ESTIMATION OF JUVENILE SALMON HABITAT IN PACIFIC RIM RIVERS USING MULTISCALAR REMOTE SENSING AND GEOSPATIAL ANALYSIS D. C. WHITED*, J. S. KIMBALL, M. S. LORANG and J. A. STANFORD Flathead Lake Biological Station, The University of Montana, Polson, Montana, USA ABSTRACT We conducted a regional classification and analysis of riverine floodplain physical features that represent key attributes of salmon rearing habitats. Riverine habitat classifications, including floodplain area and river channel complexity, were derived at moderate (30 m) spatial resolution using multispectral Landsat imagery and global terrain data (90 m) encompassing over 3 400 000 km 2 and most North Pacific Rim (NPR) salmon rivers. Similar classifications were derived using finer scale (i.e. ≤ 2.4-m resolution) remote sensing data over a smaller set of 31 regionally representative flood plains. A suite of physical habitat metrics (e.g. channel sinuosity, nodes, floodplain width) were derived from each dataset and used to assess the congruence between similar habitat features at the different spatial scales and to evaluate the utility of moderate scale geospatial data for determining abundance of selected juvenile salmon habitats relative to fine scale remote sens- ing measurements. The resulting habitat metrics corresponded favorably (p < 0.0001) between the moderate scale and the fine scale flood- plain classifications; a subset of these metrics (channel nodes and maximum floodplain width) also were strong indicators (R 2 > 0.5, p < 0.0001) of floodplain habitats defined from the finer scale analysis. These relationships were used to estimate the abundance and distri- bution of three critical shallow water floodplain habitats for juvenile salmon (parafluvial and orthofluvial springs, and shallow shore) across the entire NPR domain. The resulting database provides a potential tool to evaluate and prioritize salmon conservation efforts both within individual river systems and across major catchments on the basis of physical habitat distribution and abundance. Copyright © 2011 John Wiley & Sons, Ltd. key words: juvenile salmon; floodplain; river habitat; Landsat; Quickbird; remote sensing Received 17 May 2011; Revised 12 July 2011; Accepted 27 July 2011 INTRODUCTION River flood plains are considered to be among the most biologically productive environments and can sustain very high biodiversity (Tockner and Stanford, 2002; Opperman et al., 2010), often because of great diversity and abundance of aquatic habitats (e.g. pools, ponds, off-channel springs, tributaries). Comprehensive and systematic geospatial data describing habitats in complex aquatic-terrestrial landscapes (e.g. flood plains) are enabled and enhanced by remote sensing and Geographic Information Systems (GIS) tools (Lunetta et al., 1997; Mertes, 2002; Hall et al., 2007; Bryant and Woodsmith, 2009; Carbonneau et al., 2011) and pro- vide for better understanding of river-floodplain systems. Geospatial data applicable to the analysis of river landscapes exist at fine (≤ 5-m resolution) to moderate (10- to 100-m resolution) spatial scales. Detailed assessments have been conducted for in-stream and floodplain habitats (Whited et al., 2002; Gilvear et al., 2004; Legleiter et al., 2004; Lorang et al., 2005; Gilvear et al., 2008), riparian vegetation communities (Johansen et al., 2007; Whited et al., 2007), river geomorphology (Gupta and Liew, 2007), shallow water bathymetry (Alder-Golden et al., 2005; Fonstad and Marcus, 2005; Carbonneau et al., 2006; Conger et al., 2006; Lane et al., 2010) and grain-size distrubution (Carbonneau et al., 2004; Carbonneau et al., 2005; Dugdale et al., 2010). These assessments have largely been conducted at relatively fine scales and limited spatial extents (e.g. < 100 km 2 ) using airborne or satellite multispectral remote sensing imagery (e.g. Quickbird, IKONOS). However, this scale of analysis is generally prohibitive over larger areas because of increased cost and computa- tional requirements. Alternatively, moderate scale (e.g. 30-m resolution) satellite imagery (e.g. Landsat Thematic Mapper, ETM+) is routinely collected on a global basis and is now widely available with supporting geospatial data at limited cost from public sources (e.g. USGS http://weld. cr.usgs.gov). These data are generally too coarse to depict detailed floodplain habitats. However, moderate scale geospa- tial data can potentially be used with finer scale data to link detailed aquatic habitat assessments over limited areas to larger numbers of flood plains distributed over a more extensive geographical domain. *Correspondence to: D. C. Whited, Flathead Lake Biological Station, The University of Montana, 32125 Bio Station Lane, Polson, Montana 59860, USA. E-mail: diane.whited@umontana.edu RIVER RESEARCH AND APPLICATIONS River Res. Applic. (2011) Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/rra.1585 Copyright © 2011 John Wiley & Sons, Ltd.