An understanding of the response of a fluvial system to past climatic changes is useful for predicting its response to future shifts in temperature and precipitation. To determine the response of the Columbia River system to previous climatic conditions and transitions, a well-dated sequence of floodplain development in the Wells Reservoir region was compared with the paleoenvironmental history of the Columbia River Basin. Results of this comparison indicate that aggradation episodes, occurring approximately 9000-8000, 7000-6500, 4400-3900, and 2400-1800 yr B.P., coincided with climatic transitions that share certain characteristics. The inferred climates associated with aggradation had at least moderate rates of precipitation that occurred mainly in winter coupled with moderate winter temperatures. Such conditions would have resulted in the buildup of snowpacks and a high frequency of rain-on-snow events. The warming and precipitation increases predicted for the Pacific Northwest under most CO2-doubling scenarios are likely to repeat these conditions, which could increase the frequency of severe, sediment-laden floods in the Columbia River Basin.

No abstract

Data from prehistoric fluvial deposits can be used to extend the flood history of a river valley beyond historical records, thus increasing our understanding of variability in large, low-frequency flood events and providing a valuable means for paleoenvironmental reconstruction. We have applied this form of analysis to fluvial deposits from an archaeological site on the upper Columbia River in the state of Washington dating from 120 A.D.∗ to 1948 A.D. It was our expectation that, had flood frequencies remained constant, sedimentation event frequency would conform to an exponential function derived from the Wolman and Leopold model of vertical floodplain accretion. Our findings deviate from this model, showing that flood frequencies comparable to those of the twentieth century existed prior to 1020 A.D.∗ and after 1390 A.D.∗ Large floods were three to four times more common during the intervening centuries. On the basis of field evidence, we can rule out changing channel geometry, leaving climatic conditions as the most probable factors controlling this variation in flood frequency.

No abstract

We explored methods for extrapolating mainstream channel lengths of first‐order drainage basins from synoptic data to characterize them for numerical watershed modeling. We analyzed four catchments in a climatologically semiarid arid geologically homogeneous region east of the Cascade Mountains in Washington state. Within each of these catchments, we identified stream channel networks manually from 1:24,000‐scale topographic maps, and from 50‐m resolution digital elevation models using commercially available drainage network extraction methods. A least squares fit of logarithms of mainstream length plotted against basin area established a regression relation to use for predicting mainstream lengths in the smallest subbasins. To test our relation, we compared predicted mainstream lengths with first‐, second‐ and third‐order channel lengths measured from low‐altitude aerial photographs of a representative fourth‐order basin. Our results indicate that relations of mainstream length to basin area derived from coarsely gridded data (e.g., 30 m) cannot be used to characterize stream and basin morphometry in the smallest basins due to the presence of hydrologic and geometric controls (i.e., thresholds) that limit the mainstream channel length and total basin length in first‐, second‐, and third‐order basins. The presence of these thresholds potentially constrains the range over which theoretically self‐similar, or fractal, relationships can be applied to stream‐channel networks.