HUMAN IMPACTS ON LARGE RIVER FLOODING: A REVIEW, NEW DATA, AND IMPLICATIONS

According to a recent National Research Council panel, "there are ... few cases where human impacts on flood magnitude and frequency have been carefully documented". Although some authors assume that human activities have been proven to magnify downstream flooding, many engineers and planners dismiss evidence for rising trends and deny any smoking-gun evidence for specific human contributions. This schism can be traced from the 1850-60s Ellet and Humphreys and Abbot reports, through the “levees only” policy, to the interagency report on the 1993 flood. Challenges in documenting human magnification of flooding include: (1) separating long-term trends from short-term hydrologic variability; (2) quantifying the contributions of climate, land-use shifts, and river engineering; and (3) incorporating hydrologic shifts into flood-frequency calculations. New approaches help to overcome these challenges. Specific-gage analysis tracks secular shifts in stage for precisely fixed discharge conditions and quantifies changes in flood conveyance. Constant-discharge analysis of long-term gaging records pinpoints the causal mechanisms at work. Finally, stage indexing incorporates conveyance trends into calculations of flood frequencies and probable magnitudes. Analysis of records from several Mississippi and Missouri River stations documents rising flood levels and shrinking recurrence times for large floods driven by levees, navigation structures, and channel constriction. Current guidelines for, and implementation of, flood-frequency analysis in the U.S. largely ignore or sidestep non-stationary behavior on rivers. For example, the Corps of Engineers’ current reanalysis of the Upper Mississippi, Missouri, and Illinois Rivers assumes no effect of land-use change, climate shifts, or downstream engineering structures. Whereas indexing results suggest that the 100-year flood level at St. Louis should be raised >=4 ft, preliminary results of the Corps study lower the 100 year flood by 0.5 ft. Updated probabilities are required for a new generation of U.S. floodplain maps, the cost of which is estimated at $800 million to $1 billion nationwide. Fundamental questions about data stationarity and human- and climate-driven impacts on flood hazard must be resolved before such an enormous investment.