This article is summarized and concluded from articles as follow:
(1) Making GIS Work in Forest Management by Manij Upadhyay, unpublished work.
(2) A Hierarchical Approach to Spatial Forest Planning by Ugo Feunekes and Andrew Cogswell, USDA Forest Service, Proceeding
(3) HCVF/A Identify Within Ecoregion; Integrating Conservation Planning into Regional Spatial Planning by Barano Siswa Sulistyawan, WWF Indonesia,
(4) Assessing participatory GIS for community-based natural resources management: claiming community forests in Cameroon by Michael K. McCall and Peter A. Minang, The Geography Journal Volume 171 No.(4) pages 283-306, 2010
Previous Section Forest Resources Spatial Management and GIS
A Hierarchical Approach to Spatial Forest Planning to Create Harvest Blocks (Case study of Remsoft)
Integrating GIS into Forest Resources Spatial Management and Planning (3) — In term of forest product’s harvesting, spatial planning problems are especially difficult to solve for three reasons. First, most scheduling problems involve large numbers of stands and/or harvest block. Second, a long-term look is required to address sustainability—usually several rotations. As planning horizons increase, the decision variables and constraints necessary to represent adjacencies increase exponentially. Finally, spatial allocation and scheduling in the second-growth forest is often dubious because of the uncertainty in regeneration responses. When all of these factors are considered together, it is clear that finding a true optimal solution to an unrestricted problem is virtually impossible. As a result, every spatial planning approach has focused on finding good or near-optimal, feasible solutions, but on simplified problems (Feunekes and Cogswell, 1993).
A number of different techniques have been employed to solve spatially-constrained harvest scheduling problems. These include various mixed-integer programming formulations (Meneghin et al., 1988; Jones et al., 1991; Weintraub et al., 1994; Yoshimoto et al., 1994), binary search or inventory projection models (Baskent, 1990),simulated annealing (Lockwood and Moore, 1990), and Monte Carlo integer programming (MCIP) (O’Hara et al., 1989; C ements et al., 1990), to name a few.
Yet, all techniques developed by Nelson and Brodie, Yoshimoto et al., Clements et al., and Weintraub et al. did not satisfactory meet the requirement. Annealing simulation used by Lockwood and Moore is limiting the process to long-term sustainability issues. The inability to balance multiple product flows is a major drawback for Baskent. Others have tried to solve the spatial scheduling problems in two steps, separating the problem based on the amount of time and spatial detail considered. Beyond the problem of manually delineating blocks, the difficulty with this approach is that the only explicit linkage between the strategic and tactical planning models is the allowable harvest. Thus, the tactical solution may be incompatible with more long-term goals.
Jamnick and Walters (1991) also used a hierarchical approach in which the problem was solved in two distinct phases. The first was a strategic planning phase in which long-term (two or more rotations), non-spatial objectives and constraints were evaluated using traditional linier programming techniques. The second phase was a spatial planning phase where harvest activities were scheduled subject to adjacency delay, opening size, and harvest flow constraints over a much shorter time frame (one rotation). For a number of years, Remsoft has been refining and adapting the Jamnick and Walters approach, building on its inherent strengths, addressing its weaknesses, and turning the approach into an implementable spatial planning system.
To solve spatial problem, the design should meet the criteria included the following: the system had to (1) be flexible to allow it to operate in different jurisdictions; (2) be scalable; that is, it should work on small problems as well as operational-sized problems containing hundreds of thousands of polygons; (3) be theoretically sound in terms of ensuring that long-term outlooks are not ignored; (4) make reasonable demands in terms of hardware, time to solution, and other system requirements; and finally, (5) provide reasonable, near-optimal solutions.
In summary, the system introduced by Jamnick and Walters and improved by Remsoft can simplifies the solution of the overall problem. It demonstrates that the spatial complexity of the problem can be increased and solved without increasing the complexity of the spatial algorithm. In effect, the two phases worked together to solve a more complicated problem.
The three case studies presented show that the approach works at various spatial scales on realistic-sized problems and considers long-term trade-offs. Pseudospatial resolution, in which subforest areas are recognized, can easily be accommodated in this planning phase, allowing for the distinction of features such as watersheds, management units, forest districts, etc. An essential factor is that long-term trade-offs and constraints are handled in the strategic phase, and that the software required to allocate stands to specific periods (spatial phase) can therefore be much simpler. Also, since the long- and short-term phases are so closely linked, any changes to the strategic model are directly communicated to the spatial stage.
Remarks:
You can download the full paper in PDF at the end of the lecture series.
*Geo-information for Spatial Planning and Risk Management — Batch 6 — Faculty of Geography, Gadjah Mada University*
*Intended to fulfil Spatial Planning lecture task*