Application note
GIS tool for optimization of forest harvest-scheduling

https://doi.org/10.1016/j.compag.2015.03.001Get rights and content

Highlights

  • We created ArcGIS extension for forest harvest scheduling problems.

  • The extension is connected to GUROBI optimization software.

  • The extension facilitates spatial and temporal design of harvest units.

  • The extension provides effective way of creating forest harvest scheduling scenarios.

Abstract

This article describes GIS tool (Optimal) for spatial and temporal optimization of forest harvests. Using Optimal, forest manager can create harvest units by editing polygons of forest stands in digital map. After the harvest units are created manually by the user, the adjacency matrix is automatically produced and passed to a solver module. The solver performs optimization using integer programming and returns spatial distribution of harvest units for each harvest period. User can set number of parameters, such as number and length of harvest periods, acceptable distances and areas of harvest units. The Optimal enables the forest managers to create and explore various scenarios and increase efficiency in forest harvest-scheduling.

Introduction

There are basically two main aspects of forest harvest-scheduling: Space and time. The forest spatial structure refers to the spatial arrangement of forest stands, harvest units or patches and interconnections among them (Baskent and Keles, 2005). The spatial structure plays important role in providing ecosystem services (Kurttila, 2001) and cannot be omitted in forest harvest scheduling. Temporal aspect is important for supplying good quality timber to the market according to market demand and at the same time preserving enough of it in the forest for the future.

The clear cut forest management system is commonly used in the Central Europe because of its cost efficiency. For preserving biodiversity and other non-timber forest products, the size and spatial relationship of the clear cuts is usually limited by law. The limitations can be expressed through four constraints: (1) The maximum area of the clear cut unit. The default is 1 ha, which is legal limit for clear cuts in the Czech Republic. (2) The minimum distance between the two clear cut units harvested in the same period, which is usually set equal to height of the forest stand. This would prevent the remaining forest stands from being vulnerable by wind. (3) The maximum width of the clear cut unit, which is usually set equal to legal limit of two heights of the forest stand. (4) Adjacency relationship, which is usually set to not to allow Queen’s case (see below) as this is an official limit included in forestry legislation of the Czech Republic. Queen’s case neighboring may be allowed in special cases where reconstruction of forest stands has to be done faster than usual. A neighboring clear cut unit can only be harvested when the area is regenerated to the point where it is stable forest stand again, so called green-up constraint (Bettinger et al., 2009). All these restrictions make harvest scheduling model computationally difficult to solve even for quite small forest management area. There are basically two possible modeling approaches to solve our problem, Area Restricted Models (ARMs) and Unit Restricted Models (URMs). It has been proved that ARMs have number of advantages over URMs (Richards and Gunn, 2000). For example higher values of total harvests or lower harvest flow percentages (Murray, 1999). However, because of the harvest unit shape restrictions we used URM modeling approach.

Today, most of the forest management plans can be designed only with the use of geographic information systems (Baskent and Keles, 2005). Over the last decade, there is increasing number of approaches, which deal with spatial aspects of harvest scheduling (Ohman and Eriksson, 2002, Ohman and Lamas, 2003, Ohman and Lamas, 2005, Baskent and Kelles, 2006). A decision support systems for spatial harvest optimization were developed, e.g. SNAP (Sessions and Sessions, 1988) or HEUREKA (Wikström et al., 2011). Other solutions used for this purpose like J-Software (Lappi and Lempinen, 2013) are rather development tools, not ready to use systems. These systems can optimize the spatial distribution of the harvest units, but lack the inbuilt editing and checking capabilities needed for construction of harvest units. Law restrictions for clear cut management system in number of countries are quite different, making it difficult to adopt single solution. The main objective of this paper is to develop a GIS tool to help forest managers with spatial harvest planning, including algorithms for creating harvesting units and estimating periodically harvesting flows.

Section snippets

Components of the framework

Optimal is an extension of proprietary geographic information system ArcGIS. It is combination of geographic information system (GIS) tool and mixed integer linear programming (MIP) solver. Optimal extension is designed for forest managers who have no understanding of MIP or any mathematics used in the model. However, basic knowledge of operating GIS is assumed. The basic structure of the software is schematically described in Fig. 1.

The extension uses Add-In concept introduced with ArcGIS

Case study

The case study is presented on 46.5 ha of mature Spruce forest stands. It is based on real data, which is used with the agreement of the forest management area owner, but to comply with the rules for protection of personal data, it is not identified more specifically. The stocking volume ranges from 264 to 758 m3/ha with the average 540 m3/ha and standard deviation 61 m3/ha. The area has been divided into 92 harvest units. Several scenarios of harvest flow percentages (i.e. the differences in

Results and conclusions

Results of optimization are presented in the form of easy to understand map showing spatial distribution of harvest units in individual harvesting periods (see Fig. 5). User can repeat the simulation with different parameters and compare results. Our case study scenarios are built on various values of harvest flow. However, scenarios can be also built around different spatial and temporal constraints such as neighbor distances, number of periods and length of the periods.

We see the key value of

Software availability

Name of software: Optimal. Extension is available on request to: [email protected], [email protected]. Developers: Petr Vopěnka, Jan Kašpar. Contact address: Czech University of Life Sciences Prague, Kamycka 129, Praha 6 – Suchdol, Czech Republic. E-mail: [email protected], [email protected].

Acknowledgments

This research was supported by the project of the National Agency for Agriculture Research (No. QJ13202302) and the Internal Grant Agency of Faculty of Forestry and Wood Sciences Czech University of Life Sciences in Prague (No. B0114).

References (18)

There are more references available in the full text version of this article.

Cited by (20)

  • Forest landscape planning and management: A state-of-the-art review

    2022, Trees, Forests and People
    Citation Excerpt :

    These virtual results may concern stakeholders’ views and goals. Some examples can be observed in Meo et al. (2013) with a GIS of public participation (PPGIS) in forest landscape planning; Vopěnka et al. (2015) with a developed GIS extension for forest harvest scheduling problems with spatial and temporal design aspects; a GIS-based model to assess the impact on recreation areas caused by harvesting activities in a commercial forest (Harshaw and Sheppard, 2013); and based on a Multicriteria Decision Analysis (MCDA) methodology, Ezzati et al. (2016) found optimal locations of viable harvest zones in mountainous areas. In MCDA and spatial forest planning, the tool AHP (Analytic Hierarchy Process) methodology (Saaty, 1980) has been widely used to model problems of assessing preferences for multiple criteria (Alho et al., 2002; Kangas and Kangas, 2002; França et al., 2020; Morandi et al., 2020).

  • Operational harvest planning under forest road maintenance uncertainty

    2021, Forest Policy and Economics
    Citation Excerpt :

    In this model, the harvesting of adjacent stands is allowed as long as they do not exceed the maximum allowed cut size (Murray, 1999). The inclusion of spatial constraints can inflate the model matrices, imposing high computational costs in large-scale forest projects (Augustynczik et al., 2016; Öhman and Eriksson, 2010; Vopěnka et al., 2015). Nevertheless, several authors tried to improve these formulations and approached spatial modeling with exact methods to solving spatial forest scheduling (Gharbi et al., 2019).

  • Optimal production planning and machinery scheduling for semi-arid farms

    2021, Computers and Electronics in Agriculture
    Citation Excerpt :

    Another study proposed a two-stage analytical methodology to develop a dynamic capacity planning method for maintenance service to agricultural machinery (Hu et al., 2020). Additional scheduling research has also considered the coupling of crop assignment with vehicle routing (Graf Plessen, 2019), farmer’s demands and dispatch system (Si et al., 2019), two-echelon multi-trip vehicle routing problem (He and Li, 2019), inventory control issues (Kusumastuti et al., 2016), geographic information system (GIS) tools (Vopenka et al., 2015; Thorp et al., 2015), scheduling of composts and mineral fertilizers amendments (Ramos-Castillo et al., 2020), integer programming combine-harvester scheduling model (He et al., 2018a), and harvest scheduling (Thuankaewsing et al., 2015; Amiama et al., 2015; Robinson et al., 2016; He et al., 2018b). In spite of the current literature on production planning and machinery scheduling, research gaps still remain, especially when considering the adoption of low-water-use crops in semi-arid regions.

  • A new integer model for the management of booking stays at summer resorts

    2021, Computers and Industrial Engineering
    Citation Excerpt :

    The latter are broadly classified into two categories based on licensing. These are an open-source software such as Gecode (Schulte, Tack, & Lagerkvist, 2010), Choco (Jussien, Rochart, & Lorca, 2008), Lindo (Lin & Schrage, 2009), Coin-or Branch and Cut (CBC) (Nonaka, Erdős, Kis, Nakano, & Váncza, 2012), and commercial solvers, e.g., CPLEX (Fontaine, 2008; Lodi & Tramontani, 2013), Guroubi (Vopěnka, Kašpar, & Marušák, 2015), XPRESS (Dormer, Vazacopoulos, Verma, & Tipi, 2005). These solvers allow for providing a better design and development of optimization solutions for AP.

View all citing articles on Scopus
View full text