FLATHEAD RIVER INSTREAM FLOW INVESTIGATION

Successful management of a river system influenced by a dam requires resource managers have access to spatial information in order to assess impacts of different management scenarios. This management tool is being created for the Flathead River located in northwestern Montana. 3DI and Miller Ecological Consultants are compiling a spatial database which includes detailing historical changes in the location of the river channel, current land use/land cover adjacent to the channel, mapping channel morphometry and 3D visualization of the river channel. These layers will then be used to create a habitat suitability matrix and model for identified fish species.

1. Introduction

This paper provides background on the environmental and regulatory conditions precipitating this project, describes the Instream Flow Incremental Methodology (IFIM) and the goals of integration of this methodology with GIS, summarizes the results of the first phase of work and describes the goals for the future phases of work.

This project is funded by the U.S. Department of Energy, Bonneville Power Administration and is being conducted in cooperation with Montana State Fish, Wildlife and Parks Department, Kalispell office.

 

 

 

 

2. Background

Excerpted from the Montana Fish, Wildlife and Parks contract proposal 1999.

Construction of Hungry Horse Dam on the Flathead River (completed in 1952), caused many physical and biological changes in the Flathead River downstream (Appert and Graham 1982; Fraley et al. 1986; Fraley and Decker-Hess 1987). The dam created a complete barrier to migrating adfluvial fish species from Flathead Lake that once spawned above the dam in the South Fork Flathead River. The barrier created by the dam is a mixed blessing because it now protects a rare native species assemblage from invasion by introduced species (e.g. lake trout, eastern brook trout, northern pike and non-native forms of rainbow trout) downstream. Hypolimnetic releases from the dam artificially cooled the river from 1952 through 1996 when a selective withdrawal structure was installed on the dam. The device allows dam operators to control the water temperature in the tailwater (Marotz et al. 1994; Christenson et al. 1996).

Now that the thermal pollution from Hungry Horse Dam can be mitigated, a primary manageable threat to watershed health is dam operation (Hall et al. 1989). Power and flood control operations have essentially reversed the natural hydrograph by storing the spring melt in the reservoir, and then releasing water for power production during the cold months when natural flows are normally low. Short term flow fluctuations caused by power operations create an extensive, low productivity varial zone, greater substrate imbeddedness and species shifts in the aquatic insect community which has become less diverse and less productive (Perry 1984; Hauer et al. 1994). River flows and flow fluctuations also cause important changes in habitat availability and fish movements (Cushman 1985; Dalbey et al. 1997). A combination of man-caused factors resulted in the decline in native gamefish species (mountain whitefish, westslope cutthroat and bull trout) and a significant increase in abundance of non-game native species, the Columbia River chub or peamouth, northern squawfish and introduced rainbow trout and northern pike.

Pursuant to measure 10.3A.18 of the Northwest Power Planning Council’s Fish and Wildlife Program (FWP) (NPPC 1995), this project will use a modified form of the Instream Flow Incremental Methodology (IFIM) to examine the mechanisms by which dam operation effects the riverine biotic community and their environment, and propose operational guidelines to mitigate negative effects (Fraley et al. 1989; Montana Fish Wildlife and Parks (MFWP) and Confederated Salish Kootenai Tribe (CSKT) 1991, 1993).

An annual flow regime with tolerable flow fluctuations is needed to maximize the effectiveness of the Hungry Horse Dam’s selective withdrawal, temperature control structure and to balance riverine productivity with hydropower production and flood control. Measure 10.3A.18 of the FWP calls for BPA’s consultation with MFWP and CSKT when conflict occurs between reservoir and river requirements. An understanding of the tradeoffs between biological productivity in the reservoir and river is needed to make knowledgeable recommendations for a balanced operation. Our goal is to restore normative conditions in the Flathead Watershed (ISAB 1997).

Since 1995, recovery actions for the endangered Snake River salmon (NMFS 1995) have influenced the timing of water released from Hungry Horse Dam. Specifically, summer releases of reservoir storage, to augment downstream flows, have caused unnatural flow fluctuations in the Flathead River during the productive summer months. Unless we understand the effects of these releases, recovery actions may counter or reverse mitigation efforts to balance the needs of resident and anadromous fish (ISAB 1997b).

This project provides the physical framework for assessing physical and biological effects of various river operations on the Flathead system. It will become a component of the Hungry Horse Mitigation Program addressing operational mitigation (Integrated Rule Curve refinement and assessment: measure 10.3A of the FWP). Results of this project will expand the utility of the existing reservoir model HRMOD (Marotz et al. 1996) by improving the riverine component and refine the Integrated Rule Curves developed for Hungry Horse Dam. This modeling strategy will help federal dam operators and fisheries managers' balance dam operations for the greatest benefit, by balancing fisheries concerns with power production and flood control.

The ability to assess tradeoffs between reservoir and river operations, both locally and systemwide, is especially important now that several Columbia River fish species have been petitioned and/or proposed for listing under ESA. Also, previous investments in hydropower mitigation should be protected with special consideration when changes in system operation are implemented. Changes in dam operation for recovery actions in the lower Columbia have been shown to impact resident fish in the headwaters (ISAB 1997b) and must be balanced to benefit all native fish species. To do this, decision-makers must have tools to assess tradeoffs and make wise choices.

 

3. The Instream Flow Integrated Methodology (IFIM)

IFIM has been designed for river system management by providing an organizational framework for evaluating and formulating alternative water management options. It has been built on the philosophical foundation of hydrological analyses to understand the limits of water supply (Stalnaker et al. 1995). The IFIM methodology is comprised of five sequential phases: problem identification, study planning, study implementation, alternatives analysis and problem resolution.

3.1 Development of the Methodology

Instream flow methods have been developed predominately by biologists and hydrologists working for agencies having regulatory responsibility related to water development and management (Stalnaker and Arnette 1976). Water management problem solving has evolved from setting fixed minimum flows with no specific aquatic habitat benefit to using incremental methods in which aquatic habitats are quantified as a function of discharge.

Initial efforts to provide minimal protection for fisheries grew out of the concern resource agencies had for the loss of habitats resulting from reservoir and water development during the mid-twentieth century. Methods were developed that combined hydrologic analysis, observations of habitat quality and expertise in riverine fish ecology. The results of these early methods, however, provided less than optimal conditions for fisheries. Minimum flow levels were set for stream reaches below which no water could be drawn for consumptive use. These flows provided minimum protection and did not account for different "life-stage" needs of fish.

Following enactment of the National Environmental Policy Act (NEPA) of 1970, attention was shifted from minimum flows to the evaluation of alternative designs and operations of federally funded water projects. Methods capable of quantifying the effect of incremental changes in streamflow to evaluate a series of possible alternative development schemes were needed (Stalnaker 1993).

The need for quantified data led to research addressing life-stage-specific habitat versus flow relationships for selected species and the establishment of the key chemical and physical attributes that control fish well being. The set of attributes that consistently were shown to contribute significantly to the variation of fish population and production included water velocity, minimal water depths, instream objects such as cover, bottom substrate materials (with particular emphasis on the amount of fines in the interstitial spaces within coarse bed elements), water temperature, dissolved oxygen, total alkalinity, turbidity, and light penetration through the water column (Gosse and Helm 1981; Shirvell and Dungey 1983).

Review and assessment of federal water projects progressed into licensing application review of hundreds of small hydropower projects in the late 1970’s through early 1980’s. During this transition period the Instream Flow Incremental Methodology (IFIM) was developed under the guidance of the U.S. Fish and Wildlife Service (Trihey and Stalnaker 1985). Incremental methods became solidified as the tool of choice over the next 10 years as hundreds of federal reservoirs applied for relicensing. Multiple demands were being placed on water resources that needed to be balanced in the application review. Hydropower companies wanted to change reservoir management to enhance revenue, resource agencies looked at relicensing as a way to restore long-impacted riverine aquatic resources and recreational interests saw opportunities for improve river recreational uses.

 

3.2 Integration with GIS

Making decisions about river system management has grown to involve multiple parties with competing interests. Multi-disciplinary teams participate in IFIM projects to give voice to each interest. GIS provides the spatial context for building and analyzing the relationships between the various demands. Initial attempts at quantifying river management impacts to aquatic habitats were one-dimensional, that is they looked only at a few river cross-sections to determine habitat conditions and extrapolated this value to larger river segments. Current methods employ 2D approaches that can more accurately depict and quantify values at scales applicable for individual fish species and that can integrate data at various scales.

GIS provides the consistent base of spatial and attribute information from which alternative management scenarios can be compared. Each resource demand, (recreation, habitat, power generation, irrigation) can be compared against the other demands to strike a balance, if possible, where all demands are satisfied. Tradeoffs between alternatives can be quantified and documented for review by decision-makers.

GIS also provides the context within which the results of a selected alternative's implementation can be monitored over time, especially with respect to physical impacts to the river system. Field data collected for changes in habitat type, velocity, substrate can be compared to species mix, life-stage and total numbers of fish.

 

4. Methodology and Results of Year 1 Work

This section describes the land cover classification and change detection analysis work completed for Bonneville Power Administration (BPA) in conjunction with Montana Fish Wildlife and Parks (MFW&P) as part of the Flathead River Instream Flow Investigation Project.

The land cover classification was derived from a spatially enhanced, multispectral IRS-1C image acquired in 1998. Historic river channels were delineated from data acquired at three different time periods spanning 20 years to help identify and monitor changes in the river channel. Changes occurring between 1990 and 1998 were mapped in greater detail identifying the type (from – to) of land use / land cover change.

The Flathead River stretches approximately 42 miles through the Flathead Valley in northwest Montana. The valley consists of fertile agriculture land and timber and is bound by the Swan Mountain Range to the East. The study area encompasses the Flathead River from the confluence of the South Fork of the Flathead River, near Columbia Falls, south to the Flathead Lake inlet. It was determined at the project initiation meeting that land cover would be mapped within one-half mile of the furthest outlying river channel.

It was determined during the project initiation meeting between MFW&P and 3Di that the following land cover classes would be mapped for the study area. Classes mapped at a larger scale were facilitated through the collection of data during two days of fieldwork. Channel characteristics such as erosive banks and stabilized banks, were captured in vector format because of their linear nature. The vectors were coded by height as shown below in Table 1. See Appendix A for class definitions.

Imagery from the Indian Remote Sensing Satellite (IRS-1C) was acquired to produce the 1998 land cover classification (Tables 2 & 3). Since the IRS satellite has multiple sensors on board, it was possible to obtain both a 20-meter mutlispectral image (LISS III channel) and a 5-meter panchromatic (PAN channel) image that were acquired at the same time. The two images were then merged together to create a single image that combined the spectral resolution from the LISS III product with the higher spatial resolution of the PAN product. This spatially enhanced multispectral image allowed 3Di to map the land cover and changes at a larger scale.

Aerial photography was acquired from the USDA Forest Service at a scale of 1:15,840. These natural color photographs were flown during 1997 and 1998. The aerial photographs provided ground reference information for the land cover classification and also served as a basemap for classifying channel characteristics during fieldwork. Twenty-five 10x10 photographs were required for full coverage of the study area.

Digital Orthophoto Quarter Quads (DOQQs) were acquired and used as the data source for interpreting the 1990 river channel and detailing the change that occurred between 1990 and 1998. These panchromatic aerial photos are scanned at 1-meter spatial resolution and orthogonally rectified to remove any displacement caused by relief or camera distortions. The aerial photos were flown during the month of July 1990 and 1991. A list of the DOQQs that cover the study area can be found in Appendix B.

Digital Raster Graphics (DRGs) were used as the data source for interpreting the earliest river channel. A DRG is a standard USGS 7.5 minute topographic map scanned at a resolution of 2.8 meters and output to a georeferenced TIF file. The majority of the DRGs used for the study area were originally produced during the 1950s and photo updated in 1978. A list of the DRGs used can be found in Appendix C.

Mapping the extent of the river from aerial photographs and satellite imagery involved interpreting the boundary between water and land within the river channel. Since the Hungry Horse Dam impacts the flow of the Flathead River, the dynamic flow levels of the river influence the channel boundaries. Daily flow levels for the dates of acquisition were obtained from the USGS gauging station located at Columbia Falls, which lies on the upper portion of the study area (Table 4).

Note: The USGS DRGs cannot be linked to a specific date, therefore no daily flow measurement was available. It is estimated that the flow level is less than that of the highest flow of 27,000 cfs.

Rectification of the satellite imagery is required so that it will geographically register with a specified map coordinate system and other GIS data layers. The output land cover data will have the same geometric characteristics as the rectified satellite imagery, so rectification of the imagery is an important step. The process involves collecting ground control points from a data source with a known map coordinate system and selecting corresponding points on the satellite imagery. Over twenty points are usually collected to rectify an entire satellite scene. The geometric relationship among the points is used to calculate the root mean square (RMS) error, which is a measure of the accuracy of the rectification. A RMS error of about one half of a pixel (0.5) is considered excellent and an error of less than 1.0 is usually acceptable.

DOQQs were used to collect ground control points for rectification. No irregularities were encountered during the rectification process and a RMS error of less than 0.5 was attained for both of the satellite scenes.

Fieldwork was required to collect data that was not discernable on the satellite imagery and also to provide additional ground truth for areas adjacent to the river channel for the 1998 classification. A piloted boat was provided by MFW&P for two days to allow two image analysts from 3Di to map channel features and adjacent land uses and land cover. Natural color aerial photography printed at 1:15,840 scale was used as a basemap for mapping features such as eroded banks, stabilized banks and willows. These data were then digitized to create a digital, georeferenced data set that could be used to enhance the land cover classification derived from the IRS image. Table 5 shows the data created from the fieldwork completed in July 1999.

The enhanced IRS data was processed using an ISODATA clustering routine, which groups the image into spectral clusters. The number of spectral clusters output by the ISODATA routine is specified by the image analyst, and depends on the spectral variability of the data. For the initial clustering process, 125 clusters were specified for the study area. Once the 125 clusters were analyzed, those clusters that represented true classes were kept and those that were considered confusion clusters were run through the routine again to further refine the clustering. Once all of the clusters were assigned to a designated class, manual editing of the classification took place to correct any misclassifications of spectrally similar features. Table 6 shows the acreage for each land cover category in the study area.

The urban and rural developed lands were classified separately from the rural areas to help improve the accuracy of the classification. Urban structures such as concrete and rooftops are often spectrally similar to dry, bare soil and exposed rock. To avoid confusion between these cover types, urban areas were manually delineated on the IRS data and extracted from the imagery. An unsupervised classification was then performed on these areas. One hundred spectral clusters were specified for the urban classification regions. Clusters that represented urban areas were saved and coded appropriately. Clusters that did not represent urban areas were coded as class zero, so that categories from the rural classification would inhabit these areas when the urban and rural classifications were combined.

The land cover classification was generalized to remove "salt and pepper" from the land cover data. In most original classification products, very small pixel clumps, which fall along category boundaries, are erroneously classified. In this project, the single pixels and small clumps were dropped out using a "clump and sieve" routine and filled in by using a majority filter algorithm. Clumps less than four pixels in size were eliminated and replaced with the surrounding majority class.

Simple change was defined as a change from water to another land cover class or any land cover class to water. Simple change in the river channel was identified for three time periods: 1978-1990, 1990-1998 and 1978-1998. The river channels were digitized from the DRGs to create the 1978 channel, the DOQQs to create the 1990 channel and extracted from the IRS-1C classification to create the 1998 channel. An overlay technique was used to compare the river channels and identify areas of change and the general type of change. Table 7 shows amount of change identified.

Areas identified as change during the 1990-1998 time period were processed at a larger scale and in greater detail to produce the Detailed Change Analysis layer. The DOQQ's, aerial photographs, ground truth data and the 1998 IRS land cover classification were used to create this layer which identifies the "from and to" classes of change.

A spatial model was developed to further classify the type of change identified in the results of the simple change analysis. Areas that had changed from water in 1990 to a class other than water in 1998 were coded using the values of the1998 land cover classification as the "to" class. Since there was not any land cover established for the 1990 data beyond the water class, areas that had changed "from" a land cover class other than water had to be manually interpreted from the 1990 DOQQ data. The results produced a detailed "from and to" change analysis layer mapped at approximately 1:12,000 scale. Table 8 shows the "from and to" classes and amounts identified for each class. See Appendix D for change class definitions.

4.12. Classification Accuracy Assessment

After completion of the 1998 IRS classification, accuracy assessment was performed. This task was accomplished by analyzing 100 randomly selected assessment points within the classification. At least 10 accuracy assessment sites for each class were obtained. The number of accuracy assessment sites assigned to each information class was roughly proportional to the abundance of that category in the classification.

Accuracy assessment sites were coded based on the land cover type present in the aerial photography, where coverage was available, and the IRS image in other areas. All of the accuracy assessment sites were compared with the classification to determine an overall accuracy for a classification unit. In addition, accuracy figures were calculated for each class. Appendix E shows the accuracy assessment tables.

The land cover classification derived from the enhanced IRS-1C imagery proved to be an accurate representation of the river channel and the adjacent lands providing an overall classification accuracy of 91%.

Due to spectral similarities between Agriculture lands and Grasslands, many of the Agriculture lands were delineated manually. Grasslands were recoded to agriculture if the land use pattern indicated agricultural practices were being applied. Since these classes were very similar in the spectral response, it was not unexpected to see some Grassland and Agriculture confusion in the accuracy assessment table.

Water and Forest classes did extremely well in the accuracy table. Deep, clear water and coniferous forests are spectrally similar and are often confused with each other. Some confusion of this nature had to be edited out, but the amount was minimal. Some Forest may be confused with water where coniferous trees are growing next to the river. In this case it was sometimes challenging to separate the boundary between trees and water using the IRS imagery.

The Barren class showed slight confusion with several of the other classes. This confusion seems to represent the natural intermixing of the Barren class with Willow, Agriculture and Grassland. The willows were found mainly on rocky substrates within the channel and intermixed with the Barren class, which represented the rocks. Confusion with the Agriculture class and Grassland class appeared to be a result of the spectral reflectance of the soil being stronger than the spectral reflectance of the sparse vegetation.

The Simple Change data created accurate representations of the dynamic river channel over the span of twenty years. Differences in water levels may have contributed to false change identification. The 1990 data had a flow level that was 7.5 times the flow during the acquisition of the 1998 IRS imagery. Although certain types of change may have been influenced by the difference in water levels, true changes in the channel’s extent were detected.

The Detail Change went a step further to identify exactly what types of changes had occurred during the 1990-1998 time period. This process required manual interpretation of the historic land cover present in the 1990 DOQQ data where change was identified. A portion of the areas identified in the Simple Change layer were not identified further if there was not substantial evidence in the data that true change had occurred. Some of these uncharacterized areas were influenced by tree canopies and shadows that extended into the river channel due to the off-nadir look angle inherent in the DOQQ data.

The types of detailed changes identified can be divided into two broader categories for comparison purposes, change "from Water" in 1990 and change "to Water" in 1998. Approximately 30,000 acres of land cover fell into "from Water" while only 3,300 acres were classified as "to Water." The large amount of land cover identified as "from Water" is due to the increased flow of water in the 1990 data. Those areas classified as "to Water" are more indicative of true change representing a loss of landmass due to erosion.

Two Pentium PCs were used for image processing. Both systems were equipped with dual Pentium Pro 200 processors, 128 megabytes of RAM and 11 gigabyte hard drives. The system used for vector processing was a Pentium Pro 200 with 132 megabytes of RAM and six gigabytes of disk space. All systems were running Windows NT, version 4.0.

ERDAS Imagine software, version 8.3.1 was used for image processing and ARC/INFO version 7.2.1 was used for vector manipulation. Both software packages were developed for the NT operating system. Microsoft Word was used for word processing and Excel was utilized for mathematical operations.

This portion of the Flathead River Instream Flow Investigation Project has provided many insights into the dynamic characteristics of this river system while providing a solid baseline for future studies and data development. The next step is to transform this two-dimensional data set into three dimensions. Three reaches of the Flathead River, each representing unique channel characteristics, have been identified. Detailed channel topography has been generated for each of these reaches and will be used for three-dimensional visualization. A spatial model will also be developed to identify species habitat suitability based on river characteristics such as habitat, depth and velocity.

At the request of MFW&P, 3Di has proposed to acquire additional data to further develop this database both geographically and historically. Aerial photography acquired circa 1940 can be obtained from the National Archives. This data would provide insight to pre-dam conditions of the Flathead River while extending the timeframe of the dataset to approximately 60 years. Also, similar 1990 DOQQ and 1998 IRS-1C data are available for both the North and Middle Forks of the Flathead River. This data would provide the means to extend the channel mapping and change detection coverages into these two regions.

8.1.1. Overview

After the development of the GIS base mapping information for the project during the 1^st phase of work the focus of the GIS turns to: 1) building upon the Flathead River GIS base data, 2) developing a river habitat suitability matrix and model for identified fish species and, 3) visualizing the river system in 3D.

Over the 3-year time span of the project, team members will be conducting fieldwork and analysis of data that will be captured within the GIS. As layers are developed that describe the dynamics and key components of the river system a habitat suitability matrix and model will be developed that will be used to identify and determine the amount of habitat for fish species of concern at various flow levels. In order to improve understanding of the dynamics of this system the GIS and analysis layers will be displayed as 3D models and representations. BPA decision-makers will access Flathead River project information through a project specific and intuitive computer interface. This interface will also allow BPA’s technical staff to model and map habitats of species of concern

The technology for 3D display and visualization of spatial data exists now. Environmental Systems Research Institute (Esri) the developers of ARC/INFO, ArcView and related GIS products and Earth Resources Data Analysis System (ERDAS) the makers of ERDAS Imagine have developed 3D visualization and modeling tools that integrate GIS data and are among a growing number of 3D visualization software companies. The GIS software industry is advancing rapidly with the development of "open" systems and object-oriented technologies that offer GIS to a larger base of users without requiring a high level of technical know-how. 3DI expects that 3D visualization software products will advance over the next few years offering greater capabilities and ease of use. 3D visualization will be provided for the GIS data, however, we have not specifically determined how the data will be portrayed to allow for advances in the technology to be incorporated in the definition of these project deliverables as the project advances.

This task involves developing a river channel surface for the Flathead River. Using the hydro-acoustic and other data collected by Miller Ecological Consultants (MEC) and AYRES Consultants, 3DI will use ARC/INFO to create a triangulated irregular network (TIN) and then a continuous surface that will be imported into ArcView 3D Analyst. Using 3D Analyst 3DI will create sample views of each of the segments of the river where data was collected and plot these for BPA review. BPA will review the samples and determine which ones will be used for future modeling and plotting.

Using federal agency data, data collected from other funded projects and data generated from the MEC team, 3DI will convert the thermal and velocity information into spatial data that can be used to generate a continuous grid within the river. For example, thermal modeling done by MEC will be converted to x, y points with an associated temperature code. These points will be used to generate a grid of temperature values that can be temperature contoured for the entire river study area.

3DI will coordinate with BPA and representatives to develop a habitat suitability matrix for species management in the Flathead River. Each input thematic layer, i.e. thermal gradient, substrate, depth, protective cover and velocity will be given a value relative to its value as habitat for a particular fish species and life stage. Additionally, within each thematic layer are sub-divisions. For example, the Flathead River will have a range of velocities or substrates in the channel depending on Hungry Horse Dam releases. These intra-layer categories will be given a habitat weighting value also. 3DI will combine these layers in an overlay process to create a composite that contains all the values summed for the area of interest. This analysis can be output as a 2D map, draped over the channel morphometry to create a 3D model for on-screen visualization or hard copy plotted.

3DI will meet with BPA and the project team members to assign species and life stage-specific values to each of the thematic layers. The product of the meeting will be a matrix showing fish species and life stage along the Y-axis (row) and the thematic layers (substrate, velocity, etc.) as the X-axis (columns) with relative values assigned to the intersecting fields. The data in this chart will be incorporated into the GIS Habitat Suitability Model (HSM) for mapping species habitats.

In the final year of the study 3DI will work to refine the HSM as more information is added to the GIS through further field data collection, input from related studies funded by BPA or analysis by project team members. The 3D Visualization methods will be finalized and transferred to BPA.

As part of this task, 3DI will prepare a report documenting the GIS, analysis, modeling and results. This will include sources and methods used in the development of the data (metadata), processes used in preparing the analysis and modeling, code developed for modeling purposes and related user manuals for use by BPA.

The report will be presented to BPA in Kalispell upon completion of the project. During this time 3DI will also provide training to those BPA personnel in charge of taking over management of the GIS and providing on-going support to the analysis and modeling effort.

The following category definitions will be utilized.

1.0 Developed Lands - Lands primarily used for residential and commercial purposes that contain structures and a transportation network. Areas must have a density of at least one structure per half acre and comprise a total of at least 15 acres.

2.0 Cultivated Lands - Lands under cultivation during the year of imagery acquisition. Includes well-maintained forage crops, but not seeded and hayed grasses. Does not include lands were woody vegetation is cultivated, such as nurseries or orchards.

3.0 Grasslands - Lands dominated by grasses and forbs. These areas may be hayed or grazed and contain widely scattered shrubs. Examples: Pasture, old fields, improved pasture.

4.0 Forest - An area where the predominant vegetation is comprised of trees with a single stem, generally over six feet in height, with a crown closure of at least 75%.

5.0 Bare Lands - Areas primarily composed of rock, gravel, sand or exposed soil, with little or no vegetation. Examples: Rock outcrops, open pit mines, alluvial sand or rock bars, construction sites, roads.

6.0 Water and submerged land - Areas dominated by open water. Some aquatic vegetation may be present.

7.0 Willow – Riparian areas where Willows are present but not necessarily visible in the multi-spectral imagery. This land cover was mapped during fieldwork and manually interpreted into the classification.

8.0 Steep, Highly Erosive Banks - This class was mapped during fieldwork and further classified into three sub-classes based on estimated heights of the banks at the time of the data acquisition.

8.0.A - Erosive Bank less than 5 feet

8.0.B - Erosive Bank greater than 5 feet and less than 10 feet

8.0.C - Erosive Bank greater than 10 feet

9.0 Stabilized Banks – Areas where the riverbank has some form of material placed for stabilization purposes. This was classified as to the type of material used for stabilization.

9.1 Automobiles – areas where vintage automobiles have been placed along the river bank

9.2 Rip Rap – bank stabilization consisting of rocky material with angular features

9.3 Cobble – areas consisting of river rock averaging 10 inches in diameter

SOMERS NE

SOMERS NW

SOMERS SE

BIGFORK NW

BIGFORK SW

KALISPELL NE

KALISPELL SE

CRESTON NW

CRESTON SE

CRESTON SW

COLUMBIA FALLS SOUTH NE

COLUMBIA FALLS SOUTH NW

COLUMBIA FALLS SOUTH SW

DORIS MOUNTAIN

DORIS MOUNTAIN NW

COLUMBIA FALLS NORTH SE

COLUMBIA FALLS NORTH SW

HUNGRY HORSE SE

HUNGRY HORSE SW

SOMERS

BIGFORK

KALISPELL

CRESTON

COLUMBIA FALLS SOUTH

COLUMBIA FALLS NORTH

HUNGRY HORSE

Appendix D. Flathead River Channel – Change Class Definitions.

Appert, S. and P. Graham. 1982. The impact of Hungry Horse Dam on the aquatic invertebrates of the Flathead River. Final report of the Montana Dept. of Fish, Wildlife and Parks, Kalispell, Montana, to the U.S. Bureau of Reclamation. 90pp.

Christenson, D. J., R.L. Sund and B.L. Marotz. 1996. Hungry Horse Dam’s successful selective withdrawal system. Hydro Review / May 1996:10-15.

ERDAS. 1995. Field Guide. Third Edition. Atlanta, Georgia.

Fraley, J., S. McMullin, and P. Graham. 1986. Effects of hydroelectric operations on the kokanee population in the Flathead River system, Montana. North Am. Journal of Fish. Management 6:560:568.

Fraley, J. and J. Decker-Hess. 1987. Effects of stream and lake regulation on reproductive success of kokanee salmon in the Flathead system, Montana. Regulated Rivers, Vol. I. Pp257-265.

Fraley, J., B. Marotz, J. Decker-Hess, W. Beattie, and R. Zubik. 1989. Mitigation, compensation, and future protection for fish populations by hydropower development in the upper Columbia System, Montana, USA Regulated Rivers; Research & Management 3:3-18.

Gosse, J.C., and W.T. Helm. 1981. A method for measuring microhabitat components for lotic fishes and its application with regard to brown trout. Pages138-141 in N.B. Armantrout, editor. Acquisitions and utilization of aquatic habitat inventory information. American Fisheries Society, Bethesda, Md.

Hall, C., J. Jourdonnais and J. Stanford. 1989. Assessing the impacts of stream regulation in the Flathead River Basin, Montana, USA I. Simulation modeling of system water balance. Regulated Rivers 3:61:77.

Hauer, F.R., J.T. Gangemi and J.A. Stanford. 1994. Long-term influence of Hungry Horse Dam operation on the ecology of macrozoobenthos of the Flathead River. Prepared for Montana Fish, Wildlife and Parks, Special Projects Bureau, Kalispell, Montana.

ISAB. 1997. The Normative River. Independent Scientific Advisory Board report to the Northwest Power Planning Council and National Marine Fisheries Service. Portland, OR.

ISAB. 1997b. Ecological impacts of the flow provisions of the Biological Opinion of endangered Snake River salmon on resident fishes in the Hungry Horse, and Libby systems in Montana, Idaho, and British Columbia. Independent Scientific Advisory Board. Report 97-3 for the Northwest Power Planning Council and National Marine Fisheries Service. Portland, OR.

Marotz, B.L., C.L. Althen, And D. Gustafson. 1994. Hungry Horse Mitigation: aquatic modeling of the selective withdrawal system – Hungry Horse Dam, Montana. Montana Fish, Wildlife, and Parks. Prepared for Bonneville Power Administration. 36 pp.

Marotz, B.L., D. Gustafson, C.L. Althen, and W. Lonon. 1996. Model development to establish integrated operational rule curves for Hungry Horse and Libby Reservoirs – Montana. Montana Department of Fish, Wildlife and Parks. Prepared for Bonneville Power Administration. 114pp.

Montana Department of Fish, Wildlife, and Parks and Confederated Salish and Kootenai Tribes. 1991 Fisheries mitigation plan for losses attributable to the construction and operation of Hungry Horse Dam. Montana Department of Fish, Wildlife, and Parks and Confederated Salish and Kootenai Tribe, Kalispell, and Pablo, Montana. 71pp.

Perry, S.A. 1984. Comparative ecology of benthic communities in natural and regulated areas of the Flathead and Kootenai Rivers, Montana. North Texas State University. Denton, Texas.

Lillisand, T.M. and R.W. Kiefer. 1994. Remote Sensing and Image Interpretation. Third Edition, John Wiley & Sons.

Lunetta, R.S. and C.D. Elvidge. 1998. Remote Sensing Change Detection – Environmental Monitoring Methods and Applications. Ann Arbor Press.

Shirvell, C.S., and R.J. Dungey. 1983. Microhabitat chosen by brown trout for feeding and spawning in rivers. Transactions of the American Fisheries Society 112:355- 367

Stalnaker, C.B. 1993. Fish habitat models in environmental assessments. Pagers 140-162 in S.G. Hildebrand and J.B. Cannon, editors. Environmental Analysis: The NEPA-Experience. CRC Press, Boca Raton, Fla.

Stalnaker, C.B., and J.L. Arnette, editors. 1976. Methodologies for determination of stream resource flow requirements: An assessment. U.S. Fish and Wildlife Service FWS/OBS – 76/03. Washington, D.C.

Trihey, E.W., and C.B. Stalnaker. 1985. Evolution and application of instream flow methodologies to small hydropower development: An overview of the issues. Pages 176-183 in F.W. Olson, R.G. White, and R.H. Hamre, editors. Proceedings of the symposium on small hydropower and fisheries. The American Fisheries Society, Denver Colo.

E-mail: dgeise@3dillc.com>