American River Flood Frequencies: A Climate-Society Interaction

Kelly T. Redmond

Western Regional Climate Center

Desert Research Institute, Reno, Nevada


The American River flows westward from the crest of the Sierra Nevada above Lake Tahoe to its confluence with the Sacramento River in the city of Sacramento (Figure 1). The North, Middle and South Forks meet upstream from Folsom Dam, located on the upstream outskirts of Sacramento. Folsom Reservoir is drafted each winter to create enough flood control space to absorb a heavy runoff event. A typical very heavy flow event will take 2-3 days to fill this space, so flood managers use maximum annual 3-day average flows rather than peak instantaneous flow to track year-to-year behavior. The reservoir was sized according to flood statistics from 1905-1949. Since 1950, the river has experienced seven 3-day flow values exceeding 90,000 cubic feet per second (90 kcfs), compared with once from 1905-1949, a value of 98 kcfs in the winter of 1927-28 (Figure 2). The latter period includes the two largest, 167 kcfs in February 1986 and 164 kcfs in January 1997, only eleven years apart. What was originally thought to be "500-year" (return interval) flood protection is now estimated to be 75-80 years.

Approximately $40 billion in damageable property lies behind the levees in the city, including the state capitol building and numerous government offices. In reaction the the apparent shift in flood risk, a long-running and spirited community debate has developed over what actions to take. Options include any or all of: increasing the capacity of Folsom Reservoir, increasing the outflow capacity of Folsom Reservoir, increasing the levee system capacity, relocating residents and businesses from parts of the floodplain and limiting future development in the floodplain, and building a new dam near Auburn on the largely unregulated North Fork. None of the options are cheap, some are controversial, and all choices will depend upon estimates of the risk of various sizes of floods. The issue thus turns back to what the climate system will, or could, bring in coming decades.


The American River above Folsom Reservoir drains about 1862 square miles. Topography ranges from about 10,000 feet at the highest elevations to about 10 feet above sea level at its termination. Each of the three tributaries is 80-85 mi long, and has incised deep V-shaped canyons with gradients averaging about 100 feet per mile. Travel times are approximately 18 hours from the upper basin to Folsom Reservoir. Flood peaks from the three adjoining forks arrive almost simultaneously. Large floods are almost exclusively winter events, produced by sustained moist, rapid southwest flow originating in the subtropics, with high freezing levels and thus rain to high elevations (e.g. Redmond and Pulwarty, 1997). Some degree of snowmelt is often present, but rain on saturated soil provides the bulk of the runoff. Flow is oriented nearly perpendicular to the topographic ramp formed by the western slope of the Sierra Nevada. With a stationary pattern, orographic lifting provides a mechanism for prolonged heavy precipitation. Nearby drainages including the Sacramento River are also often in flood at the same time.


Flows are measured just downstream from Folsom Reservoir at Fair Oaks. Reconstructed natural 3-day annual maximum flows (i.e., adjusted for upstream storage) show a decided increase from the first to the second half of the 20th Century.

The change in mid-century is not simply a peculiarity of this one gage or basin. Precipitation gages at various elevations in and near the basin show a similar increase in both the number of wet years and in the maximum 3-day and 10-day totals within each winter. The California "8-station index" (covering most of the Sacramento River drainage) shows similar temporal behavior. Interestingly, the annual precipitation is increasing more slowly or not at all. The most recent two decades have brought an increase in both the number of wet years and the number of dry years. Thus, variability has increased. Floods on the eastern slopes of the Sierra, facing the Great Basin and caused by the same factors, also show an increase over the past several decades. Thus, there is little doubt that the flood frequency increase on the American River is real.

In addition, another change was first noted by Roos (1991), and later elaborated by Dettinger and Cayan (1995). The fraction of the annual runoff from the central Sierra that occurs in late spring has been decreasing for approximately the past 50 years. Relatively more of the annual runoff has been occurring in the winter. Winter and spring temperatures have become warmer in the central Sierra. This has been a more gradual trend rather than an abrupt transition.


In flood frequency analysis, past records are used to assess the likelihood (return interval) of various magnitudes of precipitation and runoff over a range of durations (IACWD, 1982). What is needed is an estimate of what these statistics will be during a time span of interest in the future. This essentially amounts to a forecast of the return interval regime that will apply during that time. This time span is likely to be on the order of a generation or two, about a half-century. This is less than the expected lifetime of control structures, but by that time a new set of societal values is likely to have evolved, with these issues brought up again for re-examination from new perspectives.

The traditional approach has been to use whatever data are available from the past to form these estimates. A dilemma is encountered, however, when the recent record (in this case, the past 50 years) is decidedly different from the previous record. What portion of the past is most likely to be representative of the next half century or beyond?

One is thus led into a consideration of the possible causes of climate variability in this region on decadal time scales. Of particular interest is whether these variations range back and forth between approximate extremes, or appear as monotonic or unidirectional trends. Natural climate variability is generally considered to resemble the former, and possible human-induced changes are usually considered to resemble the latter.

Among the natural kinds of variability are the two phases of ENSO (which produce different responses in the California cool season), regimes and decadal oscillations in the north and central Pacific (e.g., Pacific Decadal Oscillation, PDO), regimes of ENSO behavior (predominance of one phase over the other, etc.), possible modulation of ENSO effects by the PDO, thermohaline circulation variations in the world ocean, and others only imagined, or dimly known or understood at this point.

Among the possible anthropogenic sources of variability are greenhouse gasses (GHG), aerosols, and land use changes around the earth. One expectation from GHG is for a more vigorous hydrologic cycle with more instances of extreme precipitation events (Karl and Knight, 1997). GHG responses are often expected to gradually increase in a unidirectional manner (although this need not necessarily be true in such a highly nonlinear system), whereas aerosols and land use changes could lead to many possible non-linear and non-local responses (e.g., Pielke et al, 1998).


It is against this uncertain climatic backdrop that expensive decisions with potentially lasting effects must be made and defended. While physical scientists debate whether changes are taking place in the climate system, other changes are taking place in the societal value system. Views have increasingly migrated toward favoring free-running rivers, less reliance on structural solutions to flood problems, and greater reliance on behavioral solutions, such as staying off the flood plain and de-emphasizing the placement of permanent and valuable structures there (Mount, 1995). It seems likely that these societal values will continue to change even as the climate changes or varies. Although there are strong opinions on both sides about the relative desirability of behavioral versus structural solutions, the eventual path finally chosen would likely consist of a mixture of actions which preserve some flexibility for future actions.

The problem of whether the current flood regime will continue, or grow, or will relax back to values frequently seen during less active periods in the paleo records (e.g., Meko, 1998), is thus a central focus of attention.

No firm decisions have yet been made as of this writing. However, the issues are not unique to the American River. Webb and Betancourt (1992) have pointed out analogous issues in southern Arizona. Similar problems are likely to be encountered by other cities as they reassess the relationships with their rivers.


Dettinger, M.D. and D.R. Cayan, 1995. Large-scale atmospheric forcing of recent trends toward early snowmelt runoff in California. J. Climate, 8, 606-623.

Earle, C.J., 1993. Asynchronous droughts in California streamflow as reconstructed from tree rings. Quaternary Research, 39, 290-299.

IACWD, 1982. Guidelines for Determining Flood Flow Frequency. Bulletin 17B, Hydrology Subcommittee, Interagency Advisory Committee on Water Data, USGS, Reston VA, 28pp + 155pp appendices.

Karl, T.R., and R.W. Knight, 1998. Secular trends of precipitation amount, frequency, and intensity in the United States. Bull. Amer. Meteor. Soc, 77, 279-292.

Mount, J.F., 1995. California rivers and streams: The conflict between fluvial processes and land use. University of California Press, 359pp.

National Research Council, 1995. Flood Risk Management and the American River Basin: An Evaluation. National Academy Press, Washington D.C., 235pp.

Pielke, R.A. Sr, R. Avissar, M. Raupach, A.J. Dolman, X. Zeng, and A.S. Denning, 1998. Interactions between the atmosphere and terrestrial ecosystems: Influence on weather and climate. Global change Biology, 4, 461-475.

Redmond, K.T. and R.S. Pulwarty, 1997. An overview of the California/Nevada floods of 1997. Proceedings, 10th Conference on Applied Climatology, American Meteorological Society, Oct 20-23, 1997, Reno, NV, 14-17.

Roos, M., 1991. A trend of decreasing snowmelt runoff in northern California. Proceedings, 59th Western Snow Conference, Juneau, AK, 29-36.

Webb, R.H. and J.L. Betancourt, 1992. Climatic variability and flood frequency of the Santa Cruz River, Pima County, Arizona. USGS Water-Supply Paper 2379, 40pp.