Home » Projects » Aquatic/Riparian Stream Network Modeling

Modeling riparian vegetation, channel condition, and
aquatic habitat dynamics in response to alternative land-use practices.

MFJD

Middle Fork John Day River, OR

Integrating riparian zone mapping with state-and-transition
models to project the response of riparian zones, stream
channels and salmon habitat to plant succession, natural
disturbance and land-use activities.

Collaborators:
Modeling: Remote Sensing & Mapping:
Steve Wondzell Warren Cohen
Agnieszka Przeszlowska Dirk Pflugmacher
Miles Hemstrom Robert Kennedy
Pete Bisson Janet Ohmann
Kelly Burnett Jimmy Kagan





Research Description:


We are developing a decision support tool for habitat restoration planning that incorporates advanced remote-sensing technology and information about disturbance processes with existing knowledge of critical habitat needs for salmonids. We build state and transition models to simulate temporal dynamics of riparian vegetation, channel conditions, and salmonid habitat quality in response to plant succession, natural disturbance, and land-use activities. We are testing the models in two watersheds – the Wilson River in Oregon’s Coast Range and the upper Middle Fork John Day River in the Oregon Blue Mountains – to examine current conditions relative to the historic range of variability, assess the potential of passive restoration to meet recovery goals, and evaluate the potential of active restoration to accelerate recovery.

» Project One-pager (60.5 KB)
» Non-technical overview of landscape modeling applications to stream and riparian systems (789 KB)

Study Watershed:


Wilson: The Wilson River originates in the mountains of the Northern Oregon Coast Range and drains to Tillamook Bay. This coastal river is known for its runs of steelhead and Chinook salmon. The watershed is characterized by steep mountainous terrain, narrow valleys, dense Pacific Coast rainforest, and areas allocated to timber-production. Much of the watershed was burned in the 1933-1951 Tillamook Burn and was reforested primarily with Douglas fir.

MFJD: The headwaters of the Middle Fork John Day River are in the Blue Mountains of northeastern Oregon. The upper MFJD provides important habitat for Chinook salmon, steelhead, and bull trout. Dry ponderosa pine and mixed conifer forests dominate south-facing slopes while moist coniferous forests predominate on north-facing slopes. This watershed has a long history of logging, mining, and livestock grazing.

The Wilson River in Oregon’s Coast Range

The Wilson River in Oregon’s Coast Range

The upper Middle Fork John Day River in the Oregon Blue Mountains

The upper Middle Fork John Day River in the Oregon Blue Mountains

Methods Overview:


The 5-m DEM overlain with classified stream reaches and their the valley floors

Figure 2. A portion of the 5-m DEM overlain with classified stream reaches and their the valley floors

The classified stream network in the upper Middle Fork John Day River watershed

Figure1. Classified stream network in the upper Middle Fork John Day River watershed

Stream network classification and delineation:
The stream network and valley floors were delineated with Netstream and ArcGIS tools. Stream reaches were classified into potential channel morphologies based on channel gradient and reach drainage area derived from Netstream, following the Montgomery and Buffington (1997) classification. From this initial classification, a stratified random sample of ~30 stream reaches per study watershed was selected for field validation of reach morphologies and valley floor delineation. These data were used to refine channel morphology and valley floor thresholds which were then applied in a final stream network classification (Fig. 1) and riparian zones delineation (Fig. 2). The final classification correctly classified 68% of reach morphologies in the upper Middle Fork John Day River watershed and 86% of reaches in the Wilson River.


Lidar-derived Canopy Height Map for a stream reach on the MFJD

Figure 3

Initial conditions for riparian vegetation:
Spatial modeling techniques and high density lidar point data (all returns above bare earth) were used to classify and map current riparian vegetation composition and structure within each riparian polygon in the stream network. Vegetation data collected in the field were used to assess the accuracy of the riparian vegetation attributes mapped from lidar (Fig. 3). Prediction accuracy was better than 66% for all attributes except riparian shrubs. The lidar data proved poor at distinguishing between young trees and tall shrubs and short statured shrubs (height < 0.5 m) could not reliably be distinguished from the bare earth DEM. Overall, mapped vegetation provided a reasonable fit to current conditions and was used to set initial conditions for model simulations.

Figure 3. Lidar-derived canopy height map for a stream reach on the MFJD. Initial Netstream-derived riparian polygons are shown in bold straight lines, field delineations of the valley floor and vegetation patches are shown with curved lines.



Wet-Meadow simplified (cartoon)

Figure 4

Aquatic-riparian state and transition models:
We intersected the classified stream network map with a potential vegetation map of each watershed to identify all possible combinations of channel and vegetation types (strata). We built separate state and transition models for each strata using the Vegetation Development Dynamics Tool (VDDT). These aquatic-riparian VDDT models were based on prototype aquatic models developed for the upper Grande Ronde River (Fig. 4), OR and upland forest VDDT models developed for eastern and northwest Oregon. We modified these models to include riparian shrub and hardwood state classes, effects of herbivory (grazing and browsing), hydrogeomporhpic processes, and riparian restoration practices. We also expanded the pathways and transition probabilities to characterize the historic and current land-use activities present in each watershed. The completed models simulate all possible states of a stream reach, from highly disturbed states resulting from disturbances such as logging, grazing, landslides or wildfire to relatively pristine states where major disturbances have not occurred.

Figure 4. A highly simplified wet-meadow aquatic-riparian VDDT model illustrating some of the dominant transition pathways. Intermediate state classes and most transition pathways are not shown in order to keep the figure readable.



Channel conditions and aquatic habitat quality:
We qualitatively ranked channel morphologic conditions on a 4-factor scale for each state class in the models. Morphologic condition variables included: shade, erosion, undercut banks, large wood, pools, large pools, pool structure, off-channel habitat, width-depth ratio, and riparian shrub abundance. These variables were then used in an expert systems model to rank the habitat quality for key life-history stages of spring Chinook salmon and steelhead in the upper Middle Fork John Day River and coho salmon and steelhead in the Wilson River.

Model Application:


The models are being applied to two intensively monitored watersheds – the Wilson River in the Oregon Coast Range and the upper Middle Fork John Day River in eastern Oregon’s Blue Mountains to examine: 1) current conditions relative to the historic range of variability, 2) likely trajectories of aquatic and riparian habitats given current and expected land-use practices, 3) the potential of passive restoration to meet recovery goals, and 4) the potential of active restoration to accelerate recovery.

The models can be run without anthropogenic disturbances to recreate historical conditions (pre-Euro-American settlement) as shown in the examples below (Fig. 5-8). Alternatively, the models can be initialized to current riparian and channel conditions in the watersheds and then run forward to project likely outcomes of alternative management scenarios. Realistic management scenarios were developed based on current policy and land management issues faced by managers within each watershed. The effects of policy decisions can be reflected in the models by either changing the rates of selected disturbances (e.g., decreasing grazing intensity), or by either adding or removing selected disturbance pathways from the models (e.g., removing all forest harvest treatments from riparian areas or adding fuels treatments). Consequently, decision makers could use these models to evaluate changes in habitat conditions under alternative management scenarios at sub-basin to regional scales.

Riparian Forest Types

Figure 5. Example model output showing the change in
vegetation structure and composition over time under three
management scenarios, starting with historical conditions
prior to Euro-American settlement in the upper MFJD

Nonforest Vegetation Types

Figure 6. Detail of the changes in composition and
structure of the non-forest riparian vegetation from the
example shown above in the upper MFJD.

Channel Attributes

Figure 7. Example of changes in two channel morphologic attributes under two different restoration scenarios. Time periods and model runs match the examples for riparian vegetation condition shown above in the upper MFJD.

Spring Chinook Summer Rearing Habitat Quality

Figure 8. Example of the changes in the amount of
summer rearing habitat for spring Chinook salmon in
each habitat condition class in the upper MFJD.

Discussion:


The simulated changes in both riparian vegetation and channel morphology result in large simulated changes in habitat quality within the upper MFJD stream network. It is important to note that, while some improvement in habitat quality is apparent relatively quickly, even after 50 years with complete protection from most anthropogenic disturbances, the condition of habitat within the watershed is far from fully recovered.

The model results suggest that both riparian vegetation and channel morphology can be highly dynamic for many attributes. A notable exception is the time needed to grow large trees as well as the recruitment of large wood from stream-adjacent riparian forests. In general, the rapid dynamics simulated by the models agree with observations made within the watersheds. For example, growth of riparian hardwoods has been very rapid in the Camp Creek (upper MFJD) livestock grazing and deer and elk browsing exclosure (Fig. 9). Similarly, a study of changes in channel conditions in grazing exclosures located throughout the western United States documented relatively rapid changes in channel conditions through time (Fig. 10).

Camp Creek Exclosure

Figure 9. Matched photographs from the interior of the Camp Creek grazing/browsing exclosure located within the upper MFJD study site. The first photo was taken in 1977 (photo from the Umatilla National Forest), at the time the exclosure was established. The second photo was taken 32 years later (photo by Rick R. Stiner, 2009). The photos document rapid changes in both vegetation and channel conditions.

Pools Models vs Data

Figure 10. Comparison of projected changes in pool abundance predicted by the models under two different restoration scenarios with measured changes in pool abundance in grazing exclosures (data from McDowell and Magilligan 1997).

Data and Downloads:


    Upper Middle Fork John Day

  • 5M DEM
    • 5m DEM (Raster Data, 102 MB)
      • 5-m resolution Digital Elevation Model derived from lidar data acquired in August 2008.
  • Stream Network and Riparian Zones
    • Stream Network (Vector Data, 391 KB)
      • Stream reaches with drainage areas greater than 1.5 km2 classified by channel morphology types.
    • Riparian Valley Floors & Non-Forest Zones
      • (Vector Data, 624 KB)
        Total valley floors for each reach within the stream network. These zones are also boundaries for classifying current riparian non-forest vegetation from lidar.
    • Riparian Forest Zones (Vector Data, 835 KB)
      • Forest Zones used to classify current riparian forest vegetation from lidar.
  • Potential Vegetation Types
    • Watershed Potential Vegetation Types (Vector Data, 4 MB)
      • Potential Vegetation Types of riparian and upland ecosystems in the Upper Middle Fork John Day watershed.
    • Riparian Potential Vegetation Types (Vector Data, 871 KB)
      • Potential Vegetation Types of riparian zones only. These polygons are also mapping units for aggregating current non-forest and forest vegetation attributes into states used in state and transition models.
  • Current Vegetation
    • Current Riparian Vegetation (Vector Data, 871 KB)
      • Current (2008) riparian vegetation classified into vegetation states used in state and transition models. Vegetation attributes include barren, herbaceous, shrub, and tree cover, tree size classes, and forest strata.
  • Intrinsic Potential & Initial Conditions
    • IP and Initial Conditions (Vector Data)
      • Intrinsic Potential for Spring Chinook and Steelhead calculated for each reach, and used to identify reaches to be modeled and determine initial conditions for Aquatic-Riparian models runs.
  • Aquatic-Riparian Models v. 2011_02_25
    • Aquatic-Riparian Models (Access .mdb, 1.76 MB)
      • Vegetation Dynamics Development Tool (VDDT) database for running aquatic-riparian state and transition models. Database includes all Potential Vegetation Type and channel morphology combinations for the Upper Middle Fork John Day River.
    • Model Documentation (PDFs .pdf, 768 KB)
      • Summaries of each model’s structure, disturbance dynamics, vegetation and fish habitat attributes.
    • VDDT (Software)
      • Vegetation Dynamics Development Tool can be downloaded from ESSA Technologies.
    • PATH (Software)
      • Path Landscape Model tool can be downloaded from Apex Resource Management Solutions, Ltd.
  • Aquatic-Riparian Models v. 2011_03_29
    • Aquatic-Riparian Models (Zip, 89 MB)
      • Most current version of the state and transition models (both PATH & VDDT). Includes all revisions and edits made since 2011_02_25. Also includes model packages used for analysis of the Upper Middle Fork John Day River watershed.

    Wilson

  • 5M DEM
    • 5m DEM (Raster Data, 102 MB)
      • 5-m resolution Digital Elevation Model derived from lidar data acquired in Spring 2007.
  • Stream Network and Riparian Zones
    • Stream Network (Vector Data, 391 KB)
      • Stream reaches with drainage areas greater than 1.5 km2 classified by channel morphology types.
    • Riparian Valley Floors & Non-Forest Zones
      • (Vector Data, 624 KB)
        Total valley floors for each reach within the stream network. These zones are also boundaries for classifying current riparian non-forest vegetation from lidar.
    • Riparian Forest Zones (Vector Data, 835 KB)
      • Forest Zones used to classify current riparian forest vegetation from lidar.
  • Potential Vegetation Types
    • Watershed Potential Vegetation Types (Vector Data, 4 MB)
      • Potential Vegetation Types of riparian and upland ecosystems in the Wilson River watershed.
    • Riparian Potential Vegetation Types (Vector Data, 871 KB)
      • Potential Vegetation Types of riparian zones only. These polygons are also mapping units for aggregating current non-forest and forest vegetation attributes into states used in state and transition models.
  • Current Vegetation
    • Current Riparian Vegetation (Vector Data, 871 KB)
      • Current (2007) riparian vegetation classified into vegetation states used in state and transition models. Vegetation attributes include barren, herbaceous, shrub, and tree cover, tree size classes, and forest strata.
  • Intrinsic Potential & Initial Conditions
    • IP and Initial Conditions (Vector Data)
      • Intrinsic Potential for Spring Chinook and Steelhead calculated for each reach, and used to identify reaches to be modeled and determine initial conditions for Aquatic-Riparian models runs.
  • Aquatic-Riparian Models v. 2011_02_05
    • Aquatic-Riparian Models (Access .mdb, 1.4 MB)
      • Vegetation Dynamics Development Tool (VDDT) database for running aquatic-riparian state and transition models. Database includes all Potential Vegetation Type and channel morphology combinations for the Wilson River.
    • Model Documentation (PDFs .pdf, 648 KB)
      • Summaries of each model’s structure, disturbance dynamics, vegetation and fish habitat attributes.
    • VDDT (Software)
      • Vegetation Dynamics Development Tool can be downloaded from ESSA Technologies.
    • PATH (Software)
      • Path Landscape Model tool can be downloaded from Apex Resource Management Solutions, Ltd.
  • Aquatic-Riparian Models v. 2011_05_30
    • Aquatic-Riparian Models (Zip, 19 MB)
      • Most current version of the state and transition models (PATH versions only). Includes all revisions and edits made since 2011_02_25. Also includes model packages used for analysis of the Wilson River watershed.

Key References:


Beukema, S.J., Kurz,W.A., Pinkham, C.B., Milosheva, K., Frid, L., 2003. Vegetation Dynamics Development Tool, User’s Guide, Version 4.4c. ESSA Technologies Ltd., Vancouver, B.C., Canada, pp. 239.

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.

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, Oregon, USA. Landscape and Urban Planning 80:198-211.

McDowell, P. F. and Magilligan, F. J. 1997. Response of stream channels to removal of cattle grazing disturbance: Overview of western U.S. Exclosure studies. Pages 469-475. In: S.S.Y. Wang, E. J. Langendoen, and F. D. Shields (eds.). Management of Landscapes Disturbed by Channel Incision. The Center for Computational Hydrosciences and Engineering, University of Mississippi, Oxford, MS.

Montgomery, D. R. and Buffington, J. M. 1997. Channel-reach morphology in mountain drainage basins. Geological Society of America Bulletin 109:596-611.

Montgomery, D. R. and Buffington, J. M. 1998. Chapter 2: Channel processes, classification, and response. Pgs 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.

Thompson, J. 2007. Simulating the consequences of land management, based on science by Steven Wondzell, and Pete Bisson. USDA Forest Service, Pacific Northwest Research Station, Science Findings #92.

Wondzell, S. M. 2001. The Influence of Forest Health and Protection Treatments on Erosion and Stream Sedimentation in Forested Watersheds of Eastern Oregon and Washington. Northwest Science 75:128-140.

Wondzell, S. M., and Howell, P. J. 2004. Chapter 6: Developing a decision support model for assessing condition and prioritizing the restoration of aquatic habitat in the interior Columbia Basin. Pgs. 73-81. 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.

Wondzell, S. M., Burnett, K. M., and Kline, J. D. 2007. Landscape analysis: Projecting the effects of management and natural disturbances on forest and watershed resources of the Blue Mountains, OR, USA. Landscape and Urban Planning 80:193-197.

Wondzell, S. M., Hemstrom, M. A., and Bisson, P. A. 2007. 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 80:249-267.

Vavra, V., Hemstrom, M. A., and Wisdom, M. 2006. Modeling the effects of herbivores on the abundance of forest overstory states using a state-transition approach in the upper Grande Ronde River Basin, Oregon, USA. Landscape Urban Planning 80: 212-222.