ATLAS and SIMFOR both simplify and abstract reality. Their inputs are not literally represented within the model, nor can their outputs be literally transferred by decision makers into management plans. Understanding the simplifications and abstractions in creating datasets and algorithms is crucial to understanding what types of knowledge can be gained from the outputs. What aspects of a spatial harvest schedule can suffer generalization to a statement of how the landscape will actually appear in 100 years? When is accuracy error (mistakes in measurement) less important than abstraction error (mistakes in understanding)? Both abstraction and accuracy error are present in all model outputs; the challenge for modellers is to recognize how reliable knowledge can be found in spite of error. Based on the relevant sections of “Decision-support systems: it’s the question not the model” (Bunnell and Boyland, 2002), this paper explores in simple terms a framework for understanding the different types of error within models.

The many unique features of dead wood make it attractive as a place to forage, nest, and shelter. In the Pacific northwest, 69 vertebrate species commonly use cavities, and 47 species respond positively to down wood. Cavity users typically represent 25% to 30% of the terrestrial vertebrate fauna in forests of the Pacific Northwest. Cavity excavators select for heart rot – major factors governing it’s presence are tree species, site productivity, tree age, and time. Data presented shows 2-3 large snags per hectare, and 10-20 smaller snags per hectare, throughout the rotation, are required to sustain cavity nesting birds.

This paper presents an introduction to harvest schedule modelling. Harvest scheduling models are designed to help resolve the resource conflicts common to British Columbia’s forests. Understanding these models helps understand what role they can play in the decision making process. A short review of the development of harvest schedule models is presented. This is followed by presentation of a classification scheme for harvest scheduling models on: algorithm type, harvest unit formulation, and constraint formulation. Finally, model use and interpretation is analyzed, focussing on common model misuses.

As regional and sub-regional land use planning processes are completed around the province, numerous strategies for managing the crown forested land base must be implemented by the forest industry. Strategies that typically impact forest-harvesting operations are spatial in nature and require some form of forest cover to be maintained. An example could be: at least 40% of the forested area within an identified caribou zone must be mature at all times (40%>140 yrs old). This simple rule becomes more complicated as steep slopes are excluded and it must be evaluated separately for each biogeoclimatic zone. It quickly becomes apparent that computer models are a very useful tool for evaluating these types of rules and predicting impacts on timber availability through time. Spatially explicit timber supply modeling is capable of completing this task while also illustrating how it looks on the ground. In order to get realistic implementation direction (harvest blocks) from the model, sound operational planning should drive landscape level strategies for non-timber resources, which in turn drive timber harvest projections (bottom-up approach).

Revelstoke Community Forest Corporation (RCFC) has completed numerous land base assessments and total chance harvesting planning initiatives that allow a bottom up approach to be applied when modeling Tree Farm Licence (TFL) 56. This extension note describes how detailed spatial timber supply modeling was used on TFL 56 to:

•Influence the development of strategic land use decisions in the Revelstoke Forest District, and ultimately,

•Complete the TFL Management Plan Process so that all levels of planning (operational – landscape – strategic) came together to give realistic harvest projections and explicit direction for how to achieve it on the ground.

This report is structured as a chronological series of events or initiatives that represents RCFC’s experience with spatial modeling and how it was used to bridge all levels of planning on their land base.

Patch definitions are reviewed, included age-, seral-, and structural-based, as well as habitat-based systems. These systems all classify contiguous stands into patches based on similarity criteria. Patch dynamics are explored, showing how patch distributions change in a variety of classification-dependent ways as the landscape ages. Management recommendations include using objectives, not rules, to avoid the reductions in other resources from "hard" rules; to recognize the uniqueness of landscape units, and create plans specific to the objectives and landscape at hand; and to increase our diversity of patch management strategies to hedge our bets for protecting biodiversity.

The models ATLAS and SIMFOR were used together to evaluate forest management scenarios at the landscape unit planning scale. Caribou habitat, seral stage constraints, and timber harvests were measured over a 250 year planning horizon to monitor the success of three management strategies: Biodiversity Recruitment strategy, Caribou Habitat guidelines, and Old Seral Patch strategy. Previous work with ATLAS has used "hard" rules, where no harvesting is allowed when a constraint is violated. The biodiversity recruitment strategy successfully used a "soft" approach of allowing harvest while recruiting stands to fill the constraints. SIMFOR's approach of Habitat Suitability Indices were conceptually accepted by Ministry of Environment staff, including capability and suitability maps. The combination of ATLAS and SIMFOR points to an alternative management approach of balancing the caribou habitat values against the wood supply values, as opposed to simply meeting the KBLUP caribou guidelines (rules) and maximizing wood supply.

Adjacency rules are used to limit opening sizes by preventing cutting on stands neighbouring an existing clearcut.  Openings are not allowed to “grow” in size until the original opening’s regeneration has reached a specified age and/or height.  Adjacency has been used to control opening size, disperse harvest units across the landscape, and to control the rate of harvest.

This extension note describes the mechanics of adjacency, showing how different types of adjacency rules affect the landscape in different ways.  It then goes into the impacts of adjacency on timber supply, showing both the potential short- and long-term effects.  And finally, methods for mitigating the effects of adjacency on timber supply are presented.

During the past 20 years public forests in British Columbia have been managed under regulations that have steadily decreased the allowable harvest unit size (i.e. opening size).  These regulations have been generically applied over large regions, and the resulting landscape patterns are readily apparent in most publicly managed forests.   In the interest of ecosystem management and biodiversity, academics have long cautioned about the universal application of one regulation on all forests.  Forest management is entering a new era where variety in opening size is considered desirable. Introducing different opening sizes on landscapes that were altered by past regulation is a difficult task and requires a thorough analysis of the consequences on the timber supply and landscape pattern.  This extension note summarizes a case study of opening size options that were simulated with the ATLAS model.  It describes the study site, generation of harvest blocks, harvest scenarios, and timber production under each scenario.  It concludes with recommendations for introducing new harvest patterns on altered landscapes.

Patches can be divided into two types: those created by harvesting (openings) and the isolated older stands created by cutting all the stands around them (residual patches). BC has moved from a policy of progressive clearcutting, through the 3-pass cut/leave system, to the present small openings combined with adjacency and green-up rules.  These three systems have left their distinctive patch patterns on the landscape. This evolution was driven by ecological, silvicultural, and hydrological issues surrounding the earlier two policies. However, new environmental and social problems have emerged with the present system.  These include the availability of short-term timber supply, access management related to dispersed harvests, and forest fragmentation.

This report uses a recent case study done with the ATLAS/SIMFOR models to illustrate how alternate policies of harvest opening size, adjacency, and forcing new patch policy rules onto landscapes with a legacy of harvesting and natural disturbance can have significant short-term impacts on timber supply.  Recruitment into older age classes to replace naturally disturbed old seral stands is also discussed.  Tailoring specific policies for each landscape is recommended to avoid unnecessary timber supply shortages.

Strategic planning models for large forest estates can become very inefficient if proper care is not taken in defining the appropriate level of spatial detail necessary to answer the strategic questions. There is a tendency to carry spatial detail appropriate for operational and tactical planning to the strategic level, which results in cumbersome databases and models. In addition, there is a tendency to treat all inventories within GIS as though they share the same accuracy, which results in numerous resultant polygons. The overall effect is that these large databases and models discourage sensitivity analysis and strategic thinking. In this paper we explicitly define the strategic questions we are trying to answer, and then propose a method for generalizing inventory data to a set of common polygons. In a case study of the Kootenay Lake Timber Supply Area, we compare the effectiveness of detailed and generalized polygons in a strategic planning model. By generalizing, we reduce the number of polygons by approximately an order of magnitude, yet are able to generate harvest schedules with a few percent of the detailed polygon schedules. Database size is significantly reduced, as are model run times. The generalized polygons can provide a data link to tactical and operational planning where more detail is desirable through GIS overlays. Recommendations are made for improving our technique and these center on methods to minimize loss of small, irregular shaped polygons (e.g. riparian zones) during generalization, plus methods for creating different sized polygons in various compartments (small in riparian, large in the inoperable forest).