THE APPPLICATION OF GEOGRAPHIC INFORMATION SYSTEMS TO FOREST OPERATIONS:

THE INTEGRATION OF CABLE SETTING DESIGN INTO GIS

2nd Brazilian Symposium on Timber Harvesting and Forest Transportation; Salvador, Bahia, Brazil,

May 1995

Peter Schiess, Professor of Forest Engineering

Lawrence M. O'Brien, Graduate Research Assistant

University of Washington

Forest Engineering

Box 352100

Seattle, WA 98195-2100

Abstract

The Forest Engineering Group of the University of Washington, in cooperation with the Washington State Department of Natural Resources, is developing a CAE package to support tactical harvest and tranportation planning which includes spatial considerations. Its core is a GIS package around which off-the-shelf, or purpose-designed design programs, such as PLANS are arranged. Primary system functions are: logging setting location; physical harvesting feasibility analysis, transportation system location; timber harvest scheduling; and economic analysis of generated alternatives. Planning methodology integrates technology, procedures, human and organizational factors to enhance system utility and user acceptance. The system is based on a personal computer version of ARC/INFO to move data from one module to another.

Keywords: Harvest planning; cable setting analysis; geographic information systems; forest operations

1. Introduction

1.1. Planning and Design in the Absence of Certainty

Management implies the making and carrying out of plans and it is the creation of forest harvest plans that is the basis of this paper. The primary function of forest management has historically been to plan for the growth and harvest of timber in a manner that ensured a continuous supply of wood products. Viewing forest planning over an entire forest is referred to as "forest level management" (Leuschner 1990). In more recent times forest management has grown to include the planning for sustained supply of multiple forest resources including such diverse resources as water quality, wildlife, recreation and visual landscapes (Alexander 1989). In addition to the diverse and often competing range of resources that must be managed there are spatial and temporal issues that further complicate the management planning process.

Planning forest management activities includes the analysis of both forest resources and human resources. Issues such as community stability, regional economic health and fiscal responsibility are all laid on the desk of the modern forest manager (Duerr et al. 1982). In addition to these multiple and often conflicting management objectives the forest management planner also operates in an environment of incomplete information and insufficient ecological knowledge (Davis, 1966).

Given the forest planning environment, forest management planning is evidently a complex problem. Producing an optimum plan is essentially impossible since any given plan is based on incomplete information, and has multiple, often conflicting, objectives. Being unable to produce an optimum plan does not preclude forest management since waiting for perfect knowledge prior to making management decisions is infeasible (Walters 1986).

The production of forest management plans in the absence of perfect knowledge requires the plans to be modifiable and the planning process to be flexible. "The more flexibility that can be built into the planning process the less danger there will be that plans become irreversibly committed to policies, activities and investments which turn out to be wrong, or at least less than optimum" (FAO 1974). To simplify the problem planning methods focus on different levels of forest management from macro concerns such as national goals down to individual stand operations plans (USDA Forest Service 1982; FAO, 1974). Planning methods have also taken advantage of technological improvements to assist the planning effort (Iverson and Alston, 1986).

Forest harvest planning issues can be broadly divided into two categories: strategic and tactical, or operational. Strategic issues are those that consider the management of large land bases considering relatively few parameters over long time frames with the purpose of maintaining continuing viability of a resource given a particular set of management assumptions. Operational or tactical issues are those that concern the management of individual stands where management parameters are numerous and considerations are for a short time frame. Strategic issues are addressed by forest-level management planning while operational issues are addressed by stand-level management planning (Clutter et al. 1983). Procedural methods for timber harvest planning have been described by Pearce (Pearce 1961) and other authors (Silen, 1955; Klien 1954; Conway, 1982) have discussed specific principles of harvest planning, but timber harvest planning standards do not exist (Twito and McGaughey 1984).

This paper concerns a planning method that has been designed to provide flexibility, addresses a specific part of the forest management planning problem (timber harvest design) and utilizes computer technology. The method referred to is the computer based timber harvest planning and transportation system design method (hereafter referred to as the University of Washington Timber Harvest Planning System : UWTHPS), which was designed in cooperation with the Washington State DNR (Cullen and Schiess, 1992; Schiess and Fridley, 1993). The UWTHPS method depends on a modular design to provide flexibility in adapting the system to different planning situations. Modularity requires the transfer of data from module to module in order to analyze and process information. Data transfer in the UWTHPS has typically involved extensive human action to prepare data files of the required format for each module. It has been recognized that one of the major advantages of computer based systems is the ability to transfer data electronically. Electronic transfer can result in increased efficiency by eliminating the need to rekey data and preventing data errors inherent in the rekeying process (Bodner, 1987). Additionally the process is improved since the planner can focus on the data rather than on the mechanics of data transfer (Taylor ,1992).

1.2. Computer-Assisted-Engineering (CAE) in Harvest Planning

Successful forest engineering design of a forest harvest plan requires analyzing substantial geographic data and examining numerous harvest plan scenarios. Computer programs can be used to speed up the engineering design process and even improve the outcome. It has been demonstrated that judicious use of appropriate forest engineering tools, like PLANS (a computer program for the preliminary analysis of cable settings based on digitial elevation models, produced by the USDA Forest Service, also see McGaughey, 1992 ), can enable the designer to develop a far superior harvest plan.

Forest engineering design traditionally has been a paper and pencil based activity. Implementing a systematic computer aided engineering (CAE) approach to harvest planning activity enabled harvest plan designs to be produced that: 1) are better harvest plans than were produced, 2) require less elapsed time to produce, 3) allow for effective storage, retrieval and editing of harvest plans and 4) interface and integrate with other corporate databases.

The CAE approach bases all or nearly all of the engineering design and analysis activity on numerical processing and computer graphics (as opposed to traditional paper based engineering work). Elapsed and engineering time are reduced compared to traditional approaches because information is passed from one step (or computer program) in the process to another without necessarily producing "hardcopy" output or input. Better designs result because the engineer can be concerned entirely with the design instead of the mechanics of designing. Key to the successful forest engineering CAE approach are: 1) the existence of computer based geographic data, 2) the existence of cost and performance data for forest operations and roads, 3) easy data interchange between the various computer programs used by the designer, 4) easy input, output, and interactive graphical display of data (especially geographic data), and 5) an appropriate and systematic design and analysis method for harvest plan design.

Changes in environmental requirements, harvest equipment, forest products markets, etc. have required a greater amount of analysis to be incorporated into the plan (Davis and Reisinger, 1990). Designing Harvest units for rugged mountainous terrain where cable harvest systems must be used is one area where greater analytic requirements were recognized and addressed. Twito and McGaughey review the evolution of skyline harvest planning from the "chain and board method" of Lysons and Mann through to computer programs based on this technique such as LOGGERPC (Twito and McGaughey, 1984). These early methods relied on contour maps or field deflection line surveys for profile input that was then subject to analysis. With the development of digital elevation models (DEM) the potential for computer analysis was greatly increased. PLANS (Preliminary Logging ANalysis System) a computer assisted engineering (CAE) program that has a module that generates multiple profiles for a given landing was developed to use DEM technology and improve the analytical abilities of timber harvest planning (Twito and McGaughey, 1984, McGaughey, 1992). For discussion of digital elevation models and their application to harvest planning see McGaughey and Twito, 1988; or Christie and McNeel, 1993.

1.3. Planning Level Integration

Several authors recognize the problems of translating forest-level management plans into stand-level operational plans (Baltic et al., 1989; Schiess et al., 1988; Griess, 1984). Problems arise when forest-level plans, which aggregate inventory types for analysis, produce target harvests that are spatially infeasible (Baltic et al., 1989). The adjacency problem is particularly common when multiple resource goals are targeted (Iverson and Alston, 1986). Attempting to formulate additional operational parameters into forest-level planning in order to avoid the spatial feasibility problem results in large unwieldy models (Kent et al., 1991).

The need to integrate site-specific analysis with forest-level allocation can be addressed in part by considering an additional planning level. Breaking planning activities down into various smaller planning levels is a common procedure. Various terms and definitions have been used for the planning level breakdown in the forest management literature, i.e.: macro, sector and project, with divisions primarily based on the economic level effected by the plan (FAO, 1974); provincial, regional, forest management, local resource use, and operational, divided by areas and sub-areas (MacDonald, 1992); area, region and division, broken down by function (policy, coordination and management) (Smith and Prisley, 1989). The terminology that will be used in this discussion is: strategic, tactical and operational (After: Dewhurst et al., 1993; Kent et al., 1990; Nelson and Brodie, 1990; Sedjo, 1987; and others). Breaking the problem down effectively allows for a planning process to begin with the large scale, broad issue aspects of planning (such as goal definition and land allocation) and progress through an iterative process to where these broad management concerns are addressed in on-the-ground management activities (i.e. actual road construction and timber harvest).

1.4. Computer Tools

To ensure that the tactical planning level is effective in transferring strategic goals to operational activities, extensive use has been made of computer tools and methods. Managing large volumes of information is integral to tactical planning and the ability of computers to manage information systems has long been recognized (Dippel and House, 1969). The high cost of computer technology and the resulting complexity of user interfaces originally limited the utility of computer systems, but with reductions in cost came improvements in interfaces that resulted in increased use and productivity. Because of the complexity of forest resource management the quantity of information required for planning exceeds the capacity of a manual system to effectively produce relevant information for decision making (Levinsohn and Brown, 1991). Therefore the development of computer based Geographic Information Systems (GIS) with their abilities to process spatial information has been grasped as an important tool for forest land management (Heit and Shortreid, 1991).

Other computer tools developed for tactical planning include decision support systems and harvest scheduling models. Examples of decision support systems are FORMAN2000 (Jordan and Wightman 1993) and TEAMS (Dewhurst et al., 1993). These models are typically formulated for specific management goals and are designed as integral parts of tactical planning systems. Harvest scheduling models are tools that recognize the importance of spatial relationships among harvesting areas. The complexity of solving spatially constrained harvest scheduling problems has led to the use of heuristic algorithms (Hood, 1991; Torres-Rojo and Broidie, 1990; Sessions, 1987). Examples of computer programs for harvest scheduling are NETWORK (Sessions, 1986) and SNAP (Sessions and Sessions, 1988).

1.5. Computer Data Transfer

Information needs to be effectively passed from one level of planning to another; and as computer systems are being used at all levels of planning; the ability to transfer information electronically is a possible way to increase the efficiency of the entire planning process (Smith and Prisley, 1989). Data exchange whether between planning levels or between software within a level can be accomplished by: manual file manipulation and rekeying of data; text file capture, manipulation, and rekeying; or complete data file manipulation with transfer to application (Bodner 1987). Of these methods the last is preferable since it involves no manual editing of data which can introduce human error into the process. Experience with computer aided engineering (CAE) shows that the automated exchange of data results in better designs since the engineer can be concerned with the design rather than with the process of designing (Fridley and Schiess, 1990).

There are two basic alternatives to handling data transfer in an automated manner, standardizing data formats so that no data conversion is necessary, or developing conversion mechanisms (Tychon and Johnson, 1990). Standardized methods of developing software to ensure compatibility of data structures (Taylor, 1992) and procedures for documenting software to allow for modification have been developed; but it has been the use of the ASCII standard that has contributed the most to allowing for electronic data transfer (Bodner, 1987).

2 The University of Washington Timber Harvest Planning System

2.1. Overview

The University of Washington Forest Engineering department in co-operation with the Washington State Department of Natural Resources (DNR) undertook the challenge of timber harvest planning at a tactical level by using a systems approach (Cullen and Schiess 1992). The system described here has been used by the University of Washington senior spring field studies class, which is part of the Forest Engineering curriculum at the University of Washington (Fridley and Schiess 1989, Schiess and Fridley 1993), since 1992. The system is intended to be used when the objectives for a given planning area include timber harvest and a tactical plan is required.. The UWTHPS system does not implicitly require the use of computer planning tools; but as pointed out in Schiess et al. (1988): "because of the complex nature of setting analysis traditional methods of timber harvest setting design rarely utilized sound engineering planning principles; computer tools speed up and simplify the process thereby resulting in improved setting design". Figure 1 illustrates a general flow chart of the harvest planning procedure and was taken from Cullen and Schiess (1992).

The UWTHPS is a tactical planning procedure where the beginning of the project involves the input of management goals for the planning area being considered and the output is a feasible forest harvest plan that can be used as a basis for operational planning.

Figure 1. Timber harvest planning project procedures as described by Cullen and Schiess (1993).

2.2. A Simplified Overview of UWTHPS Process

To summarize the process described by Schiess and Fridley (1993), and Cullen and Schiess (1992) the planning system can be viewed as encompassing six major steps; Data Collection, Initial Analysis, Design of Harvest Settings and Access, Harvest Scheduling, Field Verification, and Plan Review/Modification (see figure 2). Each of these steps typically occur sequentially i.e. you can't perform a harvest schedule until you have designed the harvest units, however each step may be returned to in an iterative manner and some concurrent activities can be performed. The programs utilized in the UWTHPS are illustrated in figure 3.

Figure 2. The simplified six step harvest planning process

2.3. Technology Configuration and Specific Components

As figures 1 and 2 illustrate the UWTHPS has both field and office components. The office components rely on computer programs which are illustrated in their technology configuration in figure 3. The computer tools described here can not replace knowledgeable engineers and planners, they are simply tools that provide the professional with a method to improve and quantify the quality of the plan. The computer based process allows for the use of different programs to accomplish any of the various activities, for example using the SCHEDULER/NETWORK (Schiess et al. 1991; Sessions, 1986) programs to perform harvest scheduling and transportation analysis or using SNAP (Sessions 1988) which performs both functions. The approach provides the user with the ability to mix and match available programs in order to follow company or agency requirements and to change the system when improved tools become available

A computerized Geographic Information System (GIS) is at the core of the software configuration of UWTHPS. The specific GIS used by the University of Washington is P.C. ARC INFO (ESRI ltd.). Modern computerized GIS has provided the planner with greater analytic abilities and efficiencies enabling a higher quality plan with both environmental and economical savings. The GIS is used to manage data from multiple sources allowing for storage, extraction and presentation of information. It is the use of GIS

Figure 3. The technology configuration of the University of WashingtonTimber Harvest Planning System.

that allows for the modularity of the UWTHPS since data from various sources can be integrated within the GIS for analysis, or data can be extracted for use in other programs. Apart from it's data management role the GIS is also used for map production and data processing in order to produce the hard copy of the produced plan.

2.4. The PLANS Setting Design Process

The role of PLANS within the UWTHPS is the design of harvest settings. The process of designing settings within PLANS can be isolated from the UWTHPS and considered as the process depicted in figure 4. The figure illustrates the information flow of the setting design process indicating those processes that occur within PLANS (map registration, terrain evaluation and setting design) and implying the field reconnaissance phase and it's role in modifying the design. A detailed description is given in Cullen and Schiess (1992).

Figure 4. The cable setting analysis process utilizing PLANS

2.5. Data Transfer Requirements to Improve UWTHPS

Data transfer within the UWTHPS is crucial to the systems success. Data must be moved both into and out of the individual programs for each component of the system to perform it's role (O'Brien, 1994) The central issue of this study was how to transfer data from one component program to another. The automated data transfer methods were limited to DOS personal computer based software including: ARC INFO Simple Macro Language (SML) macros (ESRI Ltd.), ASCII editors, DOS commands (Microsoft Ltd.), DOS executable programs including those developed in AWK (Aho, et al. 1988), and Borland C (Borland International Inc.).

The movements of data within the UWTHPS are graphically represented by figure 3. The problem now is one of developing the appropriate data interfaces or data exchange protocols. Figure 3 is such a representation showing the data flow among the various design packages. The particular design programs such as PLANS (and others) generate numerical as well as spatial solutions which is then displayed using the power of the GIS package.

It was thought that if the data could be transferred to ARC INFO planners could take advantage of ARC INFO's role as the central data manager to provide for additional analysis of the data. Not only could maps be produced illustrating each given profiles features (payload, tailhold height, yarder type, etc.) to assist in boundary rectification or field reconnaissance, but the data could be put to other uses. Examples of how PLANS data may be utilized are: average yarding slope distance, and maximum payloads could be utilized to more accurately predict harvest costs; listings of yarders utilized coupled with harvest scheduling information could chart equipment requirements; and data regarding cable tension could be used to predict areas of greatest concern for worker safety.

3. The Cable Setting Boundary Rectification Problem and the PLANS to ARC Conversion

3.1. An Application of the Cable Setting Rectification Problem

Boundary rectification between settings designed with PLANS is performed utilizing the PLANS setting boundaries and the resource information base maps. The placement of the rectified setting boundary is entirely dependent on the skills of the engineer who must interpret the information from the base maps to determine which profile will be used to harvest which area. While PLANS performs a rigorous analysis of each profile generating and maintaining data including payload, yarder parameters and tailhold height, this information is not readily available to the engineer while performing the boundary rectification. In order to access the payload and tailhold information it is necessary to either run the PLANS program and review each profile in turn, or to peruse PLANS report files. Both of these methods are awkward and have seldom been considered worth the effort. It was thought that transferring the data from PLANS to ARC/INFO would provide the ability to present some of the analytic information graphically and therefore assist the engineer in selecting preferred profiles for boundary rectification. Figures 5, 6, 7 and 8 illustrate the boundary rectification problem where a geometrically convenient solution can result in harvest problems.

Figure 5 shows three PLANS setting boundaries with large overlaps. Selecting a boundary between settings without incorporating information from the PLANS analysis could result in the exclusion of setting 2 and a boundary between settings 1 and 3 as pictured in figure 6. Figure 7 shows three individual cable roads (profiles) that access the same area

Figure 5: Contour map showing three overlapping cable settings requiring a boundary rectification

Figure 6: One possible boundary rectification solution

Figure 7: View of cable profiles which may be color coded by payload capacity and tailhold height, imported from PLANS into ARC for further analysis

Figure 8: Profile view of the three profiles as analyzed by PLANS indicating payload and deflection potential.

each from a separate setting (landing). Figure 8 shows the profile view produced by PLANS of these cable roads. It is evident from reviewing figure 8 that the profile accessing the area from setting 3 has inferior deflection compared to either the profile from setting 1 or 2. If information pertaining to the quality of a given profile could be included in the boundary rectification procedure superior setting plans would result. Producing a plot of each profile would result in thousands of plots for a typical harvest plan which would be cumbersome. A preferable presentation would be a plan view map illustrating which profiles are marginal and which ones are superior. PLANS report files store two pieces of information that are particularly useful in inferring the relative quality of individual profiles: payload and tailhold height. It is axiomatic that the better the deflection the greater the payload that can be carried over the profile. Increasing tailhold height is a method for improving deflection which has the negative result of requiring adequate tailhold trees or the placement of a tailhold tower both of which increase the difficulty of logging a setting and increase cost (Conway 1992). Therefore a superior profile is one with the lowest tailhold and the highest payload. Figure 9 illustrates the improving profile quality axiom.

Figure 9. The relationship between payload and tailhold height with respect to profile quality

3.2. PLANS Data and The ARC INFO Data Format

It was recognized that transferring the PLANS data to ARC INFO would allow for the production of plan view maps illustrating profile quality, which would improve boundary resolution methods. The maps produced for boundary rectification used color coding to present the height of the tailhold and the amount of payload for each profile. When using PLANS the engineer attempts to design settings with overlap so that there are no unreachable areas between settings. Overlap proves troublesome in ARC INFO where polygons are defined in the geometric sense where any overlapping polygons form at least three distinct polygons. Thus rather than importing data as polygon covers line and point features were used. The combination of line and point features allows both point data and line data to be stored in the same cover by creating a cover with both an AAT.DBF and a PAT.DBF.

Some basic objectives were formulated to extract the PLANS data and make it available in ARC INFO: to produce as few covers as possible while still importing all available PLANS data; to create covers with the capability of expressing graphically the data required for boundary resolution according to the procedures developed, and to produce a method that required the minimum amount of user input as possible (O'Brien, 1994).

The method selected creates a maximum of seven covers. The basic covers for all three modules contain profile and tailhold locations and data associated with these features. Each module stores somewhat different data for these features (i.e. SYKMOBIL and SKYTOWER contain payload information while HIGHLEAD does not) thus it was advisable to let each module exist within it's own set of covers. The two additional covers created for each of the SKYTOWER and HIGHLEAD data include one cover that contains the computer generated boundary data and another that contains the user modified boundary and the associated data.

The method requires the use of several programs and associated data format files. The executable program ARCPLANS.EXE forms the core of the conversion, but in order to keep the process as seamless as possible the entire conversion is executed from within ARC INFO (O'Brien, 1994).

3.3. Additional Benefits From the PLANS_ARC Data Transfer

An example plan was selected and a PLANS to ARC INFO data transfer was performed using Getplans.sml (O'Brien, 1994). The example plan is from the 1994 University of Washington Spring Field Studies. The transfer created three covers : the spatial location of the computer generated cable setting boundaries (PLANS); spatial information regarding landing locations and the user defined boundaries; and the spatial information on profiles and tailholds. The latter cover consists of 1261 individual profiles and tailholds for the particular planning area of about 2500 hectares. These profiles were then categorized according to the profile quality as illustrated in figure 9. Table 3 describes the categorization, listing the number of profiles within each of the tailhold height - payload categories.

Table 3. A categorization of profiles from an example plan showing the number of profiles in each tailhold height - payload category

PAYLOAD WEIGHT	NUMBER OF PROFILES BY TAILHOLD HEIGHT CATEGORY(meters)
	6.0	4.5	3.0	< 1.5	Avg. profile length by weight class
	Number of Profiles
< 4000 kg	8	8	36	99	155
4000 kg	11	9	22	547	589
4 - 9000 kg	2	10	33	271	316
> 9000 kg	5	3	11	186	205
Avg. Profile Length by tail hold height	26	30	102	1103	1261

The additional benefits that can be achieved are illustrated by the ability to extend the categorization into process analysis. It was noted that the number of profiles in the "best" category, those with the lowest tailhold height and the highest payload, was rather large. One of the engineers that designed the settings indicated that the large number of preferred profiles was probably a result of the profiles being relatively short (Doug St. John, personal communication). The data imported to ARC INFO allowed a simple test of the engineers hypothesis by listing the lengths for all the profiles within each category and finding the average length. The results of the average profile length summary are shown in table 4.

Table 4. The average length in meters of profiles in each tailhold height - payload category

PAYLOAD WEIGHT	PROFILE LENGTH IN EACH TAILHOLD HEIGHT CATEGORY(meters)
	6.0	4.5	3.0	< 1.5	Avg. profile length by weight class
	Profile lenght (meters)
< 4000 kg	255	215	216	169	187
4000 kg	285	217	235	224	748
4 - 9000 kg	243	218	193	180	182
> 9000 kg	109	184	138	120	121
Avg. Profile Length by tail hold height	239	214	196	191	193

As can be seen from table 4 the hypothesis that the "best" quality profiles would be relatively short is true since the average length of these profiles is 120 m while the average length of all profiles in the plan is 193 m. It is interesting to note that the "worst" profiles have a longer average length (250 m) than the plan average which probably is a result of designing profiles to reach otherwise inaccessible areas. It is also worth noting that profiles with the largest payloads are shorter than average (121 m) while the profiles with the lowest tailholds are about average in length (191 m) indicating that the payload is inversely effected by the length of the profile while tailhold height is independent. Considering the average length of profiles with high tailholds (239 m) once again illustrates the engineers attempting to access hard to reach areas this time by raising the tailhold to improve deflection.

As can be seen from the example the transfer of data from PLANS to ARC INFO results in the ability to perform additional analysis on harvest design data.

4. Discussion and Conclusions.

The University of Washington Timber Harvest Planning process is a successful system for producing tactical level forest harvest plans. Because of it's systematic design and use of Computer Assisted Engineering (CAE) tools, the process produces superior quality plans based on rigorous quantitative analysis. Data transfer via direct electronic means has been and continues to be one area where additional gains in efficiency of the method can be made. The development of data transfer software and methods has already resulted in improvements in efficiency as evidenced by the improved productivity of the students attending Senior Field studies (Schiess and Fridley, 1993).

The improvements made in solving the PLANS setting boundary resolution problem are a direct result of data transfer capability. While there is still a significant portion of hand work to be performed (hand drawing boundaries and digitizing them into ARC INFO), the resulting boundaries are superior since data pertinent to the settings could be displayed to assist the harvest designer select the preferred boundary. The transfer of data from PLANS also results in additional benefits as other uses for PLANS information are developed.

The development of timber harvest planning has taken us from labor intensive hand methods such as the "chain and board" method for analyzing deflection through computer programs that emulated the hand procedures such as the original versions of LOGGER PC (Sessions, 1992). Programs such as PLANS moved the process one step forward toward radically different methods of harvest setting design where the computer and the DTM are standard tools. The process of developing computer programs that first emulate existing methods and later change the method itself is typical of the evolution of computer assisted design methods. The next generation of setting design tools may incorporate a paradigm shift, but it is my suspicion that the next generation will first take one more step away from the pencil and experience toward the computer and quantitative analysis.

5. Bibliography

Aho, Alfred V., Brian W. Kernighan and Peter J. Weinberger. (1988). The AWK Programming Language. Addison-Wesley Publishing Co. Reading Mass

Alexander, T.G. (1989). Timber Management, Traditional Forestry, and Multiple Use Stewardship: The Case of the Intermountain Region, 1950-85. Journal of Forestry 33 (1) : pp 21-34.

Bodner, Michael Simon (1987). Micro to Mainframe Data Interchange. TAB Professional and Reference Books, Blue Ridge Summit PA.

Christie, D.A. and J.F. McNeel (1993). Deflection-line Analysis Using Digital Elevation Models. proceedings GIS'93 Symposium, Vancouver B.C., February 1993.

Conway, Steve. (1982). Logging Practices, Principles of Timber Harvesting Systems (revised edition). Miller Freeman Publications, Inc. San Francisco, CA.

Cullen, J. and P. Schiess (1992). Integrated Computer Aided Timber Harvest Planning System. International Mountain Logging and eighth Pacific Northwest Skyline Symposium, Bellevue, December 14-16, 1992.

Clutter, Jerome L., James C. Fortson, Leon V. Pienaar, Graham H. Brister, Robert L. Bailey. (1983). Timber management : a Quantitative Approach. John Wiley and Sons. New York.

Davis, Kenneth P. (1966) Forest Management: Regulation and Valuation. McGraw-Hill Book Company, New York.

Dewhurst, Stephen M., W.W. Covington and D.B. Wood. (1993). A Flexible Decision Support Architecture for Ecosystem Based Forest Management. GIS'93 Symposium, Vancouver, B.C. February 1993.

Dippel, Gene and William C. House (1969). Information Systems, Data Processing and Evaluation. Scott, Foresman and Company. Glenview Illinois.

Duerr, William A. et al. (1982). Forest Resource Management Decision-Making Principles and Cases. Oregon State University Book Stores inc. Corvallis, Oregon.

Food and Agriculture Organization of the United Nations. (1974). An Introduction to Planning Forestry Development. Rome.

Heit, Michael and Art Shortreid eds. (1991). GIS Applications in Natural Resources. GIS World, Inc. Fort Collins, Colorado.

Hood, Steven L. (1991). A Comparison of Methods for Scheduling Timber Harvests and Road Construction Subject to Spatial Constraints. Master Thesis, University of Washington, College of Forest Resources.

Iverson, David C. and Richard M. Alston. (1986). The Genesis of FORPLAN: A Historical and Analytical Review of Forest Service Planning Models. USDA Forest Service General Technical Report INT-214. Intermountain Research Station. September, 1986.

Kent, Brian. et al. (1990). Natural Resource Land Management Planning Using Large-Scale Linear Programs: The USDA Forest Service Experience With FORPLAN. Operations Research, Vol. 39. No. 1. January-February.

Kline, Robert F. (1954). Logging Sale Layout. The Timberman. Vol. 55.

Leuschner, W. (1990). Forest Regulation, Harvest Scheduling, and Planning Techniques. Wiley and sons.

Levinsohn, Allan G. and Sandra J. Brown. (1991). GIS and Sustainable Development in Natural Resource Management. in: GIS Applications in Natural Resources. Michael Heit and Art Shortreid eds. GIS World, Inc. Fort Collins, Colorado.

MacDonald, A.J. (1992). Interfacing GIS with Harvest Planning Software. International Mountain Logging and eighth Pacific Northwest Skyline Symposium, Bellevue, December 14-16, 1992.

McGaughey, Robert J. (1992). An overview of PLANS: Preliminary logging analysis system. Proceedings: Workshop on Computer Supported Planning of Roads and Harvesting. Feldafing, Germany, August 26-28, 1992.

Nelson, John. and J.Douglas. Brodie (1990). Comparison of a Random Search Algorithm and Mixed Integer Programming for Solving Area-based Forest Plans. Canadian Journal of Forest Resources, 20: pp 934-942, 1990.

O'Brien, Lawrence M. (1994). Electronic data transfer within the GIS based University of Washington timber harvest planning system. Master's Thesis, Unversity of Washington, College of Forest Resources.

Pearce, J. Kenneth. (1961). Forest Engineering Handbook. USDA, Bureau of Land Management, Salem, Oregon.

Schiess, Peter and James .L. Fridley. (1993). Computer-Aided Timber Harvest Planning and Transportation System Design: A Capstone Design Course. IN: Innovations in Engineering Design Education. The American Society of Mechanical Engineers. Book no. 1003411993.

Schiess, Peter, Indra Gupta and Zhenyan Shen (1991). SCHEDULER, A Computer Program to Create Harvest Schedules. unpublished report, University of Washington Department of Forest Engineering, Seattle, 1991.

Schiess, Peter , Jack Cullen and Steven Brown (1988). Long-Term Timber Sales Planning. in: proceedings: International Mountain Logging and Pacific Northwest Skyline Symposium, Portland, December 12-16, 1988.

Sedjo, R.A. (1987). FORPLAN: An Evaluation of a Forest Planning Tool-summary. In: An Evaluation of a Forest Planning Tool. T.W. Hoekstra, A.A. Dyer and D.C. LeMaster eds. General Technical Report RM-140. Ft. Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Forest and Range Experiment Station.

Sessions, John and J.B. Sessions (1988). SNAP - A Scheduling and Network Analysis Program for Tactical Harvest Planning. in: Proceedings: International Mountain Logging and Pacific North West Skyline Symposium, December 12-16, 1988.

Sessions, John. (1987). A Heuristic Algorithm for the Solution of the Variable and Fixed Cost Transportation Problem. in Proceedings: The 1985 Symposium on System Analysis in Forest Resources. University of Georgia, Athens. pp.324-336, 1987.

Sessions, John. (1986). Network Analysis Using Microcomputers for Transportation Planning. Engineering Field Notes, January-February 1986.

Sedjo, Robert.A. (1987). FORPLAN: An Evaluation of a Forest Planning Tool-summary. In: An Evaluation of a Forest Planning Tool. T.W. Hoekstra, A.A. Dyer and D.C. LeMaster eds. General Technical Report RM-140. Ft. Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Forest and Range Experiment Station.

Silen, Roy R. (1955). More Efficient Road Patterns for a Douglas Fir Drainage. The Timberman. Vol. 56. No. 6. April 1955.

Smith, James L. and Stephen P. Prisley. (1989). The Ultimate Compromise: Situating GIS in the Organizational Hierarchy. in: GIS Applications in Natural Resources. Michael Heit and Art Shortreid eds. GIS World, Inc. Fort Collins, Colorado.

Taylor, Dean L. Computer Assisted Design. (1992). Addison-Wesley Publishing Co.

Torres-Rojo, Juan M. and J. Douglas Brodie. (1990). Adjacency Constraints in Harvest Scheduling: an Aggregation Heuristic. Canadian Journal of Forest Resources 20: 978-986.

Twito, Roger.H., Steven E. Reutebuch, Robert J. McGaughey and Charles N. Mann. (1987). Preliminary Logging Analysis System (PLANS): Overview. USDA Forest Service General Technical Report PNW-GTR-199.

Twito, Roger H. and Robert J. McGaughey. (1984). Computer Aided Timber Harvest Planning: An Example. Mountain Logging Symposium Proceedings, Penn A. Peters and John Luchok. eds.+ June 5-7, 1984. West Virginia University.

Tychon, G.G. and Mark R. Johnson. (1990). GIS Data Exchange: Standards and Formats. in: GIS Applications in Natural Resources. Michael Heit and Art Shortreid eds. GIS World, Inc. Fort Collins, Colorado.

U.S.D.A. Forest Service. (1982). National Forest System Land and Resource Management Planning. Federal Register. Vol. 47. No. 190. pp. 43037-43052.