Little is known about how active stream network expansion during rainstorms influences the ability of riparian buffers to improve water quality. We used aerial photographs to quantify stream network expansion during the wet winter season in five agricultural catchments in western Oregon, USA. Winter stream drainage densities were nearly two orders of magnitude greater than summer stream densities, and agricultural land use was much more abundant along transient portions (e.g. swales, road ditches) of stream networks. Water moving from agricultural fields into expanded stream networks during large hydrologic events has the opportunity to bypass downstream riparian buffers along perennial streams and contribute nonpoint-source pollutants directly into perennial stream channels.

This paper compares approaches to quantifying land cover/land use (LCLU) in riparian corridors of 23 watersheds in Oregon's Willamette Valley using color infrared aerial photography (AP) and Thematic Mapper (TM) imagery. For each imagery source, LCLU adjacent to stream networks were quantified across a range of lateral and longitudinal scales. Single‐date AP data had higher LCLU class accuracies than the multi‐date TM data, except for row crops. Correlations among LCLU classes for the two imagery sources increased with increased spatial extent. In general, LCLU proportions for AP and TM differed, but lateral/longitudinal patterns were similar. An aggregated vegetation class comprised of forest, shrub/scrub, and grass/forb was strongly associated with a fish index of biotic integrity (IBI) for both AP and TM data, although AP correlations were higher. Highest fish IBI correlations for both data sources were with the aggregated vegetation class close to the stream and for the longest longitudinal scales. The row crop class was strongly associated with stream nitrate for both data sources, although correlations for the TM data were higher. Stream nitrate correlations were strongest for the widest lateral and longest longitudinal scales. Overall, both single‐date AP and multi‐date TM imagery appear to have potential for use in estimating indicators of stream ecological condition.