Evaluating the effectiveness of floodplain restoration on the North Fork John Day River, Northeast Oregon, USA

Over the last decade hundreds of river restoration projects intended to maintain, protect, and restore watersheds, rivers, and habitat for native species in the Pacific Northwest have been implemented. By some counts, investment in watershed restoration exceeds hundreds of millions of dollars annually yet the effectiveness of these efforts remains an elusive question (Roni, 2005). Remote sensing and GIS technologies show great promise for large-scale river monitoring, however most natural resource organizations who implement these projects have limited budget and staff and would benefit from simple, low cost monitoring techniques that use readily available imagery. We used 1:24000 digitized orthorectified resource imagery from 1995, and National Agriculture Imagery Program (NAIP) digital orthophotography from 2005 to assess the effectiveness of floodplain restoration on a 16 km reach of the North Fork John Day River. Between 1993 and 1997 this section was restored by mechanically removing, reshaping, and revegetating cobble-boulder tailings piles left from dredge mining. The project was intended to directly improve floodplain function (i.e. inundation, riparian habitat) and indirectly improve instream habitat (pools, spawning) by reconnecting the active river channel with a reconstructed floodplain surface. Project effectiveness was not well documented initially in terms of quantifying floodplain functional area improvement or channel condition and response at the river-reach scale. Our objectives were to field-verify remote sensing measurements of response variables to test the applicability of available remote sensing imagery for project effectiveness monitoring, and to quantify adjustment in river response variables, using a "before-after" case study approach. Bracketing restoration activities with 1995 and 2000 imagery, we developed and tested methods for acquisition and processing of digital imagery and identified a core set of response variables to sample. Methods for sampling and data capture to characterize floodplain and channel conditions before and after treatment, and for quantifying effects of treatment were also developed. Response variables included: channel sinuosity, depositional area (bars), active channel width, floodplain area, and dominant vegetation type. We found mixed results in the post-project NAIP image interpretation due to resolution and sampling methods, but small, positive changes in some river response attributes (sinuosity, vegetation cover) were observed. We found complex mixed response in other variables (active channel width, floodplain area), presumably due to the compounding effects of previous habitat enhancement work and limited response and recovery time. Evaluating restoration treatment effectiveness on larger river systems presents numerous technical challenges. Such evaluations should consider river management history, original project objectives and design criteria, and the occurrence of channel-altering flows since treatment.