Difference between revisions of "GAYA"

Revision as of 11:13, 19 February 2010

General System description

System name: GAYA

Brief overview

The first version of GAYA was developed by Prof. Ljusk Ola Eriksson at the Swedish University of Agricultural Sciences. GAYA was "imported" to Norway and transferred to Norwegian conditions by the professors Hans Fredrik Hoen and Tron Eid at the Norwegian University of Life Sciences around 1990. GAYA is a stand simulator and different optimization tools can be connected.

Scope of the system

GAYA-JLP is used both for analyse long range forest management planning at forest level based on information for individual forest stands, and for analyses based on (aggregated) sample plot data for the national forest inventory.

It has two main modules the stand simulator and the decision model.

System origin

• GAYA model used simulations to describe silvicultural regimes in single sample plots, stands or strata, and employed linear programming (LP) to settle management strategies on a forest level. Further development has taken place since 1990 and the model, today called GAYA-JLP, is now a comprehensive tool for long-term analyses^[1].
• GAYA-JLP was developed for long-term economic analysis of forest production in Norway.
• It was used for research and education at NSK (NABUURS et PÄIVINEN 1996).

Support for specific issues

It provides support for forest management planning, through harvesting scheduling, economic evaluation, CO₂-flow evaluation, yield prediction, and its optimization.

Support for specific thematic areas of a problem type

• Silvicultural
• Certification
• Conservation
• Development choices / land use zoning
• Policy/intervention alternatives

Related systems

It has a strong resemblance to the Finish MELA model.

Data and data models

Typical spatial extent of application

Large scale model using stand level growth model.

It is used both for analyse long range forest management planning at forest level based on information for individual forest stands, and for analyses based on (aggregated) sample plot data for the national forest inventory.

Forest data input

There are a total of 28 state variables defined for the stand. They are divided between species, biologically, and economically related variables.

• Species is defined by tree number, basal are, height, age and volume.
• Stand is further defined by area, site index, and e.g. earlier treatments in the rotation.
• Economical stand related variables are the terrain transportation distance, several difficulty parameters related to felling operations and a wood quality parameter.

Type of information input from user

The following control variables can be user-specified in decision making:

• basal area of the removal as a percentage of total basal area;
• basal area in the remaining stand;
• number of trees in the removal;
• number of trees in the remaining stand;
• diameter ratio between removed trees and remaining stand;
• type of fertiliser;
• intensity of fertiliser.

Within the optimization package JLP some constraints have also to be defined, such as non-declining flow of wood, maximisation of NPV, etc.

Models

Forest models

GAYA is the simulator module. It simulates alternative forest management treatment schedules for each calculation unit. A rotation is divided in two periods, the regeneration and the thinning phase. Within the former, the development cannot be manipulated; it is endogenously defined; only the length of the period can be user defined. During the latter, up to three species per stand can be simulated, projecting its development separately. Development is simulated based on regression functions in terms of diameter, basal area, height, spacing and natural mortality. The time dimension of the calculation, the treatment unit, is user controlled, it is a discrete number of time period of uniform length of 5 or 10 years. Cost and revenues for each management unit are also calculated.

Decision Support

Definition of management interventions

Silvicultural treatments, fertilization, regeneration methods, thinnings, harvesting.

Typical temporal scale of application

Tactical planning.

Decision-making processes and models

The implemented optimization module to solve the planning problem is the JLP-package. It uses linear programming.

Output

Types of outputs

Outputs are displayed in tables, showing nearly 40 defined variables for each time period. They are divided in periodic and non-periodic variables.

Periodic variables

Treatment undertaken in the period.

For stand after treatment: total tree number, standing volume by species, basal area, dominant height, age, and NAI (Net Area Increment).

Removal: total tree number, volume for each species., basal area, mean diameter, sawwood and pulpwood proportions, price and cost information (including haulage costs), cash flow, total biomass, and net CO₂ flow.

Non-periodic variables

For the treatment schedule: NPV at time zero, ending inventory value (EIV; NPV at the end of the last period). NPV and EIV of CO₂-flow, total NPV in case CO₂-flow is priced.

Variables for the calculation unit: number of schedules, area, site, vegetation type, altitude, rentability’’’, slope, and a GIS-specified treatment code.

Spatial analysis capabilities

Spatial constraints such as adjacency cannot be taken into account. However, a GIS integration implementation is operative.

System

System requirements

Operating Systems: DOS (5.O or above), OS/2 (2.1 or above) or Windows (3.1 or above).

Architecture and major DSS components

As it was said before, the GAYA-JLP system is divided in two modules, the simulation model GAYA, and the optimization module using JLP.

Usage

It was used at research and educational level.

Computational limitations

The GAYA-JLP system has solved problems with 200,000 decision variables and 8,000 stands (NABUURS et PÄIVINEN 1996).

User interface

Command line interface. Building all the management schedules is very laborious, especially when large areas and detailed outputs are wanted.

References

Cited references

1. ↑ EID, T. et K. HOBBELSTAD (2000): AVVIRK-2000: A Large-scale Forestry Scenario Model for Long-term Investment, Income and Harvest Analyses. ‘’Scand. J. For. Res.’’ 15: 472-482.

External resources

• NABUURS G.J. et R. PÄIVINEN (1996): Large Scale Forestry Scenario Models - a compilation and review. European Forest Institute Working Paper No. 10. Joensuu, Finland.