to left). Although some plants had colonized the edges of the 1982 flood deposits, the center remained a mass of unvegetated boulders before the September 2013 flood.
Although the 1982 flood lost momentum in the Fall River valley, the floodwaters kept going, filling the sinuous, cobble-bed Fall River with sand up to 6 feet deep. Two miles downstream from the alluvial fan, the flood overtopped Cascade Lake Dam, a 17-feet-high concrete gravity dam built in 1908 to supply a pipeline and hydropower plant. The flood didn’t just wash lightly over the top: 4 feet of water flowing over the dam crest caused this dam also to fail, releasing another huge pulse of water. A diagram of peak river flow with distance downstream shows the effect of this second dam failure. The floodwaters had started to spread out and slow down in the broad Fall River valley, but the failure of Cascade Lake Dam spiked the peak flow once more.
Peak flow again declined in the relatively broad, gentle valley over the next 6 miles between Cascade Lake Dam and Lake Estes. The openness of the valley upstream decreased some of the force with which the flood hit Estes Park, but it was still bad enough. Most people had plenty of warning and escaped the floodwaters, although three people along the flood route were killed. The town of Estes Park and other entities sustained $31 million in damages at a time when the population of the town was only 6,000 during the winter.
The 1982 flood was about fifteen times the size of the largest previously known flood. The flood was a catastrophe for people and their infrastructure, but once a dam bursts, there is no longer a flood risk and neither Lawn Lake nor Cascade Lake Dam was rebuilt after 1982. What did the flood do to the river?
Floods and Rivers
The surge of floodwater down the Fall River filled the channel with sand. A great deal more sand eroded from the Lawn Lake alluvial fan during the 1983 snowmelt season, creating sediment loads a thousand times greater than those present prior to the flood. Over the succeeding decade, all of this sand slowly moved downstream, until the river once more flowed clear through alternating pools and riffles over a cobble bed.
A dam-burst flood from a human-built dam can substantially alter a river ecosystem, but the river recovers. Rivers are resilient. They redistribute sediment mobilized during the flood. River organisms—bacteria, aquatic insects, fish, riverside plants—recolonize the disturbed zone, moving in from sites downstream or from adjacent, undisturbed rivers. The flood was a disaster for the city of Estes Park and for the people who died, but not for the river.
I think about floods and disasters as I survey the view from the saddle above the Roaring River valley. North-northwest lie the Comanche Peak Wilderness and the peaks of the Rawah Range, with the Zirkel Range just showing beyond. The tundra flowers are done blooming and the alpine zone is distinctly autumnal, with tints of golden, orange, and burgundy among the olive-green leaves. Pikas chirp anxiously from a lichen-covered mound of rock, and a marmot bulked up for winter trundles across a grassy patch. The long views from the saddle foster a sense of detachment from the busy activity below, where heavy machinery shifts masses of rocks and logs dislodged onto the Fall River Road during the September 2013 rainfall. I sit on a soft, grassy tussock admiring the views until the wind picks up and hurries the scattered thunderheads together, when I retreat back down the trail toward the cover provided by forest.
The region around Rocky Mountain National Park does not get much rainfall. On average, Estes Park gets about 14 inches of precipitation, nearly half of which falls during the summer. Communities at the base of the mountains receive only about 15 inches of precipitation. All of this changed spectacularly during the week between September 9 and 15, 2013. The base of the mountains received between 10 and 18 inches of rain. Different portions of Estes Park received 6 to 11 inches and much of the national park received 6 to 8 inches of rain. That was enough rain to cause widespread flooding in Estes Park and at lower elevations. The flooding within the national park, although interesting, mostly did not do much damage to infrastructure. The exceptions were the Roaring River/Fall River area and West Creek, at the northeastern end of the park. A wildfire covering more than 1,100 acres burned a portion of the West Creek drainage during the summer of 2010. The 1982 Lawn Lake flood left an abundance of unconsolidated sediment along the Roaring River. In each case, the earlier disturbance probably contributed to the outsized response of the stream to the September 2013 rains.
The September rains created anomalously widespread flooding in Colorado. From Estes Park east to the Colorado-Nebraska border, high flows inundated towns, ripped out roads, bridges, and pipelines, and spread everything from sand and gravel eroded from floodplains to toxic chemicals leached from oil and gas operations across a broad swath.
The news media described the flood as unprecedented, but very large floods commonly receive this designation. During the week of the flood, a persistent low-pressure zone over the southwestern United States brought moisture from the tropics north into Colorado. The configuration of the jet stream and the presence of a high-pressure zone southeast of Colorado blocked this moisture from moving beyond the Front Range. A similar atmospheric configuration in September 1938 also produced widespread, damaging floods, but the population of eastern Colorado was much lower at that time.
News coverage of the 2013 flood focused largely on the damage to people and infrastructure. As always seems to happen after a flood, people spoke of “cleaning up” the rivers. They wanted to get debris out of the channels and get the rivers back into the channels present before the flood. Each of these expressions—cleaning up rivers and getting the rivers back into their channels—makes me cringe. When the debris in question is pieces of houses, cars, pipelines, or bridges, I’m all for getting it out of the channels. When the debris refers to trees toppled from the banks or sediment carried from upstream, the issue is much less straightforward.
A rainfall or snowmelt flood benefits a river. The flood may leave what looks like a mess, but that mess creates diversity of habitats—newly eroded secondary channels or cutoff meanders in which standing water or very slow flow provides nursery habitat for young fish, new logjams from trees toppled into the channel that create the backwater pools sought by fish, new moist sandbanks in which freshly deposited seeds of cottonwood and river birch can germinate. Habitat diversity is the foundation for diversity of species and ages of individuals within a species and thus for river health. An unusually large flood, such as the September 2013 flood, creates a range of new habitats that subsequent, smaller snowmelt floods help to maintain.
A really big flood does not necessarily wipe out river plants and animals, either. Animals, in particular, can migrate up tributaries, seek out zones of slower flow within the main channel, or otherwise survive the high water, ready to recolonize the newly created habitat after the flood. At least, animals can do that if human infrastructure and river engineering do not isolate individual animal populations by creating barriers to their movements. One of the sad stories I heard about the 2013 flood was the fifty or more beavers found dead on lower North St. Vrain Creek, pinned against a metal grill designed to keep wood out of a large culvert.
On this September day, the rain never materializes and I am lucky to see several large bighorn sheep that look much more impressive in these surroundings than the radio-collared bighorns seen from the road in Big Thompson Canyon. I reach the last, steep descent into the Fall River valley and look down to where the 2013 flood destroyed portions of the road and interpretive trail on the alluvial fan.
In addition to cleaning up the rivers, the other phrase I heard frequently after the 2013 flood was the need to put the river back in its place. The phrase has so many undertones: from the darkest and most anthropomorphic interpretation of unruly rivers that refuse to acknowledge their subservience to us, to a simple desire to return the river corridor to exactly what it was before the flood so that infrastructure can be repaired and people can follow established patterns. The irony of the phrase is that the rivers that migrated across their historic floodplains were in their places during and after the 2013 flood: only our misunderstanding of river dynamics made us perceive them as out of place.
Each of the cities snuggled up against the foothills—Fort Collins, Loveland, Boulder, Lyons, and others—is built on the floodplain and alluvial fan created by a river over hundreds to thousands of years. A river flowing from a mountain canyon onto a flatter plain typically loses energy and deposits some of the sediment carried in the flow. Depositing sand and cobbles on a floodplain or a fan causes the ground surface to become slightly higher in that area. Sooner or later, a flood overtops the banks and the entire channel moves sideways to another area slightly lower in elevation. This process periodically recurs over long periods of time, but if a large flood has not occurred recently and at least some vegetation has grown across the former channel courses, the dynamic history of the floodplain or alluvial fan may not be obvious to someone on the ground. The old channels do stand out with even a cursory glance at an aerial photograph: commonly, the differences in soil across the floodplain or fan appear as differences in vegetation, even if the floodplain is now plowed over for crops. If the floodplain is covered by an urban area the old channels are much more difficult to discern, until a large flood reclaims the floodplain for the river.
The repeated sideways movements of a river across its floodplain create diverse habitats for aquatic and riparian plants and animals. Historically, rivers at the base of the mountains had meandering or braided main channels surrounded by secondary channels, floodplain wetlands, and patches of cottonwood forest. The river and the adjacent floodplain were closely linked by movements of water, sediment, and nutrients. Snowmelt or rainfall floods shifted the main channel from side to side of the valley bottom, inundated the floodplain, and deposited organic-rich sediment as the floodwaters gradually receded back to the main channel. Water soaking into the valley bottom resurfaced as springs and seeps well away from the main channel, and hyporheic exchange cleansed the water of dissolved nitrogen. The floodplains were undoubtedly messy with downed, dead wood, marshes, and beaver-dam ponds. I think of them as gloriously messy in exactly the same way that a tropical rainforest is messy with lianas, epiphytes, termite mounds, and downed wood. In each case, the messiness results from the continuing, irrepressible give-and-take among rocks and soil, weather, hillslopes and rivers, and plants and animals. Scientists and environmentalists have effectively increased understanding that it may be a jungle out there, but it’s also an incredibly diverse ecosystem. I dream of an equal awareness of rivers as not just unruly pipes that convey water downstream, but as rich ecosystems entitled to occupy the spaces they have created through time.
Build It and Suffer the Consequences
One of the lessons I took from the 2013 flood was that of “build it and then you have to protect it.” Floods are natural disturbances that maintain the health of river ecosystems, but the construction of campgrounds, pipelines, roads, and bridges along rivers means that these structures must be protected from floods or rebuilt after floods. This leads to a great deal of manipulation and alteration of the river environment, even in national parks. River engineering typically creates more uniform, homogenous channels and floodplains. This decreases habitat diversity and isolates individual segments of a river network by blocking access to adjacent floodplains or blocking animal migration with structures such as irrigation intakes or culverts. How many people think about the effect of a culvert when they drive across a river? I never did until I became involved in a project to design culverts that would maintain the ability of river animals to move up- and downstream. That seemingly innocuous corrugated metal pipe can concentrate flow sufficiently to prevent upstream movement by salamanders or can create a vertical drop at the downstream end that prevents upstream migration by small fish. Each poorly designed culvert chops off another length of stream from an array of animals that might otherwise occupy the stream. With enough road crossings, the stream can become a series of disconnected fragments incapable of supporting much aquatic life. One study of national forest lands in Oregon and Washington identified over 6,250 road-stream crossings on fish-bearing streams, which equates to about one crossing every 3.6 miles of stream. Fishery biologists considered 90 percent of the culvert crossings to be at least partial barriers to fish passage, so that culverts blocked access to about 15 percent of the potential fish habitat on national forest lands in the region. For native fish species already under stress from other factors such as introduced fish, excess sediment coming from logged hillslopes and unpaved roads, and warming water temperatures, loss of access to 15 percent of their historic habitat is a serious blow.
Just like land animals, fish and other aquatic wildlife need to move. Movement of aquatic animals upstream and downstream helps to maintain a balance between predators and prey and allows animals to move to new food sources and habitat. Animals dispersing to new areas maintain genetic diversity between populations and can supplement populations in which birth of new animals is not keeping pace with losses to predation. Room to move ensures that animals can come and go as habitat is created or lost by floods, drought, or other disturbances. On larger rivers, animals repeatedly move between the margins and the center of the channel to reach areas of shallow and deeper water, swift and slow velocity, or cooler and warmer water. In the small streams most likely to have culverts, cross-channel habitat diversity can be limited, making movements upstream and downstream especially important.
As I look down on the Fall River detouring across the valley around the alluvial fan, I think that the greatest favor any society can do for its rivers, and itself, is to stay out of their way. Floods are natural disasters only because people live and build within floodplains that are prone to periodic flooding. The underlying folly of this practice is not unique to Colorado. Graphs of flood damage through time in the United States and in the world, adjusted for inflation, shoot upward alarmingly despite the enormous amounts of money invested in flood protection in the form of levees, bank stabilization, and dams. The explanation is simple: no structure is completely flood-proof, but structures provide a false sense of security that encourages people to continue to live and build within floodplains.
The Big Thompson River downstream from Estes Park exemplifies the folly of thinking that rebuilding stronger and better within the flood zone will prevent future damages. During the night of July 31 to August 1, 1976, up to 12 inches of rain fell in the area between Estes Park and the town of Drake. The resulting flood killed 139 people and caused $35 million in damages. It was the flood of record, an unprecedented catastrophe, a 1-in-10,000-year event, a freak of nature. People got busy right after the flood, rebuilding the road and the many houses along the river. Persistence and determination in the face of adversity can be admirable traits in some contexts, but not in this one. Less than forty years later, a flood of similar magnitude happened again and people seemed to be thunderstruck and outraged that the road was largely destroyed and houses were smashed and carried away. George Santayana famously remarked that those who cannot remember the past are condemned to repeat it. The Lawn Lake alluvial fan and Fall River valley are a form of living history different than those present in other national park units such as Antietam Battlefield or the Hubbell Trading Post National Historic Site, but the historical lessons of past floods are no less relevant than those of battlefields and cultural monuments.
Part of the problem is our limited ability to estimate how frequently really large floods will occur. Scientists initially interpreted the 1976 Big Thompson flood as the largest flood to have occurred within the past 10,000 years based on sediments exposed at the small town of Waltonia. Waltonia sits on an alluvial fan formed where a tributary creek joins the Big Thompson River within the river’s canyon. Entering from river right, the fan forms a sideways bulge of sediment that constricts the Big Thompson River channel. The energetic waters of the 1976 flood eroded this constriction, exposing 10,000-year-old charcoal buried in the fan. Geologists reasoned that, given the age of the organic remains, this was the only flood to have reached sufficient depth to erode that portion of the fan. Continued research over the next few years, however, indicated that the Waltonia alluvial fan is analogous to a conveyor belt of sediment moving very slowly down into the canyon. A big flood can erode the lower part of the fan, but the continued downslope movement of sediment will gradually rebuild the fan. The charcoal exposed by the 1976 flood might have been deposited on the very upstream-most portion of the fan and taken 10,000 years to work its way downslope before being exposed by flood erosion.
Scientists continued to develop other lines of evidence—from flood scars on very old trees growing along rivers in Front Range canyons to the ages obtained from wood buried in gravel bars during floods and then left undisturbed as the channel shifted to another part of the floodplain—to understand how frequently the largest floods return to the Front Range rivers. Gradually, the average time between occurrences of a 1976-magnitude flood declined from 10,000 years to a few hundred years, to less than a hundred years.
Room for Rivers
Thinking about the September 2013 flood in the context of history raises an interesting question. How relevant is the occurrence of past floods if climate is changing? Precipitation varies through time. By any measure—annual rainfall, rainfall within a particular month, maximum rainfall within a 24-hour period, and over any time span—years, decades, centuries—rainfall is more changeable than consistent.
Widespread acceptance among hydrologists of the idea that indirect flood records such as flood scars on trees or the age of exposed sediments can be used to estimate the frequency of floods has occurred only during the past twenty years. Without such geologic and botanical evidence, the only way to estimate flood frequency is to use measurements of floods from stream gages. The oldest stream gages in the United States date to the last decade of the nineteenth century, creating a relatively short record from which to extrapolate the frequency of a large flood that may occur only once every 100 or 150 years. Most stream gage records cover time spans much shorter than a century, yet a flood that recurs on average once every hundred years is the basis for regulatory zoning throughout the United States. Faced with this dilemma, hydrologists made the convenient assumption that floods are random in time and space. This allowed them to use floods during any interval of time—say, 1950 to 1970—and assume that the relationship between flood magnitude and frequency can be extrapolated to longer time periods.
This assumption, known as stationarity, was never anything more than a convenient fallacy. The more we learn about variations in stream flow through time, the clearer it becomes that stream flow is nonstationary. Fluctuations in the jet stream produce decades of enhanced flash floods across the continental United States and intervening decades of few flash floods. Extrapolate from a flood-rich decade and you get one estimate of the so-called 100-year flood. Extrapolate from a flood-poor decade and you get a very different estimate. In 2008 a group of hydrologists published a paper with the wonderful title “Stationarity is dead: whither water management?” in the prestigious journal Science. That was the death knell of the convenient fallacy, and engineers and flood planners have been scrambling to adjust ever since. Beyond the natural fluctuations in stream flow associated with weather and climate, we have variations associated with changes in land use (urbanization typically increases the flood resulting from a particular amount of rainfall because of the increase in paved area, loss of floodplains, and efficient routing of water through storm sewers), as well as global warming.
This makes it difficult to definitively determine whether the rainfall of September 2013 is a phenomenon of warming climate or would have occurred without human-induced atmospheric increases in greenhouse gases. However, atmospheric scientists agree that a warmer atmosphere holds more moisture, so warming climate likely contributed to the large amounts of rain that fell during the September flood and, more importantly, may result in more frequent large floods in future.
Warming climate and associated changes in floods is one more uncertainty adding to the challenge of managing natural environments in Rocky Mountain National Park. We are too far along the path of increasing atmospheric carbon dioxide levels to stop climate warming, but we can give rivers the room to move and adjust to changing levels of flooding. The European Union has adopted a river restoration program named Room for Rivers. A national park is the preeminent place to leave room for rivers, and I smile at the thought of what a good chant the phrase makes: What do we want? Room for rivers. When do we want it? Now.