Integrating wood quality considerations in landscape-level forest planning Tuomo Kalliokoski, Jordi Garcia-Gonzalo, Ane Zubizarreta-Gerendiain, José G. Borges & Annikki Mäkelä 6.5.2013 1
Outline Backgroung/Research question/motivation/aims PipeQual: principle and structure PipeQual: simulation results SADfLOR platform SADfLOR: Preliminary results Conclusions/Lessons learned 6.5.2013 2
Background Wood quality definition: Wood "quality" means different things to different people. - foresters think of tree size and form, - lumber manufacturers see large, straight and clear logs, - customers associate wood quality with other attributes. 6.5.2013 3
Background Wood quality definition: Mitchell (1961) appears to be the most widely cited: "Wood quality is the resultant of physical and chemical characteristics possessed by a tree or a part of a tree that enable it to meet the property requirements for different end products". 6.5.2013 4
Background The interests of forest industries are nowadays beyond wood quantity, and they also look for the raw material quality. We need tools to provide information on the impacts of management alternatives on the raw material properties. Until recently wood quality modelling got much less attention than growth and yield modelling. Integrating these models in DSS may help to identify a set of proper management alternatives to take into account wood quality into forest planning in order to produce desired raw material under. 6.5.2013 5
Motivation Conference in Bordeaux in 2010. Anniki Makela presented an application of PipeQual to predict wood quality. I presented SADfLOR platform. Discussion started 6.5.2013 6
Motivation What can you do with simulators? Check the outputs from a specified prescription (wood quality information) Scenario analysis Answer what if questions What you can not do or would be difficult to do with growth and yield simulator is to answer questions as: What is the optimal plan for a landscape taking into account wood quality? Together with other constraints (e.g. even-flow of harvests, budget constraints, adjacency constraints ) Why not Too to develop much work a DSS for a with single optimization STSM!!! techniques to address wood quality on forest management planning? 7
AIMS Combine a process-based stand simulation model PipeQual and forest management planning tool DECfLOR (module of SADfLOR platform) to include wood quality issues in forest planning. To generate optimal forest management for different scenarios at the forest property and landscape level in Finnish boreal forests. 6.5.2013 8
PipeQual: principle Relates foliage biomass to fine root biomass and sapwood area of stem, branches and coarse roots (Shinozaki et al. 1964) Stand growth from establishment to harvest with annual time step Resource acquisition (light) and treetree interactions stand structure at initialization differentiation of trees over time 6.5.2013 9
PipeQual: structure Growth of trees from photosynthesis, carbon balance on stand level Stand is divided to size classes Tree module describes the biomass of different compartments, and dimensions of crown length, height of crown base and sapwood area at crown base Whorl module contains models for the vertical structure of stem and branches. Module updates annually the total foliage biomass in each whorl based on branch sapwood areas Branch module calculates the annual dynamics of individual branches 6.5.2013 10
Modeled stand characteristics: no thinnings 6.5.2013 11
Modeled stand characteristics : five even thinnings, thinning intensity 10% 6.5.2013 12
Modeled stand characteristics : five even thinnings, thinning intensity 50% 6.5.2013 13
Information about impact of forest management 6.5.2013 14
SADfLOR structure Graphical user interface Management information module Prescription generator module Optimization module (DECfLOR) Inventory and sptatial data, Climate Costs scenarios and Prices, Simulation outputs Prescription writer Growth and yield Simulator Management model generator Solver (Optimizator) Results visualization and reporting module Report Graphs The SADfLOR platform that integrates specific systems to support management planning (Falcão and Borges 2005). The system used here integrates four independent and compatible modules, encapsulated in one single graphical interface.
Integration 6.5.2013 16
Integration 6.5.2013 17
Dataset used for initial stages of stands Mäkinen & Isomäki 2004, Mäkinen et al. 2006 Landscape formed by 128 stands Initial stage of stands Ages from 26 to 57 Mean height from 7.3 to 22.2 m Stand density from 780 to 3984 trees/ha Volume from 87 to 569 m3/ha 6.5.2013 18
Management alternatives used in simulations business-as-usual (Hyvän metsänhoidon suositukset- 2006, TAPIO) Rotation times, 60 to 100 years Two thinning intensities, 10% and 50% of stand density removed in each thinning Different type of thinnings (above even below) 6.5.2013 19
Linking wood quality and economic calculations For comparison purposes two different analyses were made: In first, only one price for produced logs, 56,80 /m3 In second, the produced logs were classified on basis of their maximum branch diameter and prices were given accordingly: Quality A: live branch max diam = 10 mm, dead branch max diam = 10 mm, price 113,60 /m3 Quality B: live branch max diam = 30 mm, dead branch max diam = 20 mm, price 85,20 /m3 Quality C: live branch max diam = 50 mm, dead branch max diam = 40 mm, price 56.80 /m3 In both analyses same costs were used 6.5.2013 20
Model formulation The forest management problem is defined according to a typical Model I formulation: MIP 6.5.2013 21
Preliminary results Results from the optimization indicate that: When considering different quality assortments: Heavy thinnings are preferred (except the first thinning) Longer rotations. Thinnings from below. If considering only one log wood quality: Light and more frequent thinnings. Shorter rotations may be considered. 6.5.2013 22
Conclussions/Lessons learned The inclusion of wood quality models in DSS may help to identify a set of proper management alternatives to produce desired raw material with certain characteristics. Preliminary results look promising. G&Y modelers and us do not allways speak the same language Understanding the input data needed for the optimizer and the modification needed by the G&Y models is not a trivial task e.g. new thinning routines Lesson: do not underestimate time needed to achieve the integration of data and tools. 6.5.2013 23
Acknowledgements Partial support for this research was provided by: MOTIVE (Models for Adaptive Forest Management) and INTEGRAL (Future-oriented integrated management of European forest landscapes) funded by 7th EU FP as well as from the ForEAdapt project, funded by the European Union Seventh Framework Programme (FP7-PEOPLE-2010- IRSES) under grant agreement n PIRSES-GA-2010-269257 FORSYS Forest Management Decision Support Systems Cost Action (FP0804). (STSM) 6.5.2013 24