Simulation of landowner behavior and potential forest policies at a large scale in Coastal Oregon

1. Simulation of landowner behavior and potential forest policies at a large scale in Coastal Oregon Pete Bettinger Warnell School of Forest Resources University of Georgia Marie Lennette Department of Forest Resources Oregon State University

2. Overall Simulation Process Tools and data for policy analysis Existing forest inventories Habitat condition for focal species Stand structure projections Management intentions Successional stages LAMPS Response models Prices and costs Recreation opportunities Synthesis of effects of alternative management scenarios Policy guidance Aquatic habitat / watershed potential GIS databases Landslide / debris flow Land use change Land use pattern Timber volume and value Employment and income Center for Forest Business, Warnell School of Forest Resources, University of Georgia

3. Landscape Modeled State-managed forest land Private industrial forest land Other ownership classes Study area Center for Forest Business, Warnell School of Forest Resources, University of Georgia

4. Hierarchical Spatial framework Basic simulation units Management units Harvest blocks Interior habitat blocks Land ownerships Landscape Softwood Mixed forest Hardwood Management unit boundary Center for Forest Business, Warnell School of Forest Resources, University of Georgia

5. Hierarchical Spatial framework Basic simulation units Management units Harvest blocks Interior habitat blocks Land ownerships Landscape Center for Forest Business, Warnell School of Forest Resources, University of Georgia

6. LAMPS simulation model Seeks to emulate management behavior of four major landowner groups represented in the Oregon Coast Range: federal, state, forest industry, non-industrial private. Facilitates an evaluation of forest policies across large areas, and over a long time frame (100 years). Merges strategic and tactical planning processes. Formulates problems as "Model I." Center for Forest Business, Warnell School of Forest Resources, University of Georgia

7. LAMPS simulation model Interface: VB 6.0 Scheduling model: C Center for Forest Business, Warnell School of Forest Resources, University of Georgia

8. LAMPS simulation model Four types of scheduling processes can be used. Federal State Forest Industry Non-industrial private The processes include spatial harvest scheduling aspects, yet the overall goal is to emulate management behavior across a broad scale. Center for Forest Business, Warnell School of Forest Resources, University of Georgia

9. LAMPS simulation model “Federal scheduling process” Monte Carlo approach to scheduling clearcuts. No clear objective function to optimize. Constraints: A maximum percentage of “matrix” land area can be clearcut in any one 5-year time period. A minimum amount of “older forest” within a watershed must be present before any clearcuts can be scheduled in that watershed. Center for Forest Business, Warnell School of Forest Resources, University of Georgia

10. LAMPS simulation model “State scheduling process” Monte Carlo approach to scheduling clearcuts. Objective is measured by the achievement of maximum even-flow of timber harvest volume, using binary search. Constraints: Achievement of structural conditions by State management District. Maintain a distribution of sizes of Interior Habitat Area (IHA) patches (older forest conditions spatially connected). Unit restriction adjacency. Center for Forest Business, Warnell School of Forest Resources, University of Georgia

11. LAMPS simulation model “Forest Industry scheduling process” Objective can be measured by the achievement of maximum even-flow of timber harvest volume, using binary search, or target harvest levels can be set by the user as goals to achieve. Constraints: Attempt to emulate a historical clearcut size distribution by “blocking” management units together for harvest using a dynamic process (block patterns are not fixed). Area restriction adjacency. Center for Forest Business, Warnell School of Forest Resources, University of Georgia

12. LAMPS simulation model “Non-Industrial Private scheduling process” Harvest probability approach to scheduling clearcuts. No clear objective function to maximize, although target harvest levels can be set by the user as goals to achieve. Constraints: Attempt to emulate a historical clearcut size distribution. Area restriction adjacency. Center for Forest Business, Warnell School of Forest Resources, University of Georgia

13. LAMPS simulation model Riparian policies: 1) No harvest within Oregon Forest Practices Act buffers. 2) Allow partial cutting within Oregon Forest Practices Act buffers, to the extent allowed by the Act. 3) Choice #2, yet no harvest of hardwoods within 100 feet of a stream. 4) Choice #1, and no harvest of hardwoods within 100 feet of a stream. No harvest in FPA buffers Harvest of other hardwoods within 100 ft Harvest in FPA buffers Harvest of other hardwoods within 100 ft 2 1 No harvest in FPA buffers No harvest of other hardwoods within 100 ft Harvest in FPA buffers No harvest of other hardwoods within 100 ft 4 3 Center for Forest Business, Warnell School of Forest Resources, University of Georgia

14. LAMPS simulation model Leave tree strategies: 1) Leave Oregon Forest Practices Act minimums (2 small TPA). 2) Leave 2 large TPA. 3) Leave 5 TPA, according to State Forest Plan guidelines. 4) Leave 14 TPA according to State Forest Plan guidelines. 5) Leave clumps of trees (BSUs, as a percentage of land area) uncut. Center for Forest Business, Warnell School of Forest Resources, University of Georgia

15. LAMPS simulation model Transition probabilities: After clearcutting, each BSU within a clearcut unit has a probability of being regenerated as one of four types: open, hardwood, mixed, or conifer. Probability = f (previous forest type, owner, ecoregion, distance from stream) Center for Forest Business, Warnell School of Forest Resources, University of Georgia

16. LAMPS simulation model Simulation notes: Projections are for 100 years, with 5-year time steps. All ownerships must be classified as either federal, state, forest industry, or non-industrial private. Classification of owners and owner policies is flexible. Forested area in the landscape modeled: 1,457,602 acres Center for Forest Business, Warnell School of Forest Resources, University of Georgia

17. Policies modeled 1. Base Case. Simulations tailored for the Coast Range of Oregon. 2. Extended green-up requirements, from 5 years (base case) to 10, 15, and 20 years. 3. Require leaving 10% or 20% of clearcut area in un-cut clumps. This is modeled at the BSU level. 4. Require, through a riparian policy, that no harvesting occurs within the Oregon Forest Practices Act buffers, and that no hardwoods are cut within 100 feet of the stream system. 5. Change the transition probabilities: Increase them 10% towards the successful establishment of conifers; decrease them 10% away from the successful establishment of conifers. Center for Forest Business, Warnell School of Forest Resources, University of Georgia

18. Results Basic simulation units Management units Harvest blocks Interior habitat blocks Land ownerships Landscape Center for Forest Business, Warnell School of Forest Resources, University of Georgia

19. Results Basic simulation units Management units Harvest blocks Interior habitat blocks Land ownerships Landscape Center for Forest Business, Warnell School of Forest Resources, University of Georgia

20. Results Base case Extended green-up requirements Center for Forest Business, Warnell School of Forest Resources, University of Georgia

21. Results Base case Leave tree clump requirements Center for Forest Business, Warnell School of Forest Resources, University of Georgia

22. Results Base case Riparian policy Center for Forest Business, Warnell School of Forest Resources, University of Georgia

23. Results Base case Changes in transition probabilities Center for Forest Business, Warnell School of Forest Resources, University of Georgia

24. Results Average difference in periodic volume from the base case 10-year green-up 15-year green-up 20-year green-up 10% leave tree clumping 20% leave tree clumping Riparian policy Lower conifer transition probabilities Higher conifer transition probabilities -19.4% -26.9% -27.9% -7.2% -14.0% -6.9% -6.1% +0.1% Center for Forest Business, Warnell School of Forest Resources, University of Georgia

25. Discussion LAMPS uses a hierarchical spatial structure to allow the modeling of large-scale, long-term forest policies for landscapes with multiple landowners. The framework is, of course, a simplification of a complex system of forest management. Recognition of management behavior, however, increases the credibility and realism of the simulations. Center for Forest Business, Warnell School of Forest Resources, University of Georgia

26. Discussion Neither the even-flow assumptions nor the constraints examined in these policies are Law. The even-flow objective for forest industry landowners was obtained from evidence of recent behavior of the industrial group as a whole. In the past, industrial landowners in Oregon had the ability to utilize federal timber sales to buffer changes in the market. Since federal timber sales have declined dramatically in the past decade, it remains to be seen whether these landowners will continue to harvest a relatively even amount of timber per year. Center for Forest Business, Warnell School of Forest Resources, University of Georgia

27. Discussion We have simulated a higher amount of volume harvested when even-flow objective is relaxed. Most simulations show a build-up of standing inventory after the "bottleneck" time period that constrains achievement of even-flow. Center for Forest Business, Warnell School of Forest Resources, University of Georgia

28. Summary The LAMPS simulation model is a spatial simulation model of landowner management behavior, and is associated with a quantitative projection of stand structures, which as nested in a hierarchical spatial framework. The model was designed to help policy makers and land managers “think through” forest policies before implementing them. Center for Forest Business, Warnell School of Forest Resources, University of Georgia

29. Acknowledgements • Funding for the development of LAMPS was provided by: • USDA Forest Service, Pacific Northwest Research Station • College of Forestry, Oregon State University • Oregon Department of Forestry • Support for this modeling effort provided by: • College of Forestry, Oregon State University • Warnell School of Forest Resources, University of Georgia Center for Forest Business, Warnell School of Forest Resources, University of Georgia