Experimental Flooding

in Grand Canyon
Scientists monitor a controlled deluge that was staged
in the early spring of 1996 solely for the benefit
of the environment in and around the Colorado River

by Michael P. Collier, Robert H. Webb and Edmund D. Andrews

F or more than three decades, Glen

Canyon Dam has impounded
the flow of the Colorado River
above Grand Canyon, the vast winding
chasm in America’s southwestern desert
pated that there might be serious envi-
ronmental consequences downstream.
But over the years, scientists, government
officials and professional river guides
have become increasingly aware of
ing with a gentle daily ebb and flow dic-
tated entirely by the electrical power
demands of distant cities. The dam thus
eliminated the normal seasonal varia-
tion in river flow and ended the im-
that ranks as one of the wonders of the troubling changes within Grand Canyon. mense floods that had annually washed
natural world. Although many people These alterations have occurred be- through the canyon. Although these
recognized that damming the flow would cause the dam replaced the Colorado’s floods had lasted only a few weeks of the
destroy the river upstream, few antici- natural pattern of forceful summer flood- year, they had been the principal force

Copyright 1998 Scientific American, Inc.

 sculpting the river corridor. The flood- tamarisk, camelthorn and even bermu- sis passed, and engineers were able to re-
waters routinely stripped all but the high- da grass, took root on beaches that had design the spillways to minimize “cavi-
est vegetation from the channel banks, previously been bare. Mature mesquite tation.” This phenomenon (formation
deposited sandbars and plucked boul- trees growing at the old high waterline of a partial vacuum within a moving
ders out of rapids. After Glen Canyon began to die. liquid) had sucked material from the
Dam went into service, exotic flora en- But not all changes brought about by tunnel walls and caused them to erode
croached, sandbars disappeared and the damming of the river were obviously with startling speed.
boulder piles clogged the main channel. undesirable. Trout—which did not live The downstream effects of the 1983
So as dozens of scientific observers there before in the relatively warm, tur- flood also took others by surprise. When
(including the three of us) made ready, bid waters of the free-running river— the waters receded from the flooded
the secretary of the interior, Bruce Bab- flourished in the cold, clear water below banks, scientists and guides familiar with
bitt, launched a bold experiment in en- the dam. Stabilization of flow favored the river were stunned to see many of
vironmental restoration at 6:20 A.M. on trees and shrubs on the riverside, which the formerly shrunken beaches blanket-
March 26, 1996, opening the first of provided new homes for some endan- ed with fresh sand. The flood had killed
four giant “jet tubes” at Glen Canyon gered birds. The green ribbon of new some exotic vegetation that had grown
Dam. Over the next nine hours, the oth- vegetation made the once barren canyon artificially lush and had also partially re-
er three tubes and the eight hydroelec- appear more hospitable to other kinds of stored riverine animal habitats in many
tric turbines added to the torrent, which wildlife as well—and to countless camp- spots. Had several years of ordinary
grew to 1,270 cubic meters per second— ers who traveled the river for recreation. dam operations followed, many people
a discharge 50 percent greater than the Indeed, the many beneficial changes would have hailed the 1983 flood for
maximum flow through the turbines to the canyon ecosystem may have di- improving conditions in the canyon. In-
alone. As the surge of water mounted, verted attention from some of the more stead runoff in the Colorado River ba-
the surface of the river rose five meters disturbing trends. It was not until 1983 sin during the next three years remained
higher than usual. In all, 900 million that many scientists and environmen- quite high, and the operators of Glen
cubic meters of water coursed through talists took full notice of the important Canyon Dam were forced to release huge
the canyon during the weeklong exper- role that floods originally played in shap- amounts—an average of 23 billion cubic
iment. Never before had an intentional ing the canyon. In June of that year, a meters of water every year. Flows com-
flood been released specifically for envi- sudden melting of the winter snowpack monly reached 1,270 cubic meters per
ronmental benefit, and we were eager rapidly filled Lake Powell and forced the second, for at least brief periods, each
to help assess the results. operators of Glen Canyon
Dam to release water at a
MICHAEL P. COLLIER
MICHAEL P. COLLIER

A Changed River rate of 2,750 cubic meters

per second. This flow was

A long with many other scientists who

monitored the experimental flood-
ing, we have been aware that conditions
far smaller than some re-
corded flood episodes, but
it still constituted a mo-
in the canyon had been evolving dramat- mentous event.
ically since Glen Canyon Dam began This emergency release
operations in 1963. After construction in 1983 required the first
of the dam, virtually all sediment com- use of the “spillways”—
ing from upstream was trapped above giant drain tunnels carved
the dam, in the newly created Lake Pow- into the walls of Glen Can-
ell, and most sandy beaches in Grand yon alongside the dam.
Canyon began slowly but steadily to The operators of the dam
vanish. By the time the test flood was had at first been dis-
planned, some rapids in the river had mayed—and then gravely
become so obstructed by coarse debris concerned—to see the out-
swept down from side canyons that par- flow turn red as swiftly
ticular stretches had become extremely moving water plucked first
difficult to navigate. The bridled river concrete and then great
did not have sufficient power to clear blocks of sandstone from
away the boulder-filled deposits. Many the spillway tunnels. Some
people familiar with the canyon had feared that destruction of
also observed dramatic alterations to the spillways could catas-
the vegetation since the dam was built. trophically undermine the
Native coyote willow, as well as exotic dam. Fortunately, the cri-

JETS OF WATER (opposite page) emerge during the experimental flood

from four steel drainpipes built into the base of Glen Canyon Dam
(right). The water stored in the adjacent reservoir, Lake Powell, can also
flow through the hydroelectric turbines situated underneath the dam or,
if there is urgent need to lower the lake, through the two “spillway tun-
nels” carved into the canyon walls (visible at right, below the dam).

Experimental Flooding in Grand Canyon Copyright 1996 Scientific American, Inc.

year through 1986. The beaches that had that the beaches nearest Glen Canyon yon, but it was not accumulating on the
been built up in 1983 soon washed away. Dam might be inadvertently washed banks. Rather it was settling out of sight
A single flood, it seemed, could create away. And anthropologists working in on the bottom of the river.
beaches; frequently repeated floods could the canyon expressed concern that new Along with others at the U.S. Geolog-
destroy them. flooding would accelerate erosion and ical Survey (USGS) and the Bureau of
threaten the integrity of archaeological Reclamation’s Glen Canyon Environ-
Time for Another Flood? sites next to the river. mental Studies program, we were cer-
Yet the overall sentiment was that pur- tain that a flood would stir up these de-

A s scientists learned more about risks

and benefits of flooding in the can-
yon, many of those interested in the fate
poseful flooding would be more bene-
ficial than harmful and should be ar-
ranged. By 1993 the murmurs in favor
posits and drape them along the banks,
just as the river had done before the
dam had reined it in. But what sort of
of the river began to recognize the need of a flood had turned to shouts. Some of flood would be most appropriate? The
to restore flooding of some type. Most the loudest voices came from river guides people debating that question agreed
geologists who had studied the move- who were being forced to find camping that the best time of year for an initial
ment of sediment by the Colorado River sites on smaller and smaller beaches— test would be during a narrow window
were convinced that an artificial flood despite the fact that millions of metric in late March, when fish were least like-
would benefit the canyon, and they be- tons of sand were reaching the Colora- ly to be spawning and troublesome
gan championing that idea within the do every year by way of its two main tamarisk plants would not yet be able
scientific community in 1991. But dur- tributaries below the dam, the Paria and to germinate. A date at that time would
ing discussions, some biologists worried Little Colorado rivers. Under the normal also assure that most bald eagles and
aloud that a flood would jeopardize the operating regime of Glen Canyon Dam, waterfowl that had wintered in the can-
gains that had been made within the only 450,000 metric tons of this sand yon would have already left. Still, the
canyon by several endangered species. wash downstream and out of Grand optimum choice for the size of the flood
A few geologists, too, were concerned Canyon. So sand was filling the can- remained elusive.
One reason for that difficulty was that
4,000 the quantity of sand moved by a river

JENNIFER C. CHRISTIANSEN
NATURAL FLOOD varies quite strongly with the rate of dis-
3,500 1957 charge: when the discharge rate doubles,
the flux of sand increases eightfold. Con-
(CUBIC METERS PER SECOND)

3,000 sequently, for a given flood volume, more

sand will be stirred up and deposited
RATE OF DISCHARGE

2,500 on the banks by a large flood that runs

for a short time than by a lesser flood of
2,000 longer duration. One of us (Andrews)
CONTROLLED FLOOD argued for a release at the rate of 1,500
1,500
1996 cubic meters per second, which would
have been close to two thirds the size of
the typical annual flood before the dam
1,000
was built. After all, if the goal was to re-
store a critical natural process, why not
500
try roughly to approximate that level?
But there was an important logistical
0 constraint: flows greater than 1,270 cu-
MARCH APRIL MAY JUNE JULY AUGUST
bic meters per second through the dam
SOURCE: Robert H. Webb would require the use of the spillways.
Despite having made repairs and im-
MICHAEL P. COLLIER

provements, officials at the Bureau of

Reclamation were reluctant to risk rep-
etition of the frightening experience of
1983. Restricting the flood to 1,270 cu-
bic meters per second would also mini-
mize the threat to an endangered species
of snail that lived near the dam. Most
proponents of flooding felt that this lev-
el was a reasonable compromise. They

DISCHARGE of water during the experi-

mental flood of 1996 may have appeared
extremely powerful (photograph), but the
flow maintained for that one week is
dwarfed by natural events of the past, such
as the flood of 1957 (chart), which endured
for much of the spring and summer.

84 Scientific American January 1997 Copyright 1996 Scientific American, Inc. Experimental Flooding in Grand Canyon
 agreed that the flood would last one
week—enough time to redistribute a
considerable amount of sand but not so
long as to deplete all sand reserves at the
bottom of the river.
On the eve of the test, our biggest fear
was that the water would not have the
power needed to build sizable beaches.
But John C. Schmidt, a geologist at Utah
State University who had also favored
the flooding experiment, had a bigger
concern. He worried that something
might unexpectedly go wrong: Would
scientists in their arrogance ruin what
was left of the heart of Grand Canyon?

On a Rising Tide

O n March 26 the flood began on

schedule. The waters of the river
rose and raced down the canyon. On
signal, scientists from the USGS released
30 kilograms of a nontoxic fluorescent
dye into the river a short distance down-
stream from the dam. They used the
chemical to track the velocity of the wa-
ter by measuring the arrival of this dye
at six sites spaced throughout the can-
yon, where they had placed sensitive
fluorometers. A numerical model devel-
oped by researchers at the USGS accu-
rately predicted the progress of the
MICHAEL P. COLLIER

flood. The model and measurements

showed that the floodwaters accelerated
as they ran through the canyon, pushing
riverwater so far ahead that the first crest
reached Lake Mead at the downstream
end of the canyon almost a day before
LAKE
the actual waters of the flood arrived. POWELL
On its way west to Lake Mead, the PARIA RIVER
flood reshaped many parts of the river. NORTH
For example, at a stretch of rapids called
GLEN CANYON
DAM
COLORADO RIVER flows westward
across Arizona from Lake Powell to Lake ARIZONA
Mead (map). Between these points, the
river receives massive injections of sandy
sediment from the Paria River (photo-
graph) and the Little Colorado River, its
two main tributaries.
COLORADO RIVER
LAVA FALLS

LITTLE COLORADO
RIVER

LAKE GRAND CANYON

MEAD NATIONAL PARK
BOUNDARY
ROBERTO OSTI

0 10 20 30 40 50 MILES

0 10 20 30 40 50 60 70 KILOMETERS

Copyright 1996 Scientific American, Inc. Scientific American January 1997 85

 Lava Falls, about 300 kilome- positioned originally near the top of the
ters below the dam, the river rapids. Despite their considerable size
rose against a fan-shaped (up to 0.75 meter across), all 10 rocks
bank of loose mud and boul- traveled downstream during the flood.
ders that had been formed Using directional antennas, we subse-
one year earlier after a debris quently located eight of the boulders.
flow roared down a small The great stones had moved, on average,
side canyon. The material 230 meters.
deposited by that cascade of Besides tracking boulders at Lava
rock and mud had narrowed Falls, we worked with several colleagues
the Colorado—normally 50 to measure the deposition of sand at
meters wide there—by almost some other key locales. For those stud-
20 meters. Although some ies, we chose five eddies—places where
geologists had previously con- the river widens abruptly, and water
cluded that very large floods swirls upstream near the banks. Em-
would be required to clean ploying laser tracking equipment and a
out such constrictions, we small boat equipped with a sonar depth
believed this flood would be finder, we charted the sandy bottom dur-
sufficient to do the job. ing the flood. The results were quite sur-
And so we were quite prising. We found that a great deal of
pleased to see just how effec- sand accumulated in the first 36 to 48
tive the experimental flood hours. But as the influx of sand slowed,
proved. As discharge of the the bottom of the eddy began to lose
river surpassed 850 cubic me- sand back into the main channel.
ters per second at Lava Falls This behavior initially puzzled us, but
PHOTOGRAPHS BY MICHAEL P. COLLIER

on March 27, the energized after we examined the measurements

water quickly cut through the more carefully, we realized that much
new debris fan, reducing its of the sediment had originally settled
size by one third. We studied above its so-called angle of repose, an
that event by placing radio unstable configuration that resulted in
transmitters in 10 large stones some newly deposited sand slumping

SAND DEPOSITION within an eddy, a place where water swirls in the up-
stream direction near the banks, raised the bed of the river along one mar-
gin (tan areas in diagrams) in the first days of the flood. Later during the
flood, much of that sand escaped back into the main channel (blue areas in
diagrams). To collect this record of sediment accumulation and removal, a
boat fitted with an acoustic echo sounder (photograph at left ) measured
the depth of the water, and surveying equipment on land tracked the posi-
tion of the boat (photograph at top).

PREFLOOD DAY 1 DAY 2 DAY 3 DAY 4 DAY 5 DAY 6 DAY 7 POSTFLOOD

822
(SHALLOWER WATER)

821
820
819
METERS ABOVE SEA LEVEL

818
JENNIFER C. CHRISTIANSEN; SOURCE: CHRISTOPHER E. JOHNSTON

817
816
815
814
(DEEPER WATER)

812

810

NORTH 808
ELEVATION
0 100 METERS
OF RIVER BOTTOM

86 Scientific American January 1997 Copyright 1996 Scientific American, Inc. Experimental Flooding in Grand Canyon
 back into the main channel. Still, we form in a line where the main
found that the overall amount of sand downstream current rushes
after the flood had increased in all five past a slower, upstream-flow-
places we mapped. ing eddy. Nelson observed
Many other scientists made impor- that as the main current push-
tant observations during the course of es these vortices downstream,
the flood. Near the lower end of Grand the vortices tip over, because
Canyon, our colleague J. Dungan Smith flow is slowed near the chan-
measured the velocity of the river and nel bottom where friction is
concentration of sediment held in sus- greatest. In this canted posi-
pension by the turbulent water. His goal tion, he reasoned, the vorti-
is to compare the quantity of sediment ces should then act to sweep
washed out of the canyon during the sediment out of the main cur-
flood with the amount normally deliv- rent and into the eddy.
ered into the canyon by the Paria and But sediment came and
Little Colorado rivers. Smith is still an- went within the eddy at rates
alyzing his data, but he should soon be far greater than anticipated.
able to predict how often floods could With a sinking feeling, Ru-
be staged without depleting the existing bin and Nelson watched as
sand reserve. $70,000 worth of borrowed
Several other scientists took special equipment was first buried,
interest in the movement of sand. Using then excavated and finally
optical sensors and sonar equipment carried away by the water.
borrowed from his oceanographer col- They were fortunate enough
leagues at the USGS, David M. Rubin to have collected sufficient
studied the sediment concentration of data to show that the vortex
water entering an eddy and character- “sediment pump” operated
ized the fine-scale patterns in the depo- as they had predicted. So
sition of this sand. Working at the same their ideas withstood the test
site, Jon M. Nelson documented the cu- flood, even though much of
rious behavior of swirling vortices that their equipment did not.

SCIENTIFIC STUDIES carried out during the experimental flood included

documentation of fine-scale patterns of sand deposition using plaster molds
(bottom right), time-lapse videography of the floodwaters (bottom left) and
measurement, by means of a directional antenna (top right), of the displace-
ment of boulders that were fitted with radio transmitters.
PHOTOGRAPHS BY MICHAEL P. COLLIER

Experimental Flooding in Grand Canyon Copyright 1996 Scientific American, Inc. Scientific American January 1997 87
 As expected, a good deal to protected areas along the riverbank.
of the newly deposited sand Other investigators determined that the
quickly eroded, but months floodwaters had hardly disturbed the
later much of it still remained ubiquitous cladophora algae and asso-
at those sites monitored by ciated invertebrates, which constitute an
scientists—and at many oth- important source of food for fish.
er places as well. During the But the effects on other components
summer of 1996, many long- of the local biota are still a matter of in-
time observers believed the tense debate. Lawrence E. Stevens, a bi-
Colorado River had taken ologist with the Bureau of Reclama-
on something of its original tion, has studied the river for 25 years
appearance. Those impres- as an entire suite of animals—some en-
sions echoed the more care- dangered—migrated into the canyon and
ful assessment of Lisa H. survived in the artificial environment
Kearsley, a biologist working created by Glen Canyon Dam. He is
for the Bureau of Reclama- concerned that intentional flooding may
tion. She tracked the fate of threaten the existence of some species
almost 100 beaches through- protected by the Endangered Species Act,
out the canyon and conclud- such as the humpback chub (a fish), the
ed that 10 percent of them southwestern Willow Flycatcher (a bird)
were diminished by the flood, or the Kanab ambersnail. But we would
whereas 50 percent were aug- argue that floods were part of the natu-
mented, and the remainder ral cycle of the Colorado River in the

MICHAEL P. COLLIER
were unaffected. Six months past, and many species, both common
after the flood, she found and endangered, have adapted to this
that much sand had slipped process as long as there has been a Grand
back into the river, but there Canyon—for about five million years.
was still more beach area Restoration of flooding may be detri-
than before. mental to some organisms, but we and
The expanded beaches many of our colleagues hypothesize
should please campers in that in the long run a collection more
years to come, but scientists resembling the native fauna will return.
are also anxious to know
how the flood might have af- Epilogue
fected many less vocal resi-
dents of the canyon. Because
the earlier unintentional flood
of 1983 had hurt the trout
D id the flood work? It deposited sig-
nificant amounts of sand above the
normal high-water line and rejuvenated
fishery, some biologists were some backwater habitats important to
particularly concerned that spawning fish. The flood widened the
the experimental flood of two largest rapids on the river. Archae-
1996 would wash many fish ological sites along the edge of the river
far downstream. To find out, were neither helped nor hurt by the high
biologists stationed below water; most of the encroaching vegeta-
Lava Falls during the experi- tion was similarly unaffected.
mental flood placed nets in So in our view the environmental ben-
the river. These scientists cap- efits outweighed any damage. But one
tured a few more trout than needs to consider other costs as well.
they would have otherwise Five months after the flood, David A.
GLEN CANYON ENVIRONMENTAL STUDIES

done, but their tests did not Harpman, an economist with the Bu-
show any flushing of native reau of Reclamation, was analyzing fac-
fishes, whose ancestors had, tors that bear on the final price tag. Be-
after all, survived many larg- cause power had been continuously gen-
er natural floods. The biolo- erated during the flood even at times
gists surmised that the native when demand was low, and because the
species (and most of the trout) huge quantity of water sent through jet
must have quickly retreated tubes produced no electricity at all, he

LAVA FALLS, a stretch of rapids in the Colorado, was narrowed by coarse,

rocky material that had washed down a side canyon and spread into a fan-
shaped deposit. An aerial photograph taken before the flood (top left) shows
an obvious constriction in the river. A matching photograph taken after the
flood (bottom left) reveals that much of the debris has been cleared away.

88 Scientific American January 1997 Copyright 1996 Scientific American, Inc. Experimental Flooding in Grand Canyon
REJUVENATED BEACHES, such as the one enjoyed by these edge to a more natural configuration. Such changes should, for
kayakers, signal that the flood restored habitats along the river’s example, benefit native fish, which spawn in the shallows.

estimates that the Bureau of Reclama- flood need not last seven days. Smith to vary in magnitude from year to year,
tion has foregone about $1.8 million in believes Grand Canyon beaches can be the better to mimic natural variability.
lost revenue (about 1 percent of the to- improved by floods staged perhaps ev- Will there be more floods? Probably—
tal yearly income from the sale of elec- ery year, as long as incoming sediment both in Grand Canyon and elsewhere.
tricity). Add to this expense the price of from the Paria and Little Colorado riv- We have studied several other Ameri-
the scientific studies, and the total cost ers is at least as great as the amount of can rivers controlled by dams, and they,
of the experiment almost doubles. sediment carried out of the canyon dur- too, would benefit from periodic floods.
Because similar expenditures will be ing a flood. One of us (Webb) argues for So the ideas and instrumentation devel-
incurred during future floods, the Bu- an initial release of as much as 2,800 oped by scientists working within Grand
reau of Reclamation will want to know cubic meters per second to scour debris Canyon during the 1996 experimental
precisely how big and how often floods fans, followed by an immediate drop to flood could soon help restore natural
will be needed to support the environ- more moderate beach-building levels. conditions within and around many oth-
ment. The answers are far from clear. Andrews emphasizes that under any sce- er rivers across the nation and, perhaps,
All scientists involved agree that a future nario, artificial floods should be made throughout the world. SA

The Authors Further Reading

MICHAEL P. COLLIER, ROBERT H. WEBB and ED- Recirculating Flow and Sedimentation in the Colorado River
MUND D. ANDREWS have long cherished the splendor of in Grand Canyon, Arizona. J. C. Schmidt in Journal of Geology,
Grand Canyon. Collier, who considers himself a writer and Vol. 98, No. 5, pages 709–724; September 1990.
photographer rather than a true scientist, also maintains an ac- Sediment Transport in the Colorado River Basin. E. D. Andrews
tive medical practice in Flagstaff, Ariz. He earned a master’s de- in Colorado River Ecology and Dam Management. National Research
gree in geology from Stanford University in 1978 and worked Council, Committee to Review the Glen Canyon Environmental Stud-
for six years as a river guide in Grand Canyon before he began ies. National Academy Press, 1991.
collaborating with U.S. Geological Survey scientists. Webb re- Flow Regulation, Geomorphology, and Colorado River
ceived a doctorate from the University of Arizona in 1985 and Marsh Development in the Grand Canyon, Arizona. L. E.
then joined the staff of the USGS as a hydrologist. Since 1989 he Stevens, J. C. Schmidt, T. J. Ayers and B. T. Brown in Ecological Ap-
has also taught at the University of Arizona. Andrews worked plications, Vol. 5, No. 4, pages 1025–1039; November 1995.
as a river guide in Grand Canyon from 1969 to 1974. Three Dams and Rivers: A Primer on the Downstream Effects of
years later he earned a doctorate from the University of Califor- Dams. Michael Collier, R. H. Webb and J. C. Schmidt in U.S. Geolog-
nia, Berkeley, and has since done research for the USGS in its ical Survey Circular, Vol. 1126; June 1996.
water resources division. Andrews also maintains an ongoing Grand Canyon: A Century of Change. Robert H. Webb. Universi-
affiliation with the University of Colorado at Boulder. ty of Arizona Press, 1996.

Experimental Flooding in Grand Canyon Copyright 1996 Scientific American, Inc. Scientific American January 1997 89