Epsilon Open Archive

Abstract

Traditionally forest planning is divided into a hierarchy of planning phases. Strategic planning is conducted to make decisions about sustainable harvest levels while taking into account legislation and policy issues. Within the frame of the strategic plan, the purpose of tactical planning is to schedule harvest operations to specific areas in the immediate few years and on a finer time scale than in the strategic plan. The operative phase focuses on scheduling harvest crews on a monthly or weekly basis, truck scheduling and choosing bucking instructions. Decisions at each level are to a varying degree supported by computerized tools. A problem that may arise when planning is divided into levels and that is noted in the literature focusing on decision support tools is that solutions at one level may be inconsistent with the results of another level. When moving from the strategic plan to the tactical plan, three sources of inconsistencies are often present; spatial discrepancies, temporal discrepancies and discrepancies due to different levels of constraint. The models used in the papers presented in this thesis approaches two of these discrepancies. To address the spatial discrepancies, the same spatial resolution has been used at both levels, i.e., stands. Temporal discrepancies are addressed by modelling the tactical and strategic issues simultaneously. Integrated approaches can yield large models. One way of circumventing this is to aggregate time and/or space. The first paper addresses the consequences of temporal aggregation in the strategic part of a mixed integer programming integrated strategic/tactical model. For reference, linear programming based strategic models are also used. The results of the first paper provide information on what temporal resolutions could be used and indicate that outputs from strategic and integrated plans are not particularly affected by the number of equal length strategic periods when more than five periods, i.e. about 20 year period length, are used. The approach used in the first paper could produce models that are very large, and the second paper provides a two-stage procedure that can reduce the number of variables and preserve the allocation of stands to the first 10 years provided by a linear programming based strategic plan, while concentrating tactical harvest activities using a penalty concept in a mixed integer programming formulation. Results show that it is possible to use the approach to concentrate harvest activities at the tactical level in a full scale forest management scenario. In the case study, the effects of concentration on strategic outputs were small, and the number of harvest tracts declined towards a minimum level. Furthermore, the discrepancies between the two planning levels were small.