Overview

Science Team:
Jack Stanford (PI, UMT); John Kimball (UMT); Ric Hauer (UMT); Mark Lorang (UMT); Niels Maumenee (UMT); Dan Goodman (MSU); Diane Whited (UMT); Kyle McDonald (JPL); Erika Podest (JPL); Andrew Neuschwander (UMT); Don Schenck (UMT); David Kuhn (GTG); Eric Sack (GTG)

Funding:

The Gordon and Betty Moore Foundation, the National Science Foundation

Activities:
Major science activities conducted for the Typology Project thus far include:

1. Development of a data processing workflow to derive basin and floodplain complexity metrics
1. Development of an accurate and reproducible large water feature extraction technique
2. Development of a floodplain delineation process
2. Estimation of juvenile fish populations based on classification results
3. Development of an enterprise geodatabase

The Typology database is now available!!

Project Summary:

The Pacific Rim River Typology Project is a remote-sensing based classification of salmon producing rivers across the north Pacific Rim (NPR).

Typology study area with outlines of JPL Landsat mosaic coverage.

The goal of this work is to produce a web-accessible decision support database that will assist salmon conservation around the Pacific Rim, based on a robust classification (typology) of rivers and river habitats and aimed at conserving the existing and potential production of salmon in the context of the ocean domains influencing the rivers and salmon that spawn and rear in them.

A key component to this study is that salmon productivity in freshwater is linked to complex biologic and hydrogeomorphic pathways that we refer to generally as a shifting habitat mosaic (SHM) (Stanford, J. A., M. S. Lorang, et al. 2005). Specifically, our approach assumes that the greater the biophysical complexity of the river systems, the greater the production potential. Potential correlations between complexity and production will be examined through the use of spatially explicit relational databases including spawner-recruitment data where available.

Google Earth ® sample of initial Typology floodplain processing

Samarga Floodplain, Kamchatka (160kb) Full .KML (2mb)

This file demonstrates some of the outcomes of river classifications within the Typology project. Click here if you don't have Google Earth ® .

Details

Basin Feature Delineation
By leveraging certain characteristics of two orthorectified datasets, Digital Elevation Model (DEM) and Landsat, a new and innovative method for delineating water feature information is possible. The technique builds off of the existing DEM based flow models for developing potential stream networks and a new method for delineating and classifying water body features visible in digital multi-spectral satellite remote sensing data.

Floodplain Delineation
An automated flood plain delineation algorithm has been developed using existing DEM information and derived stream networks generated from satellite remote sensing imagery. In mountainous regions (e.g Kitlope, Skeena basins), the floodplain algorithm works very well (Fig. 2a).  However in relatively flat tundra regions (e.g. Kuskokwim, Kol basins), the lack of significant elevation change across the landscape overestimates the extent of the flood plains.  To improve the floodplain delineation in these regions, a riparian vegetation classification is used to refine floodplain extent. (Fig. 2b). 

Basin Habitat Complexity Metrics
By combining information from DEM and remote sensing based water classification results, an accurate and reproducible method for characterizing basins across the north Pacific is now possible. We have implemented a preliminary set of metrics for several SaRON basins (Figure 3) in order to evaluate the biophysical significance of each variable and refine the top-down metrics relative to detailed biological surveys of salmonid habitat and population dynamics along regional productivity gradients. Our initial selection of variables for defining habitat complexity is based on a well developed body of literature (e.g. Leopold et al. 1964, Brown 2002, Whited et al. 2006), and the constraints of the regional datasets.  The results in Figure 3 identify marked differences in geomorphological patterns and associated habitat characteristics among the various basins that coincide with the regional biological productivity gradient defined from intensive field surveys. As expected, areas with a large amount of accessible floodplain area relative to basin size show the greatest productivity. For a given floodplain geomorphic domain, a moderate level of channel separations and returns (e.g. Taku) is indicative of a dynamic shifting habitat mosaic, relatively complex riparian vegetation and parafluvial and orthofluvial habitats, and greater biological productivity than similar sized floodplains with less complex structure. Greater levels of complexity (e.g. Skeena) are indicative of more frequent scouring and a greater rate of change in habitat dynamics, less developed riparian vegetation communities, reduced habitat gradients and corresponding lower productivity levels. The relative size and distribution of on-channel lakes is also a particularly relevant metric for some salmonid species (e.g. Sockeye), and less important for others. Continued development of these metrics and associated complexity analysis will include an assessment of the biological significance of habitat location and distribution within each basin, and an analysis of estuary structure.

Estimation of Juvenile Fish Populations
Juvenile fish production for SaRON field sites were estimated using electro-fish data coupled with habitat delineation results from relatively fine scale (1-4m resolution) Quickbird multi-spectral satellite imagery.  Mean juvenile fish densities were estimated for key habitat types (e.g. main channel - shallow shore, parafluvail springbrooks, and orthofluvial springbrooks) from electro-fishing surveys (Figure 4). These habitat types were classified from the satellite imagery using automated classification techniques for relative depth and velocity (Whited et al. 2003, Lorang et al. 2005) and delineation of para vs. orthofluvial springbrook habitats.  Habitat types were quantified for each SaRON flood plain and then juvenile fish production was estimated using the mean juvenile fish density per habitat type (Figures 4-5).

The estimates of juvenile fish production for the habitat types were then scaled up to the floodplain scale to determine the number of juvenile fish per square kilometer of floodplain (Figure 8).

Using these estimates, the potential number of fish per flood plain habitat segment will be estimated across all drainage basins. Quality flood plains will be defined by a number of metrics including degree of channel complexity, number of off channel water body types, stream order, and degree of human impacts.