Stormwater | Ancient Erosion and Sediment Management

Accumulation in reservoir yields clues to runoff patterns and soil gradation.

By Kenneth R. Wright, Eric Bikis, and Ernest Pemberton

Our engineers and scientists cut through a treasure house of ancient sediment layers to uncover clues to prehistoric Indian life, sediment dredging, forest fires, and even long-ago farming practices. The work was part of a 10-year-long research project by Wright Water Engineers Inc. (WWE) to learn about water handling and water use by people who lived in southwestern Colorado more than 1,000 years ago.

It was by studying a 350-year-long sediment accumulation in Morefield Reservoir, layer by layer, that we uncovered the secrets of this prehistoric period. Runoff patterns, ancient vegetation, farming activity, erosion, soil gradation, and other engineering parameters documented this long-ago period.

We learned much about the Ancestral Puebloans of Mesa Verde; we learned that these people lived with, and managed, erosion and sedimentation problems just like modern people—in fact, perhaps better than we do now.

To keep our engineering findings on track within the known archaeological framework of Mesa Verde, we enlisted Dr. Jack Smith for our research team. He formerly served as chief archaeologist for Mesa Verde National Park (MVNP) and has a wealth of knowledge of Mesa Verde and its former inhabitants. Thirty years earlier, Dr. Smith excavated our site while with the University of Colorado.

Reservoir Excavation
The stage was set for our sediment layer research when two MVNP archaeologists, Cynthia and Jack Brisbin, invited us into the park to study an enigmatic mound in Morefield Canyon. WWE was then given a permit to explore the site.

Figure 1. Looking north up Morefield Canyon. Morefield Reservoir is in the lower left. Note the stream channel, access road, and the broad valley and canyon walls.

The mound was unusual. It rose up 16 feet from the canyon floor, and we measured a diameter of 220 feet at its circular base and 130 feet across the top (Figure 1). We noted a long berm from the north that was connected to the mound.

Some scientists thought the structure was an ancient dance platform and that the berm was a walkway to its top. Another scientist said it was an erosional remnant of a Pleistocene terrace. Dr. Smith opined that it seemed to be a reservoir, but that there just was no water source."

Figure 2. Morefield Reservoir has the shape of an inverted frying pan.

It Was a Reservoir
Our excavation of the Morefield mound proved, without doubt, that it was a domestic water storage reservoir dating from AD 750 to 1100. The archaeological site had the appearance of a huge, inverted quarter-mile-long frying pan, the handle being the water supply inlet canal (Figure 2).

Why was a reservoir sitting above the canyon bottom rather than below? Why was the canal also above ground on a berm? The reason for the elevated structures was simple; there were 350 years of sediment inflow and deposition with which the Ancestral Puebloans had to contend.

In about AD 750, the Pueblo I people of Mesa Verde excavated a 4-foot-deep pond in the broad canyon bottom channel to intercept the groundwater table. It did not take long for sediment to accumulate. They dredged out the pond and cast the material to the edges. Gradually, the pond bottom rose because they never cleaned out all the accumulated material.

Within several decades, the storage vessel was above ground, and it was necessary to rely on the confining berms to contain the water. Year by year, the reservoir pond rose in elevation.

Figure 3. Mesa Verde National Park's 80 square miles contain prehistoric treasures.

Site Location and Characteristics
Morefield Reservoir (Site 5MV1931) of MVNP (Figure 3) is in Morefield Canyon in the southwest quarter of Section 33, Township 35 North, Range 14 West, of the New Mexico Meridian. It is approximately 12.8 miles southeast of Cortez, CO, in Montezuma County. Morefield Canyon is closed to the public to preserve its extensive, abundant, and valuable ruins for future study. The canyon held a large prehistoric population. The watershed consists of 4.2 square miles. The drainage basin ranges from 7,200 to 8,300 feet in elevation, with a mean elevation of 7,800 feet. The current land use is characterized as ungrazed rangeland.

Morefield Canyon is a broad valley with the side ridges rising 400 to 600 feet on each side of the nearly flat bottom that ranges from 600 to 1,100 feet wide. At the 5MV1931 archaeological site, the valley bottom is approximately 800 feet wide. The Morefield channel gradient slopes to the south at approximately 135 feet per mile, for a slope of 0.25 foot per foot. Approximately 10,000 feet north of the site, a decrease in gradient coincides with the change in bedrock from Mancos shale to the Point Lookout sandstone.

The Mesa Verde plateau is formed by the sandstones of the Cretaceous Age Mesa Verde Formation that are more erosion resistant than the underlying Mancos. The top of the plateau is essentially a dip slope lying on the Cliff House sandstone of the Mesa Verde Group. Morefield Canyon has cut through the Cliff House sandstone and is floored in the sands and shales of the Menefee Formation. No bedrock outcrops were observed along the valley bottom. No data have been found to establish the thickness of the deposits in Morefield Canyon, but the alluvium is estimated to be about 30 feet thick.

Evidence in the Sediment
Jack Smith, working for the University of Colorado, first excavated at the Morefield site in 1967. In 1997, our team was allowed to bring a backhoe to the Morefield Canyon site to excavate a trench with 1,900 feet of surface area through the reservoir berm.

Using oilfield geologic methods, we found that there were six berm-building phases. The huge trench face of the excavation showed that the ancient people did a lot of dredging. There were many truncated sediment layers; there were dumping areas for the dredged material, and we could identify old reservoir berms.

Figure 4. This sand layer from the Morefield Reservoir deposits shows ripples caused by the wind in the 9th century.

Sediment layers in the reservoir mound were clearly evident; some layers were sandy and some were of dense clay (Figure 4). The sandy zones told of high-flow periods, and the clay represented the deposition during low-flow periods.

The sand layers contained 14 different continuous thin layers of charcoal that represented ash washoff from forest fires in the 4-square-mile watershed. We know that forest fires cause more frequent and larger runoff peaks, which is one reason that the ash was in the sandy layers and not in the clay.

The tightness of the clay would have precluded any measurable reservoir storage seepage losses. The clay deposits would have been deposited in the reservoir during small rates of water diversion from the thalweg of Morefield Canyon and from some windblown sediment.

Figure 5. Prehistoric berm failure showing classical slip plane with clockwise rotation. The mid-truncated layer represents the berm shape after the failure.

Berm Building
Near the middle of the trench, the lower part of the fill contains a disconformity with several feet of variation in elevation. Slump blocks overlying this disconformity probably moved toward a void created by removal of sediment from the lower middle area. Because the lowest part of the detachment surface is tangential to lower, undisturbed couplets, the void was apparently created by manual excavation. This disconformity represents an ancient berm failure from over-excavation during dredging ( Figure 5) .

Figure 6. Click here for larger vierw.

Six reservoir-forming berm phases were exposed in the 1997 reservoir trench. The overall distribution of berm and fill sediments suggests that the width of the water surface varied with time as the reservoir rose progressively upward. At both ends of the trench, weakly bedded clayey material overlaps the berm sediments and is overlain by well-bedded fill sediments. Some parts of this clayey material contain thin or discontinuous beds of silt or fine sand. This unit may consist of peripheral fill sediments in which bedding was mostly destroyed during maintenance work on the berm. Overlapping, well-bedded fill sediments are consistent with this interpretation (Figure 6).

Figure 7. The trench provided a look at sediment deposition high inflows.

Flood Periods
The open trench wall that exposed the long period of sediment deposition provided us with a continuous record of reliable data (Figure 7). For instance, by counting the sandy layers, we could determine that there were 21 periods of high inflow spread out over the 350-year record period. On the average, there was heavy runoff once every 17 years. Some sand deposition zones were so thick that we concluded that the inflow had to be too much for the reservoir capacity. As a result, the berms would have been overtopped; they likely failed.

Prehistoric Vegetation
Where did the water come from? This question nagged us, because our present-day geomorphic studies showed that Morefield Canyon had not carried surface stream flows for at least several decades. Morefield Reservoir would have needed a fairly reliable storm runoff pattern.

Hydrologists know that tilled agricultural land has greater runoff frequencies and volume than natural vegetative ground cover. We found significant evidence of ancient agriculture upstream of the reservoir. The evidence was gathered from the sediment layers.

Figure 8. Soil samples were used for gradation analysis and pollen testing.

A vertical profile of soil samples was collected from the exposed trench wall, from top to bottom. This permitted us to define the layers, one by one, as to gradation. Sand, silt, and clay content were measured for each layer by our team member from the Natural Resources Conservation Service in Cortez, CO (Figure 8). The data were entered into the National Soil Inventory.

Soil samples were also used for palynology studies where the soil sample is carefully disintegrated to recover pollen grains. Soil can be a rich source of pollen. In this case, we recovered pollen that was carried into the reservoir, from upstream, via the inlet canal. Each of the soil samples from the profile gave up significant maize pollen; we knew that over 1,000 years ago, the Pueblo I and II people of Mesa Verde tended agricultural crops upstream of the reservoir. Besides maize, the soil sampling showed us the degree of pine and juniper foresting, the presence of sagebrush, Mormon tea, medicinal plants, and other vegetation. We could tell that the forest density varied from period to period.

The pollen studies also made us realize how well soil can preserve pollen grains and how valuable it is to study erosion-caused soil deposits to learn about long-ago environmental field conditions.

Prehistoric Precipitation
WWE studied dendroclimatic reconstruction for the Mesa Verde area to estimate average annual precipitation during the AD 750 to 1100 period of operation of Morefield Reservoir and to compare the prehistoric precipitation with modern records. Dr. Jeffrey Dean of the University of Arizona Laboratory of Tree-Ring Research in Tucson, AZ, provided the dendroclimatic reconstruction data for the Mesa Verde area from the AD 481 through 1988 period that was evaluated. The long-term average annual precipitation for the period of record was estimated to be 18.1 inches. This agrees with the 74 years of modern record from the weather station at nearby Chapin Mesa. The average annual precipitation from AD 750 through 1100 was 18 inches. For this reason, precipitation records from modern times were considered suitable for use in analyses of the ancient period.

Prehistoric Runoff Characteristics
Archaeological evidence indicates that up to 500 people lived in Morefield Canyon. Farming was practiced, and doubtless there was removal of trees and brush for cooking, construction, and heat. Rough estimates suggest it is likely that at least 125 acres were farmed (1 acre for every four people). In addition, people may have denuded a portion of the watershed by gathering firewood. Forest fires occurred in the canyon, as evidenced by layers of charcoal found in the reservoir excavation.

A study of land use/type suggests that a rainfall of 0.5 inch would produce a runoff of 0.033 inch from the 125 acres of farmland. The volume of runoff from this would-be 0.34 acre-foot provides adequate water to fill the reservoir at the site. The estimated runoff does not consider the area that would have been denuded for fuel or by forest fires. If these denuded areas were upstream of Site 5MV1931, then additional runoff would have been generated that could have been stored. For instance, it was estimated that the 1996 Chapin #5 Fire in MVNP temporarily changed runoff characteristics so as to increase the peak runoff potential some 500% to 600% because of loss of forest floor cover and hydrophobic soil conditions resulting from the fire. The Buffalo Creek (tributary of the South Platte River in Colorado) Fire of 1996 resulted in twelve 100-year floods in the first month following the fire. Forest fires cause a significant increase in runoff and sediment yield characteristics.

The Chapin Mesa rainfall records were evaluated to determine the frequency of rainfall events of 0.5 inch or greater. In the 48 years of record analyzed, there were approximately 200 days of recorded precipitation greater than 0.5 inch per day during the summer months. Under the prehistoric hydrologic conditions (farming, fires, and human activity), such events would have produced runoff. This suggests that direct runoff water would have occurred and would have been available for capture and storage about four to five times each summer. The water diversion canal route, defined in the field using instrument surveys, would have been able to intercept the canyon flow and deliver the water to the reservoir for storage.

Paleoflood Hydrology

WWE team member Dr. Robert Jarrett of the US Geological Survey made paleoflood estimates at 12 sites along Morefield Canyon in 1997 from Morefield campground near the head of the canyon to Site 5MV1931. Maximum paleoflood discharges range from about 250 to 250 cubic feet per second near Site 5MV1931. The paleoflood evidence is at least 100 years old and likely reflects the largest flooding in several hundred years. One of the largest flash-flood-producing rainstorms during the past century in southwestern Colorado occurred at MVNP. On August 3, 1924, 3.5 inches of rain fell in one to two hours at the US Weather Bureau gage located at MVNP. The estimated paleoflood discharge for nearby Spruce Canyon is about 1,000 cubic feet per second for a drainage area of about 2 square miles.

Figure 9. The ancient people lined the inlet canal right of way with stones to prevent erosion.

Intake Canal
The Morefield Reservoir evolved into an off-stream reservoir during its early life after the original excavated pond filled with sediment and was dredged out to form perimeter berms. An off-stream reservoir requires an intake canal for water delivery. The existing route of such an intake canal represents the final canal alignment at the time the reservoir was abandoned. Field instrument route surveys were conducted for mapping purposes. However, an inspection of the area in the vicinity of the final canal heading and the drainage channel revealed only modest evidence of a diversion structure (Figure 9).

In situ aligned stones defined the canal route. The canal was 1,425 feet long with an average slope of 1.0%—the slope ranging between 0.5% and 2.0%, not unlike modern farm irrigation systems in southwestern Colorado. Almost all of the observed stones along the right of way of the canal were likely used for bank erosion control and were rectangular in shape. Most were from the Menefee Sandstone. Some of the sideslope stones served to protect the canal bank from stream erosion. Excavations of the canal cross section were earlier conducted by Smith of the University of Colorado. These excavations indicated a bottom width of 3 feet, 2:1 sideslopes, and a maximum depth of about 1 foot. With these data, and by use of Manning's Equation, the computed bank full canal capacity under subcritical flow conditions was 19 cubic feet per second at a velocity of 3.8 feet per second. The roughness coefficient was estimated at 0.03, with the canal having an average slope of 0.01 foot per foot. The canal and the reservoir were integrated into a single operating structure for diversion, transport, and storage of water.

The inlet canal kept getting longer and higher as the reservoir grew in elevation. Smith's 1967 canal excavations showed numerous lower elevation canals that served the reservoir during earlier times. The final period of reservoir operation required a 0.25-mile-long canal that was able to carry 19 cubic feet per second. It was capable of transporting an instantaneous rate of sediment of about 700 tons per day.

Sediment Inflow and Deposition
The 1997 excavation of the Morefield Reservoir mound penetrated to the original ground surface on the east and west portions of the trench. In the trench midsection, the original ground surface was reached by auger at about 5 feet below the trench bottom. The auger at this deeper elevation penetrated the original pond bottom dug by the early Americans for groundwater and surface-water collection prior to the building of the reservoir.

Profile of 1997 Trench Cut
Figure 6 identifies the long horizontal layers of alluvial sediments exposed throughout the trench walls, along with gently upward-sloping layers at the edges where the reservoir embankment existed (Figure 5). At one location, evidence of embankment slope failure was evident. The distinct layers, while initially considered to be fine sand, were found to be mostly sandy clay or sandy silt. Approximately 14 thin continuous layers of charcoal deposits were found, representing fluvial-transported charcoal from forest fires.

Sediment Volume Computations
The sediment gradation and estimated transport velocities associated with peak runoff from estimated ancient thunderstorm events are compatible with an inlet canal capacity of about 19 cubic feet per second. Two selected sediment transport formulas showed for the 19-cubic-foot-per-second canal discharge an average concentration of 14,000 milligrams per liter and an instantaneous transport rate of about 700 tons per day. We concluded that the sandy layers represent erosion at the higher elevations of the drainage area. The heavy clay material represents materials eroded from the channel banks in the upstream valley alluvium representing previously deposited materials from the Mancos shale. The clay was judged to have been deposited during the dominant low-flow conditions. Of the total sediment deposited over the life of the reservoir, it was estimated that, on the average, about 1,230 cubic feet were diverted and deposited per year. About 270,000 cubic feet of the mound are clay, while 160,000 cubic feet are sandy silt representing high-flow periods. The reservoir rose in elevation because of sediment accumulation of about 1.6 inches per year. However, due to dredging efforts, the overall net rise averaged only 0.7 inch per year until in AD 1100 the reservoir fill was about 21 feet above the original pond bottom.

Figure 10. Broken pottery from jars indicates water was collected at the reservoir.

Pottery and Artifacts
The sediment deposits of Morefield Reservoir held lots of cultural evidence; there were pieces of pottery that we called potsherds (Figure 10), along with flat digging stones. We had the advantage of both our 1997 potsherds and the pottery from Smith's 1967 excavation. Trained ceramic experts can identify the period of the pottery, as well as the type of pottery. What was uncovered fell into the Pueblo I period (AD 750 to 900) and Pueblo II period (AD 900 to 1100). The oldest pottery was deepest, and the youngest was near the top of the open trench wall. On the surface, there were a few Pueblo III period potsherds, which came from later activity.

Of all the potsherds collected, 91% were from jars that would have been used for collecting and carrying water back to the nearby pueblos. Ladle potsherds accounted for 1% and the remainder of the ceramics was from bowls.

The lithic artifacts retrieved from the excavation included fragments of manos (flat grain-grinding stones), hammerstones, and flakes from chipping. One larger lithic was a large, flat, thin stone that would have been useful for dredging.

Fortunately, the pottery provided good cultural evidence of human activity at the site, along with an opportunity to age-date the site and its period of active service as a water storage reservoir.

About two-thirds of the way down in the reservoir sediments, we uncovered a deer antler fragment that could have been used for digging while dredging the sediments. The antler was age-dated by the University of Colorado using Carbon-14 dating technology. The antler originated in about AD 860.

Using the antler age and the ceramics found in the excavation, we were able to estimate the time period of the reservoir. Then, by estimating the average annual sediment accumulation rate and allowing for sediment dredging operations, we were able to roughly correlate intermediate years with sediment strata, relying on the location of the deer antler for verification.

What We Learned, in a Nutshell
Taking on scientific research at MVNP taught us much about ancient water harvesting and how the Ancestral Puebloans dealt with sediment problems. It also taught us that the early people of Mesa Verde had a tough life in a sometimes harsh environment and that they were good at water management.

We learned that these prehistoric people could collect and store water where modern engineers would say there was none. They did this, in part, by inadvertently creating higher runoff coefficients through land surface modification.

The Morefield Reservoir excavation by the WWE paleohydrologists showed just how the people of Mesa Verde were able to keep their reservoir functioning over a 350-year period; they used extensive dredging of sediments to help maintain storage capacity.

Specific technical findings provided a long list of research results dealing with water storage and sedimentation. More detailed findings were published in Ken Wright's essay, Water for the Anasazi: How the Ancients of Mesa Verde Engineered Public Works," number 22 in Essays in Public Works History , published by the Public Works Historical Society in June 2003.

Kenneth R. Wright, P.E., is the founder of Wright Water Engineers Inc. in Denver, CO, and serves as chief engineer and financial officer. Eric Bikis, P.G., is vice president with Wright Water Engineers' Durango, CO, office. Ernest Pemberton, P.E., is retired chief sedimentation engineer with the US Bureau of Reclamation who serves as an adjunct scientist for WWE.

SW September/October 2

Home + About + Subscribe + News + Calendar + Glossary
Talk + Images + Advertise + Contact Us + Search + Register + Services