Project Title: Mapping current conditions and modeling the dynamic responses of riparian vegetation and salmon habitat in Oregon. Proposed Project Period: Oct 1, 2007 to Sep 30, 2009 Principal Investigator: Dr. Steve Wondzell USDA Forest Service, PNW Research Station 3625 93rd. Ave., S.W. Olympia, WA 98512 Phone: 360-753-7691 Fax: 360-753-7737 Email: swondzell@fs.fed.us Co-Principal Investigator(s): Dr. Kelly Burnett USDA Forest Service, PNW Research Station 3200 Jefferson Way Corvallis, Oregon 97331 Phone: 541-750-7309 Email: kelly.burnett@oregonstate.edu Dr. Janet Ohmann USDA Forest Service, Pacific Northwest Research Station 3200 SW Jefferson Way Corvallis, Oregon 97331 Phone: 541-750-7487 Fax: 541-758-7760 Email: janet.ohmann@oregonstate.edu Dr. Warren Cohen UDSA Forest Service, PNW Research Station 3200 SW Jefferson Way Corvallis, Oregon 97331 Phone: 541-750-7322 Fax: 541-758-7760 Email: warren.cohen@oregonstate.edu Mr. Jimmy Kagan Oregon State University, Institute for Natural Resources 1322 SE Morrison St. Portland, Oregon 97214 Phone: 503-731-3070 ext. 111 Fax: 503-731-3070 ext. 118 Email: jimmy.kagan@oregonstate.edu Dr. Miles Hemstrom USDA Forest Service, PNW Research Station 620 SW Main St. Suite 400 Portland, OR 97205 USA Phone: 503-808-2006 Fax: 503-808-2020 Email: mhemstrom@fs.fed.us Dr. Peter Bisson USDA Forest Service, PNW Research Station 3625 93rd Ave., S.W. Olympia, WA 98512 USA Phone: 360-753-7671 Fax: 360-753-7737 Email: pbisson@fs.fed.us Proposal Summary: Key Words (1000 characters): Riparian zone; restoration planning; riparian and aquatic assessment; effectiveness monitoring; intensively monitored watershed; Nehalem River, Middle Fork John Day River; lidar; NAIP digital photography; Landsat; GIS; salmon; steelhead; bull trout; habitat quality; active restoration; passive restoration; vegetation composition and structure. Objectives (2000 characters): The proposed research integrates riparian zone mapping with dynamic models to project the response of riparian zones, stream channels and salmon habitat to natural disturbance and land-use activities. The overall objective of this work is to produce a decision support tool for habitat restoration planning that incorporates advanced remote-sensing technology and information about disturbance-recovery processes with existing knowledge of critical habitat needs for salmonids. The proposal has two components: 1) remote sensing and riparian mapping, 2) riparian and aquatic modeling. The objective of the mapping component is to explore different methods for mapping riparian and instream conditions using Landsat, lidar, and NAIP imagery and to use these methods to delineate, classify and map the attributes of riparian zones needed for riparian assessment and monitoring and to support the modeling component. The objective of the modeling component is to develop a Dynamic Riparian and Aquatic Integrated Network (DRAIN) planning model that includes background processes and land-use activities to simulate temporal dynamics of riparian vegetation, channel conditions, and habitat quality for salmonids. We will apply the remote sensing and mapping methods to two intensively monitored watersheds – Nehalem and Middle Fork John Day (MFJD) Rivers – and apply the DRAIN model to the to examine current conditions relative to the historic range of variability, examine potential of passive restoration to meet recovery goals, and examine the potential of active restoration to accelerate recovery. Methodology (2000 characters): Remote sensing imagery, including Landsat, lidar, and digital airphotos, will be combined with field data to characterize riparian vegetation. Spatial modeling techniques will be used to classify and map vegetation composition and structure for riparian characterization and for initializing the DRAIN model. A time-series of Landsat data will be used to characterize disturbance patterns over recent decades and their impact on riparian habitat. Riparian maps will be integrated with basin-wide maps of existing upland vegetation. A high-resolution DEM (<10 M) will be built from lidar imagery. We will extract watershed area, longitudinal gradient, and valley-floor width and use these to delineate and classify stream reaches into the potential channel morphologic groups for the DRAIN model. Finally, we will assess the accuracy of the mapped riparian attributes using a variety of established methods. We will compare the quality of maps developed with and without lidar data to determine the feasibility and cost for scaling up from watershed to the entire state. The mapping component of our research will be linked to the modeling component through the delineation, classification, and assessment of current condition of streams and riparian zones in each watershed. Development of the DRAIN planning model is based on a prototype aquatic and riparian model we developed using the Vegetation Development Dynamics Tool (VDDT). We will develop additional VDDT models for riparian vegetation and channel types missing from our suite of prototype models. We will also expand the pathways and transition probabilities still needed to characterize many historic and current land-use activities. We will use TELSA to develop a dynamic stream network model that is linked with spatially explicit upland models. We will modify and automate our habitat evaluation model and, using known habitat preferences of salmonids, project the effects of changes in riparian and channel condition on fish species. Rationale (2000 characters): Pacific salmon and steelhead have declined in abundance or been eliminated from large parts of their historical range, and many populations are now listed under the U.S. Endangered Species Act. Maintenance of existing high-quality habitat and restoration of degraded habitat have become cornerstones of many salmon recovery efforts. To support these efforts, there is a critical need for sound information on riparian areas – particularly over the broad spatial extents that are relevant to recovering salmonid populations. However, providing relatively detailed information on current conditions and likely future riparian conditions for large watersheds (e.g., 4th-field HUC) over large regions poses substantial logistical and technical challenges. Combining remote sensing and ground-based surveys offers a potential method for characterizing current conditions of riparian zones over large areas. However, much of the available remotely sensed imagery and DEMs are generally too coarse in spatial resolution to adequately characterize the detail of narrow riparian zones. A variety of fine-scale remotely sensed data are becoming widely available (e.g., lidar, NAIP, QuickBird) and have great potential for detailed landscape assessments. While combining newly available remotely-sensed and field-collected data offers a means of assessing the current condition of riparian areas at fine detail over large areas, these assessments only provide a static “snapshot” of a watershed at a single point in time. We need the ability to use this snapshot as a starting point for projecting changes in habitat resulting from processes that shape streams, riparian forests, and the upland systems to which they are connected. Thus, there is also critical need for an integrated stream network planning model that can support the design, implementation and monitoring of restoration projects in large watersheds. Expected Outcomes (2000 characters): The mapping methods and decision support model we develop will allow watershed councils and OWEB to assess current riparian conditions and evaluate how conditions may change in response to land-use activities and restoration efforts. These products will contribute to whole-watershed restoration planning, examining both passive and active restoration approaches to develop long-term restoration strategies for large, intensively monitored watersheds. Our expected outcomes are: 1) maps displaying geomorphic and vegetation attributes of the Nehalem and Middle Fork John Day watersheds with associated accuracy assessment, a time-series of disturbance events in the watersheds since 1972, and vegetation and land cover of the surrounding uplands; 2) methods for integrating both remotely-sensed and field data for mapping riparian, in-stream, and associated upland conditions, including recommendations for use of tested remote sensing products for riparian monitoring; 3) the Dynamic Riparian-Aquatic Integrated Network (DRAIN) planning model will be made publicly available along with a PNW General Technical Report fully documenting the model and describing its application for whole-watershed riparian assessment and restoration planning; and 4) the results of our final, validated analyses of habitat potential, current stream and riparian conditions and projected trends of tributary habitats of the Nehalem and Middle Fork John Day River Watersheds. We will establish a web site where the mapping tools and both the DRAIN model and model documentation can be freely downloaded by interested parties. We will archive collected field and remote-sensing data, either on our project’s web site or other relevant data archival site. Finally, the study’s findings will be reported in peer reviewed literature and at scientific and professional meetings. Results will also be presented to land managers, watershed councils, and others interested in stream restoration planning. 7.A. Introduction: The proposed research integrates riparian zone mapping with dynamic models to project the response of riparian zones, stream channels and salmon habitat to natural disturbance and land-use activities. The intent of this work is to produce a decision support tool for habitat restoration planning that incorporates advanced remote-sensing technology and information about disturbance-recovery processes with existing knowledge of critical habitat needs for salmonids. Pacific salmon and steelhead have declined in abundance or been eliminated from large parts of their historical range (Nehlsen et al., 1991), and many populations are now listed under the U.S. Endangered Species Act (USDA and USDI, 2000). Multiple factors have contributed to these declines, including the degradation of spawning and rearing habitat in tributary streams (Federal Caucus, 2000). Maintenance of existing high-quality habitat and restoration of degraded habitat have become cornerstones of many salmon recovery efforts. Science generally supports that riparian protection and restoration are effective ways to conserve stream ecosystems (NRC 2002). However, there is a critical need for sound information on riparian areas – particularly over the broad spatial extents that are relevant to recovering salmonid populations. Better understanding is necessary on the role of riparian areas in creating and maintaining salmon habitats and how these habitats change over time in response to natural and anthropogenic disturbance. Landscape-scale perspectives of historical, current, and potential future conditions of upland, riparian, and aquatic systems resulting from plant succession, natural disturbances, and land-use practices can help inform policy directions. Further, such information is essential to developing strategic restoration approaches for large regions (e.g., Oregon Plan for Salmon and Watersheds for coho salmon) that make the most of limited funds. However, providing relatively detailed information on current conditions and likely future riparian conditions for large watersheds (e.g., 4th-field HUC) over large regions poses substantial logistical and technical challenges. Combining remote sensing and ground-based surveys offers a potential method for characterizing current conditions of riparian zones over large areas (Gould, 2000). Such a baseline could be compared to available historical information to estimate the relative health of watersheds. Further, the development of 1 standardized methods based on remotely sensed data would allow trend assessments of habitat conditions in the future. Remote sensing is a rapidly growing discipline, and several newer approaches have not been fully exploited for analyses of riparian zones. Widely used Landsat data are generally too coarse to adequately characterize the detail of narrow riparian zones. A variety of fine-scale remotely sensed data are becoming more available (e.g., lidar, NAIP, QuickBird) and have greater potential for detailed landscape assessments. In particular, lidar (light detecting and ranging) technology, which can both create a high-resolution digital elevation model and landform map along with a three-dimensional picture of the vegetation, is becoming more available and affordable. A planned interagency purchase may make lidar available for most of western Oregon within a few years. In addition, statewide, high-resolution (0.5 meter pixel), geo-referenced, digital air photography was acquired in 2005 through the National Agricultural Imagery Program (NAIP). We propose to develop and test methods for integrating remote sensing, other spatial data, and field plots to characterize current riparian and in-stream conditions. As context for current conditions, we will combine the riparian data with information on current upland vegetation and historical patterns of disturbance and vegetation change. We will test the methods for two large, intensively monitored watersheds – the Nehalem and Middle Fork of the John Day (MFJD) Rivers – with the goal of developing rigorous and repeatable methods that should be scalable statewide to meet a variety of information needs. In particular, the methods will deliver maps showing the extent and condition of riparian vegetation, a high-resolution map of the valley floor and riparian-related landforms, and the range of vegetation structure and ecological diversity of the riparian areas within these basins. We also will test the capability of methods for mapping riparian wetlands and wetland conditions. These methods will be specifically directed toward supplying data for aquatic habitat assessments and for parameterizing the aquatic-riparian models we are developing. A secondary goal will be to create data directed at monitoring the effectiveness of restoration projects in these basins, as well as testing monitoring methods to evaluate changes resulting from restoration efforts. 2 The mapping and classification of riparian zones using a combination of remotely-sensed and fieldcollected data offers a means of assessing the current condition of riparian areas at fine detail over large areas. However, these assessments only provide a static “snapshot” of a watershed at a single point in time. We need the ability to use this snapshot as a starting point from which we can project changes in habitat resulting from processes that shape streams, riparian vegetation, and the upland systems to which they are connected. Therefore we propose to develop a Dynamic Riparian and Aquatic Integrated Network (DRAIN) planning model designed to support design, implementation and monitoring of restoration projects in large watersheds. We will build from a prototype state and transition aquatic and riparian model we developed using the Vegetation Development Dynamics Tool (VDDT; Beukema et al. 2003). VDDT models are now commonly used for upland forest and rangeland assessments (Hann et al., 1997; Hemstrom et al., 2001 & 2006). In our previous work, we developed a prototype framework extending VDDT to include aquatic and riparian ecosystems (further details can be found in Wondzell et al., 2006). We propose to further develop our prototype model into a functional planning tool. The DRAIN model will be designed to forecast network-scale changes resulting from plant succession, hydro-geomorphic processes, and natural disturbances. Simulations using a background natural disturbance regime can be used to “hind-cast” the historical range of variability and to forecast the outcomes of “passive restoration”, i.e., the strategy of allowing natural ecosystem processes to dictate the pace and trajectory of habitat recovery. The model will also accommodate land-use activities such as forest harvest, grazing, and stream, riparian, and upland restoration activities (e.g., large wood additions to streams, riparian fencing and fuels treatment projects). The model can thus be used to forecast the outcome of alternative management policies and to project the benefits of “active restoration” actions focused on key habitats, i.e., those activities designed to create missing habitat elements or to support future habitat-forming events. The DRAIN model will be designed to include upland components and will thereby include processes that control conditions of streams and watersheds, as a whole, in addition to the conditions of specific reaches or project areas. Critical watershed processes that will be included in our model are: 1) riparian 3 vegetation dynamics stemming from land-use activities, natural disturbance, and plant succession; 2) episodic disturbances (e.g., floods and wildfires) that shape stream channels and valley floors; and 3) linkages to landscape-scale factors (e.g., background upland processes and management policies). The assessment of current riparian conditions represented by mapping and classification of riparian zones and models that can project future conditions are essential tools for restoration planning, implementation, and monitoring. Our research and development of mapping tools will test the relative value of expensive, high-resolution, remotely sensed data (lidar and digital airphotos) versus more affordable but coarser-scale data (Landsat) that are already available for watershed assessment and can provide time-series information for recent decades. The mapping tools, used alone, will be configured to provide an accurate assessment of current conditions. Comparing current conditions with vegetation disturbance and dynamics over the past 30 years from Landsat data, and with model “hind-casts” of the historical range of variability in habitat from natural disturbance regimes, will support watershed-scale assessment of the extent of present-day habitat alteration and help set realistic restoration goals. Alternatively, current conditions can be used as a starting point to forecast likely outcomes of alternative land-use decisions, including restoration activities. For example, a watershed council could project likely trajectories of habitat conditions given “passive” restoration resulting from management policies as determined by the State Forest Practices Act or the Northwest Forest Plan and compare those with projections that employ “active” restoration projects. The DRAIN model can thus be used to examine both passive and active restoration approaches (alone or in combination) to develop a long-term restoration strategy for large, intensively monitored watersheds. 7.B. Objectives: 7.B.1. Objective 1 – Mapping: We will explore different methods to map the distribution and condition of riparian and upland vegetation. Vegetation and related geomorphic attributes will be used to define riparian and channel conditions for stream reaches throughout the stream network in two pilot basins. Methods will include integration of the two types of vegetation information historically collected in riparian areas: (1) plot data collected to classify riparian vegetation (Crowe et al 2004, McCain 2004); and 4 (2) riparian and stream habitat survey data from several agencies (ODFW, EPA, USFS, BLM) which provide data on channel and valley floor characteristics. Specific objectives are to: 1. Explore methods for mapping riparian and in-stream conditions using Landsat, lidar, and NAIP imagery. Evaluate the potential to characterize and map riparian and aquatic wetlands with these data. 2. Integrate detailed maps of riparian conditions (from #1, above) with existing maps of vegetation composition and structure for surrounding uplands, to provide additional basin-level information important to riparian habitat assessment. 3. Use time series of Landsat data (1972- present) to characterize recent disturbances and vegetation succession to inform riparian assessment for potential stream sedimentation and recruitment of large in-stream wood. 4. Develop map attributes specifically designed to support the DRAIN model. 7.B.2. Objective 2 – Modeling: The overall objective of the modeling component is to develop a Dynamic Riparian and Aquatic Integrated Network (DRAIN) planning model that includes background processes and land-use activities to simulate temporal dynamics of riparian vegetation, channel conditions, and habitat quality for salmonids. The proposed research builds on a foundation of existing state-andtransition models we have developed for the upper Grande Ronde River basin over the last 4 years (Wondzell et al., 2006; Hemstrom et al., 2006). The models are not yet linked into a spatially explicit, dynamic aquatic-riparian network model that can be used as a functional planning tool. We stress that these accomplishments serve only as a proof of concept and provide the building blocks for the work proposed here. Our work is focused on three specific objectives: 1. Adapt the existing aquatic and riparian state and transition models to another setting in eastern Oregon (Middle Fork John Day River) and to a setting in western Oregon (Nehalem River). 2. Develop a spatially explicit framework in which reach-scale models are linked to a dynamic stream network and incorporate existing upland models to link riparian zones with adjacent uplands. 5 3. Apply the DRAIN model to the Nehalem and MFJD Rivers to examine current conditions relative to the historical range of variability, examine potential of passive restoration to meet recovery goals, and examine the potential of active restoration to accelerate recovery. The DRAIN model will be designed for use by natural resources specialists and land-use planners to compare the effects of alternative management scenarios on riparian restoration. The modeling framework makes the models relatively easy to modify and use in a variety of landscape analyses. Further, the models are based on geomorphic processes and riparian vegetation dynamics that are well described for eco-regions across Oregon. Thus, the models will be broadly portable to land planning efforts throughout the state. Analysis of restoration outcomes will be supported by designing representative active restoration transitions into the model. These transition pathways and probabilities will be based on the watershedand reach-scale restoration priorities described in the Oregon Plan for Salmon and Watersheds and the John Day sub-basin plan (2005), and on specific restoration activities conducted within the Nehalem and MFJD intensively monitored watersheds. 7.C. Approach and Methods: 7.C.1. Mapping Methods: To characterize vegetation physiognomy, cover, and structure we will use a combination of remote sensing datasets: Landsat, lidar, and digital airphotos. Although each dataset has its own advantages for estimating and mapping vegetation characteristics, their synergy will provide more detailed and accurate measures. We also will test spatial modeling techniques such as classification trees and gradient imputation (Ohmann and Gregory 2002) for mapping classes of vegetation composition and structure (including successional status) that are suitable for general riparian characterization and for initializing the DRAIN model. We will evaluate and modify existing riparian vegetation classifications (National Vegetation Classification System, Plant Association Group, Ecological Systems, etc.) for our mapping efforts. The selected classification will be cross-walked to riparian communities already described for northwest Oregon (McCain 2004) and eastern Oregon (Crowe et al. 2004). 6 Target physiognomic types include trees (conifer vs. broadleaf) and shrubs, and to the extent possible a grass/forb/herbaceous complex. Percent cover will be characterized by physiognomic type. Structure is a more complex variable set, and will be quantified in a number of ways. For woody plants we will determine density, size, biomass, etc; for trees we will also calculate basal area. Landsat data have spectral properties that enable distinguishing between needleleaf and broadleaf vegetation, but do not do very well at distinguishing between shrubs and trees if they share broad physiognomic characteristics (Cohen et al. 2001). NAIP digital airphotos (http://165.221.201.14/NAIP.html), at 0.5 m resolution, can be used to assist in this, but lidar data are superior to NAIP in that they provide a direct measure of height. Similarly, percent vegetation cover can be estimated from Landsat (Cohen and Goward 2004), but not as well as from lidar or NAIP data. Significant challenges exist when using digitally processed airphoto data to derive cover information, but with use of modern morphological filtering tools such as eCognition this is becoming easier (Arroyo et al. 2006). Lidar data have significant advantages over Landsat and airphoto data for deriving cover and structure information (Lefsky et al. 2001), and will be the primary source of this information for our work. We will investigate several means of integrating the three primary datasets to take advantage of the unique and complementary contributions of each to prove detailed, current vegetation information. Changes in vegetation cover over time are informative of riparian condition, in terms of sedimentation, stream chemistry, and accumulation of large dead wood. We will use time series of Landsat data (1972- present) to define both disturbance (Cohen et al 2002, Healey et al. 2005) and succession (Schroeder et al in press). This will include translation of time series spectral data into vegetation cover trajectories and fitting of curves to these trajectories to automatically define disturbance data and severity (Kennedy et al. in press). From the lidar data we will derive a high-resolution DEM (<10 M) from which we can extract watershed area, longitudinal gradient, and valley-floor width needed to delineate and classify stream reaches into the potential channel morphologic groups for the habitat modeling. Other attributes we will attempt to extract from the DEM for use in the modeling component include: channel gradient, width, and 7 sinuosity; channel incision; frequency of pools, riffles, and off-channel habitat; riparian forest structure and canopy shade; channel-road crossings (e.g., culverts) and waterfalls that limit fish passage. We will use existing algorithms to extract these patterns from the digital elevation data (e.g., Miller 2003; Benda et al. 2006). We will also evaluate the potential to use lidar data and NAIP imagery to identify and quantify in-stream large wood, and to locate and map riparian-associated wetlands. Field data to be used as reference for the riparian mapping include plots established in riparian areas by the regional ecology program of the Pacific Northwest Region (McCain 2004, Crowe et al. 2004). Other field data sources (e.g., riparian and stream surveys of the USDA FS, EPA, INR, ODFW, and BLM) will be evaluated and used if adequate for our purposes. We will collect additional validation data from field plots we will measure as part of this study. We will assess the accuracy of the mapped riparian attributes using a variety of established methods. We will compare the quality of maps developed with and without lidar data as one means of determining feasibility and cost for scaling our methods up from intensively monitored watersheds to statewide. We will combine the detailed maps of riparian vegetation with existing maps of upland vegetation developed using gradient imputation (Ohmann and Gregory 2002) and classification and regression trees (CART) (www.fsl.orst.edu/lemma/gnnpac). These maps are now available for all of eastern Oregon, and western Oregon will be completed by late 2007. 7.C.2. Linking the Modeling and Mapping Components: The riparian mapping component of our research will be linked to the modeling component through the delineation, classification, and assessment of current condition of stream reaches in the stream network of each intensively monitored watershed. The stream network will be subdivided into reaches following the classification of Montgomery and Buffington (1997 & 1998) using watershed area, longitudinal gradient, and valley-floor width (Montgomery et al., 1999). The reaches will be classified into the following potential channelmorphologic type: 1) cascade, 2) step-pool, 3) plane-bed, and 4) pool-riffle (Montgomery et al. 1999). Each reach will be further classified into a discreet state based on riparian and channel-morphologic features described in the mapping component – namely active channel width and sinuosity, channel 8 incision, frequency of pools, riffles, and off-channel habitat, riparian forest structure and canopy shade, channel-road crossings (e.g., culverts), and waterfalls that limit fish passage, and if available, in-stream large wood. 7.C.3. Modeling Methods: We have developed prototype aquatic and riparian planning models using VDDT (Wondzell et al., 2006). These models describe changes in riparian vegetation and channel condition that occur through vegetative succession, natural disturbance, and a limited group of land-use activities. Our prototype models are not spatially explicit, they are designed to provide the building blocks for constructing a dynamic landscape and network model. These models remain under development. They require more complete treatment of the effects of land-use practices, development of a modeling framework that stitches individual reach-scale models into a dynamic stream network, and links to parallel landscape modeling efforts in Oregon using VDDT-based state and transition models such as the Interagency Mapping and Assessment Project (IMAP: www.reo.gov/ecoshare/mapping/index-issues.asp). Task 1 – complete additional reach-scale state-and-transition models following methods described in Wondzell et al., (2006). We will develop additional models for channel morphologic groups missing from our suite of prototype models. We will expand the pathways and associated transition probabilities still needed to characterize many historical and current land-use activities. Critical missing transitions include the effects of historical logging of riparian areas, splash dams and log drives, wood removal from river networks, prescribed fire and fuel treatments, logging and other silvicultural treatments, and active restoration practices. The final models will characterize disturbance regimes for 1) the historical range of variation; 2) recent land management, and 3) restoration, including effects for both passive restoration and active restoration practices. Finally, we also will develop new VDDT models, or adapt existing models, for riparian vegetation types missing from our prototype models, especially for Coast Range streams. Task 2 – develop a spatially explicit framework. The existing VDDT models simulate local-scale processes and larger-scale disturbances impacting a given stream reach. Because the models are not spatially explicit, hydro-geomorphic disturbances, such as floods, cannot propagate downstream nor can large-scale upland disturbances, such as fire, propagate into riparian areas. There are several modeling 9 approaches that would allow translation of the current VDDT model framework into a spatially explicit, dynamic stream network (e.g., discrete event simulation, Fonnesbeck, 2006). However, a major goal of the proposed work is to develop models that can be linked to existing, large-scale landscape analyses (USFS R6 Forest Plan Revisions, IMAP). These models are built on the VDDT framework and utilize the Tool for Exploratory Landscape Scenario Analysis (TELSA; Kurz et al. 2000) for spatially explicit simulations. New revisions to TELSA allow simulation of directional movements of disturbances across landscapes. Our work under this task will focus on developing a dynamic stream network model that is linked with spatially explicit upland models. Task 3 – develop stream habitat evaluation models. We developed a prototype expert systems model to evaluate channel conditions and habitat suitability for Chinook salmon (Oncorhynchus tshawytscha) and steelhead (O. mykiss) (Wondzell et al., 2006). Channel conditions were evaluated using 14 indicator variables commonly measured in aquatic assessment programs in the Pacific Northwest (MacDonald et al., 1991; Moore et al., 2002; Kershner et al., 2004; Reeves et al., 2004). These variables are related to habitat suitability for Chinook salmon and steelhead using a qualitative 4-factor scale based on professional expertise and available literature. Currently, the use of our habitat suitability models is limited because they are not directly linked with the VDDT models. Therefore, we will modify the existing modeling framework to automate habitat evaluations. We have made preliminary projections of habitat suitability with these models (Wondzell et al., 2006), however, the models are not yet validated. We will test model predictions with a retrospective examination of actual long-term changes in stream habitat condition using National Forest inventories, Oregon Department of Fish and Wildlife surveys, or tribal records. Task 4 – apply the DRAIN model to the Nehalem and MFJD Rivers. We will use the maps of current riparian and in-channel conditions derived from the mapping component of our research as a starting point for model simulations. We will use model analyses to compare and contrast future trends expected from: 1) continuing current land-use practices; 2) wide-scale passive restoration on public lands, and 3) combining the projected changes in land-use management called for in the Unified Aquatic Conservation 10 Strategy currently being developed for USFS and BLM lands with high priority active restoration treatments called for in restoration plans for the intensively monitored watersheds (i.e., Oregon Plan for Salmon and Watersheds and the John Day Sub-Basin Plan). The historical range of variability in the distribution of riparian and channel conditions will be projected from model runs using only the background natural disturbance regime. The historical range of variability will be used as the baseline for model comparisons. 7.C.4. Field Data Collection Methods: Models for both quantitative mapping of riparian conditions and for projecting aquatic-riparian conditions require field data for parameterization and validation. To the extent possible, we will use existing data. However, the distribution of available data within our intensively monitored watersheds may be limited. Our sampling strategy must balance the requirement to provide a statistically representative distribution of current channel and riparian conditions with the logistical limitations forced by the resources available and time needed to sample each reach. Therefore, our sampling will be focused on those channel morphologic groups that have the highest potential to provide spawning and rearing habitat for salmonids, and where existing data are most limited.. We will sample the channel and riparian conditions at each sample reach using sampling protocols widely used in stream and riparian assessments in the Pacific Northwest (MacDonald et al., 1991; Moore et al., 2002; Kershner et al., 2004; Reeves et al., 2004) to measure the variables used in our models. These sampling protocols have been extensively tested for application to status and trend monitoring (e.g., Archer et al., 2004). Our sampling protocol is specifically designed to measure composition and structure of vegetation for riparian mapping and to measure the indicator variables used in the DRAIN model. The data will be used to refine thresholds used in the GIS-based classification and reach delineation of potential channel morphologic groups. The data will also be summarized to characterize the current distribution of riparian and channel conditions and then compared to predictions from the DRAIN model to see if they support the discrete states and transformation pathways used in the VDDT models or if these models need to be modified or if new models need to be developed. The vegetation data will be used as 11 reference data for training the remotely sensed data, and in accuracy assessment of resulting vegetation maps. 7.D. Work Plan & Time Table: 7.D.1. Mapping – In the first year we will concentrate on acquiring existing plot and spatial data, acquiring new lidar data for the two basins, developing plot and spatial databases for analysis, and identifying additional field sampling needs (Table 1). Starting in the first quarter of 2008, we will begin developing the DEM, stream topology, and vegetation information from the lidar data. We will then explore methods for integrating plot and spatial data to characterize riparian and in-stream habitat. In the final phase of the study we will examine the history of disturbance and succession in the basins using Landsat data, and integrate riparian and upland (disturbance and GNN/ReGAP) map products to characterize conditions over entire basins. All phases of the work will involve close collaboration with the modeling component to assure data needs for project integration are met. 7.D.2. Modeling – In the first year of this study we will complete the development of the vegetation and hydro-geomorphic VDDT models and the spatially-explicit dynamic-network modeling framework. We will also acquire existing stream survey data and fish abundance data. Finally we will use data from the mapping component to design our summer field sampling campaign for the Nehalem and MFJDR watersheds. We will spend the summer field season collecting stream and riparian data (Table 1). During the following winter, we will use the DEM, stream topology, and vegetation information extracted from the lidar data to parameterize the DRAIN model and make initial model runs. Model projections will be tested against our collected field data, stream survey data, and fish abundance data to test the model projections. We will modify and recalibrate models as needed. Once models are completed, we will conduct a final analysis of the watersheds, projecting alternative future states expected from alternative management strategies. 12 Table 1. Schedule of proposed work. Benchmarks & Milestones Q 4 Q 1 2008 Q Q 2 3 Q 4 2009 Q Q Q 1 2 3 Mapping: Acquire existing remote sensing and other spatial data Acquire existing plot and survey data and develop database Fly lidar over Middle Fork John Day River Identify additional field sampling needs Develop DEM, stream, and vegetation layers from lidar Integrate spatial and plot data to generate rasters of vegetation attributes and classified riparian network Examine history of disturbance and succession (Landsat) Integrate riparian and upland maps (GNN and ReGAP) Assess accuracy and error structures of map products Modeling: Complete VDDT models (vegetation + hydro-geomorphic) Develop spatially-explicit dynamic modeling framework Acquire existing fish population monitoring data Automate stream-habitat evaluation modeling component Complete initial watershed analysis and locate field sites Collect validation field data Validate & finalize model analyses Generate a holistic, long-term restoration strategy for each basin Final report, documentation, manuscripts ► 7.E. Project Management, Collaboration and Matching Funds: 7.E.1. Matching Funds: Matching funds exceed 29% (see budget justification for details). 7.E.2. Working on private land: Not applicable. We will not be collecting field data on private lands. 7.E.3. Collaborations with others: Mapping - The mapping of riparian vegetation will be coordinated with other ongoing programs to map existing vegetation of Oregon and the Pacific Coast States. These regional mapping efforts span all land ownerships, at 30-m resolution using Landsat imagery, and do not target riparian areas. Ecological Systems of all land cover types are being mapped for the USGS Gap Analysis Program by the USDA Forest Service (USFS) Pacific Northwest Research Station (PNW), the Institute for Natural Resources at Oregon State University, and the Oregon Natural Heritage Information Center. The composition and structure of forest vegetation is being mapped using Gradient Nearest Neighbor imputation by PNW in 13 collaboration with the USFS Western Wildland Environmental Threats Assessment Center, the Interagency Mapping and Assessment Project (IMAP), the Oregon Department of Forestry, the USFS Region 6 (Pacific Northwest Region), Effectiveness Monitoring for the Northwest Forest Plan, and PNW's Forest Inventory and Analysis Program. The proposed riparian mapping and modeling will benefit directly from this additional information on the uplands. The riparian project will utilize extensive plot and spatial data developed for the other projects, in addition to final map products. Modeling – The models developed here will be designed for use by natural resources specialists and land-use planners to compare the effects of alternative management scenarios on riparian restoration. The DRAIN models will be designed to link with existing, broad-scale landscape modeling efforts that also use VDDT models such as the USFS Region 6 Forest Plan Revisions and the multi-agency IMAP project. Development of the DRAIN model will allow riparian areas to be included in these ongoing efforts. The models we are developing will contribute substantially to planning and assessment efforts for salmon recovery. OWEB, Watershed Councils, and private partners; other State and Federal Agencies; the Tribes; and many NGO’s have invested heavily in salmon recovery efforts. Recovery efforts are typically focused on active restoration treatments. But to be successful, these efforts must be integrated with improved tributary habitat and habitat-forming processes throughout the whole watershed. Thus there is critical need for a large-scale planning and assessment model that can evaluate alternative management scenarios being considered by landowners or land management agencies and link these to on-going and planned restoration activities within intensively monitored watersheds. Our models will provide useful feedbacks between planned and ongoing monitoring and evaluation efforts within the intensively monitored watersheds. Trends in stream, riparian, and watershed conditions measured in these monitoring programs can be compared to trajectories projected in our simulation models. These comparisons will help determine if revised management direction and restoration efforts are resulting in expected habitat recovery and thereby help inform adaptive management decisions. 14 7.G. Expected Outcomes and Deliverables: The mapping methods and decision support model we develop will allow watershed councils and OWEB to assess current riparian conditions and evaluate how conditions may change in response to landuse activities and restoration efforts. These products will contribute to whole-watershed restoration planning, examining both passive and active restoration approaches to develop long-term restoration strategies for large, intensively monitored watersheds. Mapping 1: We will provide maps with accuracy assessment for use in assessing and monitoring riparian and upland conditions that affect riparian and in-stream habitat. The maps will display: geomorphic and vegetation attributes of the two study basins; time series of disturbance events in the basins since 1972; vegetation and land cover of the surrounding uplands. Mapping 2: We will provide methods for integrating remotely-sensed and field data for mapping riparian, in-stream, and associated upland conditions, including recommendations for use of tested remote sensing products for riparian monitoring. We will also provide information about alternative imagery types, evaluating the quality of information provided vs. costs of acquisition and processing. Modeling 1: We will make the Dynamic Riparian-Aquatic Integrated Network (DRAIN) planning model publicly available. We will publish a PNW General Technical Report documenting the model and describing its application for whole-watershed riparian assessment and restoration planning. Modeling 2: We will report the results of our final, validated analyses of habitat potential, current stream and riparian conditions and projected trends of tributary habitats of the Nehalem and Middle Fork John Day River Watersheds. Overall 1: We will establish a web site where the mapping tools and both the DRAIN model and model documentation can be freely downloaded by interested parties. We will archive collected field and remote sensing data either on our project’s web site or other relevant data archival sites (e.g., H. J. Andrews LTER data base or the Puget Sound Lidar Consortium web sites). Overall 2: Study findings will be reported in peer reviewed literature and at scientific and professional meetings. Results will also be presented to land managers, watershed councils, and others. 15 7.H. References: Archer, E. K., Roper, B. B., Henderson, R. C., Bouwes, N., Mellison, S. C., and Kershner, J. L. 2004. Testing common stream sampling methods for broad-scale, long-term monitoring. USDA Forest Service, Rocky Mountain Research Station, General Technical Report RMRS-GTR-122 http://www.fs.fed.us/rm/pubs/rmrs_gtr122.pdf Arroyo, L.A., S. Healey, W.B. Cohen, D. Cocero, J.A. Manzanera. 2006. Using object-oriented classification and high-resolution imagery to map fuel types in the Mediterranean region, Journal of Geophysical Research. Benda, L. E., D. J. Miller, K. Andras, P. Bigelow, G. Reeves, and D. Michael. In press. NetMap: A new tool in support of watershed science and resource management. Forest Science. Beukema, S. J., Kurz, W. A., Pinkham, C. B., Milosheva, K., and Frid, L. 2003. Vegetation Dynamics Development Tool, User's Guide, Version 4.4c. Vancouver, BC, Canada: ESSA Technologies Ltd. 239 pp. Cohen , W. B. and S. N. Goward. 2004. "Landsat's role in ecological applications of remote sensing." BioScience 54: 535-545. Cohen, W. B., T. K. Maiersperger, T. A. Spies and D. R. Oetter. 2001. "Modeling forest cover attributes as continuous variables in a regional context with Thematic Mapper data." International Journal of Remote Sensing 22: 2279-2310. Cohen, W. B., T. A. Spies, R. J. Alig, D. R. Oetter, T. K. Maiersperger and M. Fiorella. 2002. "Characterizing 23 years (1972-1995) of stand replacement disturbance in western Oregon forests with Landsat imagery." Ecosystems 5: 122-137. Crowe, E.A., B.L. Kovalchik, and M. Kerr. 2004. Riparian and Wetland Vegetation of Central and Eastern Oregon. Oregon Natural Heritage Program, Portland, OR. 483 pp. Federal Caucus. 2000. Vols. 1-2. Conservation of Columbia basin fish, final basinwide salmon recovery strategy. Portland, OR: Federal Caucus/Bonneville Power Administration. http://www.salmonrecovery.gov/Final_Strategy_Vol_1.pdf, http://www.salmonrecovery.gov/Final_Strategy_Vol_2.pdf. (June 9, 2003). Fonnesbeck, C. J. 2006. Spatial modeling of riparian state dynamics in eastern Oregon, USA by using discrete event simulation. Landscape and Urban Planning, DOI:10.1016/j.landurbplan.2006.10.007 (Published on-line 11 December, 2006). Gould, W. 2000. Remote sensing of vegetation, plant species richness, and regional biodiversity hotspots. Ecological Applications. 10:1861-1870. Healey, S.P., Y. Zhiqiang, W.B. Cohen, and J. Pierce. 2006. Application of two regression-based methods to estimate the effects of partial harvest on forest structure using Landsat data, Remote Sensing of Environment 101:115-126. Hemstrom, M. A., Merzenich, J., Reger, A., and Wales, B. C. 2006. Integrated analysis of landscape management scenarios using state and transition models in the upper Grande Ronde River Subbasin, 16 Oregon, USA. Landscape and Urban Planning DOI:10.1016/j.landurbplan.2006.10.004 (Published on-line 28 November, 2006). Hemstrom, M., Ager, A. A., Vavra, M., Wales, B. C., and Wisdom, M. J. 2004. Chapter 2: A state and transition approach for integrating landscape models. Pgs. 17-32. in J. L. Hayes, A. A. Ager, and R. J. Barbour, Technical Editors. Methods for integrating modeling of landscape change: Interior Northwest Landscape Analysis System. PNW-GTR-610. USDA Forest Service, Pacific Northwest Research Station, General Technical Report 610. Herger, L.G. and G. Hayslip. 2000. Ecological Condition of Streams in the Coast Range Ecoregion of Oregon and Washington. EPA-910-R-00-002. U.S. Environmental Protection Agency, Region 10, Seattle, Washington. John Day Sub-Basin Revised Draft Plan. 2005. Report to the Northwest Power and Conservation Council. Prepared by the Columbia-Blue Mountain Resource Conservation and Development Area., and available at: http://www.nwcouncil.org/fw/subbasinplanning/johnday/plan/ Kershner, J. L., Archer, E. K., Coles-Ritchie, M., Cowley, E. R., Henderson, R. C., Kratz, K., Quimby, C. M., Turner, D. L., Ulmer, L. C., and Vinson, M. R. 2004. Guide to effective monitoring of aquatic and riparian resources. USDA Forest Service, Rocky Mountain Research Station, General Technical Report RMRS-GTR-121. Kurz, W. A., Beukema, S. J., Klenner, W., Greenough, J. A., Robinson, D. C. E., Sharpe, A. D., and Webb, T. M. 2000. TELSA: the tool for exploratory landscape scenario analysis. Computers and Electronics in Agriculture 27:227-242. Lefsky, M. A., W. B. Cohen and T. A. Spies. 2001. "An evaluation of alternative remote sensing products for forest inventory, monitoring, and mapping in Douglas-fir forests of western Oregon." Canadian Journal of Forest Research 31: 78-87. MacDonald, L. H., Smart, A. W., and Wissmar, R. C. 1991. Monitoring guidelines to evaluate effects of forestry activities on streams in the Pacific Northwest and Alaska. US Environmental Protection Agency, Water Division, EPA/910.9-91-001. 166 p. McCain, C. 2004. Riparian plant communities of northwest Oregon: streamside plant communities. USDA Forest Service, Pacific Northwest Region, Technical Paper R6-NR-ECOL-TP-10-04. Miller, D.J. 2003. Programs for DEM analysis. In Landscape dynamics and forest management. Gen. Tech. Rep. RMRS-GTR-101CD. USDA Forest Service, Rocky Mountain Research Station, Fort Collins, CO. CD-ROM. (http://www.fsl.orst.edu/clams/prj_wtr_millerprg.html. Last Accessed on October 28, 2006) Montgomery, D. R. and Buffington, J. M. 1998. Chapter 2: Channel processes, classification, and response. Pages 13-42. In: R. J. Naiman and R. E. Bilby (eds.) River Ecology and Management Lessons from the Pacific Coastal Ecoregion. Springer-Verlag. New York, NY. Montgomery, D. R. and Buffington, J. M. 1997. Channel-reach morphology in mountain drainage basins. Geological Society of America Bulletin 109:596-611. 17 Montgomery, D. R., Beamer, E. M., Pess, G. R., and Quinn, T. P. 1999. Channel type and salmonid spawning distribution and abundance. Canadian Journal of Fisheries and Aquatic Sciences. 56:377387 Moore, K. M. S., Jones, K. K., and Dambacher, J. M. 2002. Methods for stream habitat surveys, version 12.1. Oregon Department of Fish and Wildlife, Aquatic Inventories Project, Natural Production Program, Corvallis, OR. 59 p. National Research Council (NRC). 2002. Riparian areas: functions and strategies for management. National Academy Press, Washington, DC. 428p. Nehlsen, W., Williams, J. E., and Lichatowich, J. A. 1991. Pacific salmon at the crossroads: Stocks at risk from California, Oregon, Idaho and Washington. Fisheries 16:4-21. Ohmann, J.L., and M.J. Gregory. 2002. Predictive mapping of forest composition and structure with direct gradient analysis and nearest neighbor imputation in coastal Oregon, USA. Canadian Journal of Forest Research 32:725-741. Reeves, G. H., Hohler, D. B., Larsen, D. P., Busch, D. E., Kratz, K., Reynolds, K., Stein, K. F., Atzet, T., Hays, P., and Tehan, M. 2004. Effectiveness monitoring for the aquatic and riparian component of the Northwest Forest Plan: conceptual framework and options. Gen. Tech. Rep. PNW-GTR-577. Portland, OR: U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station. 71 p. Schroeder, T.A. W.B. Cohen, Z. Yang. In press. Patterns of forest regrowth following clearcutting in western Oregon as determined from multi-temporal Landsat data, Forest Ecology and Management. U.S. Department of Agriculture, Forest Service; U.S. Department of the Interior, Bureau of Land Management. 2000. Interior Columbia basin supplemental draft environmental impact statement. The Interior Columbia Basin Ecosystem Management Project. Washington, DC: U.S. Government Printing Office. [Irregular pagination]. Wondzell, S.M., Hemstrom, M. A., and Bisson, P. A. 2006. Simulating riparian vegetation and aquatic habitat dynamics in response to natural and anthropogenic disturbance regimes in the Upper Grande Ronde River, Oregon, USA, Landscape and Urban Planning, DOI:10.1016/j.landurbplan.2006.10.012 (Published on-line 6 December, 2006). 18 Grant Administration: The overall grant budget is split into two separate contracts. While this is a tightly integrated joint project, the mapping component will be centered in the Corvallis Forestry Science Laboratory’s – Laboratory for Application of Remote Sensing in Ecology – under the direction of Dr. Warren Cohen; the modeling component will be centered at the Olympia Forestry Sciences Laboratory’s – Riparian Ecology and EcoHydrology Group – under the direction of Dr. Steve Wondzell. Logistical considerations and the high degree of independence between the two groups in the USFS Pacific Northwest Research Station’s organizational structure will create accounting problems making the grant difficult to administer under a single contract. Therefore, we request that the overall grant budget be broken into two separate contracts. The Statement of Work, shown below, reflects the joint accomplishments of both groups and contracts. Objectives: The proposed research integrates riparian zone mapping with dynamic models to project the response of riparian zones, stream channels and salmon habitat to natural disturbance and land-use activities. The overall objective of this work is to produce a decision support tool for habitat restoration planning that incorporates advanced remote-sensing technology and information about disturbance-recovery processes with existing knowledge of critical habitat needs for salmonids. The proposal has two components: 1) remote sensing and riparian mapping, 2) riparian and aquatic modeling. The objective of the mapping component is to explore different methods for mapping riparian and instream conditions using Landsat, lidar, and NAIP imagery and to use these methods to delineate, classify and map the attributes of riparian zones needed for riparian assessment and monitoring and to support the modeling component. The objective of the modeling component is to develop a Dynamic Riparian and Aquatic Integrated Network (DRAIN) planning model that includes background processes and land-use activities to simulate temporal dynamics of riparian vegetation, channel conditions, and habitat quality for salmonids. We will apply the remote sensing and mapping methods to two intensively monitored watersheds – Nehalem and Middle Fork John Day (MFJD) Rivers – and apply the DRAIN model to the to examine current conditions relative to the historic range of variability, examine potential of passive restoration to meet recovery goals, and examine the potential of active restoration to accelerate recovery. Methods: Remote sensing imagery, including Landsat, lidar, and digital airphotos, will be combined with field data to characterize riparian vegetation. Spatial modeling techniques will be used to classify and map vegetation composition and structure for riparian characterization and for initializing the DRAIN model. Riparian maps will be integrated with basin-wide maps of existing upland vegetation. A high-resolution DEM (<10 M) will be built from lidar imagery. We will extract watershed area, longitudinal gradient, and valley-floor width and use these to delineate and classify stream reaches into the potential channel morphologic groups for the DRAIN model. Finally, we will assess the accuracy of the mapped riparian attributes using a variety of established methods. We will compare the quality of maps developed with and without lidar data to determine the feasibility and cost for scaling up from watershed to the entire state. The mapping component of our research will be linked to the modeling component through the delineation, classification, and assessment of current condition of streams and riparian zones in each watershed. Development of the DRAIN planning model is based on a prototype aquatic and riparian model we developed using the Vegetation Development Dynamics Tool (VDDT). We will develop additional VDDT models for riparian vegetation and channel types missing from our suite of prototype models. We will also expand the pathways and transition probabilities still needed to characterize many historic and current land-use activities. We will use TELSA to develop a dynamic stream network model that is linked with spatially explicit upland models. We will modify and automate our habitat evaluation model and, using known habitat preferences of salmonids, project the effects of changes in riparian and channel condition on fish species. Rationale: Pacific salmon and steelhead have declined in abundance or been eliminated from large parts of their historical range, and many populations are now listed under the U.S. Endangered Species Act. Maintenance of existing high-quality habitat and restoration of degraded habitat have become cornerstones of many salmon recovery efforts. To support these efforts, there is a critical need for sound information on riparian areas – particularly over the broad spatial extents that are relevant to recovering salmonid populations. However, providing relatively detailed information on current conditions and likely future riparian conditions for large watersheds (e.g., 4th-field HUC) over large regions poses substantial logistical and technical challenges. Combining remote sensing and ground-based surveys offers a potential method for characterizing current conditions of riparian zones over large areas. However, much of the available remotely sensed imagery and DEMs are generally too coarse in spatial resolution to adequately characterize the detail of narrow riparian zones. A variety of fine-scale remotely sensed data are becoming widely available (e.g., lidar, NAIP, QuickBird) and have great potential for detailed landscape assessments. While combining newly available remotely-sensed and field-collected data offers a means of assessing the current condition of riparian areas at fine detail over large areas, these assessments only provide a static “snapshot” of a watershed at a single point in time. We need the ability to use this snapshot as a starting point for projecting changes in habitat resulting from processes that shape streams, riparian forests, and the upland systems to which they are connected. Thus, there is also critical need for an integrated stream network planning model that can support the design, implementation and monitoring of restoration projects in large watersheds. Expected Outcomes: The mapping methods and decision support model we develop will allow watershed councils and OWEB to assess current riparian conditions and evaluate how conditions may change in response to land-use activities and restoration efforts. These products will contribute to whole-watershed restoration planning, examining both passive and active restoration approaches to develop long-term restoration strategies for large, intensively monitored watersheds. Our expected outcomes are: 1) maps displaying geomorphic and vegetation attributes of the Nehalem and Middle Fork John Day watersheds with associated accuracy assessment, a time-series of disturbance events in the watersheds since 1972, and vegetation and land cover of the surrounding uplands; 2) methods for integrating both remotely-sensed and field data for mapping riparian, in-stream, and associated upland conditions, including recommendations for use of tested remote sensing products for riparian monitoring; 3) the Dynamic Riparian-Aquatic Integrated Network (DRAIN) planning model will be made publicly available; and 4) the results of our final, validated analyses of habitat potential, current stream and riparian conditions and projected trends of tributary habitats of the Nehalem and Middle Fork John Day River Watersheds. We will establish a web site where the mapping tools and both the DRAIN model and model documentation can be freely downloaded by interested parties. We will archive collected field and remote-sensing data, either on our project’s web site or other relevant data archival site. Finally, the study’s findings will be reported in peer reviewed literature and at scientific and professional meetings. Results will also be presented to land managers, watershed councils, and others interested in stream restoration planning.