The results of recent fire seasons make it clear that fuel treatment decisions need to be an integral part of the land management planning process. The issues facing land managers are enormous and the tasks overwhelming when one considers the large number of acres with fuel buildups, the budgets needed to treat all those acres, and the trade-offs with regard to resource and environmental effects.

Only sound analysis to determine efficient and effective fuel treatment strategies will do the job. Intellectually rigorous but user-friendly decision support systems are the key to developing fuel treatment alternatives and quantifying the trade-offs among those alternatives.

Currently under development, the Multi-resource Analysis and Geographic Information System (MAGIS) is such a decision support system. MAGIS is designed to spatially schedule treatments that effectively meet resource and management objectives and compute trade-offs associated with those treatment schedules. MAGIS also addresses trade-offs regarding access, such as resource effects associated with roads and the effect of available access on the ability to conduct vegetation treatments.

An analytical methodology combining MAGIS with SIMPPLLE (a stochastic vegetation simulation model) provides a powerful approach for:

• predicting the extent and likely location of disturbance processes (such as fire) both in the presence and absence of treatments.
  
  
• developing spatial and temporal treatment scenarios for addressing fuels treatment and other resource objectives.
  
  
• evaluating those scenarios in a manner that captures the combined effects of treatments and disturbances processes.

Additional development and testing, however, is needed for MAGIS to reach its potential as a truly effective decision support system. However, this development has been hampered by the funding uncertainties associated with soft money and the lack of permanent staff to provide continuity in the development process.

Close partnerships will be sought with key National Forest Systems staffs and forest managers for input on the design and capabilities for data input, scenario specification, and display of results. These partnerships are also important for testing and the eventual handoff of training and support responsibilities. Data sets will be developed, tested, and validated for a variety of geographic areas and conditions, beginning in the first year. The modeling approach will be compared with other approaches for analyzing fuel treatment options at landscape scales, and peer review will be sought for various aspects of the modeling approach.

• First year: Testing, debugging, and documentation will be completed for the current features and processes in MAGIS. New processes for quickly and easily moving information between MAGIS and SIMPPLLE will be completed and documented. Data and resource relationships will be developed and tested for a minimum of two new geographical areas in the country.
  
  
• Second year: Improved interfaces will be developed with GIS to: 1) enter all area data by directly accessing data in GIS coverages; 2) interact with GIS for building management scenarios; 3) export results to GIS for display, including the location and timing of treatments, the location and timing of road activities, the importance of specific segments roads for treatment access, and the location and amounts of resources effects from both roads and polygons. Data and resource relationships will be developed and tested for additional geographical areas and a variety of forest conditions.
  
  
• Three to Five years out: A fully documented operating decision support system with user-friendly interfaces with GIS and SIMPPLLE (or other vegetation simulation models). Data and resource relationships will be developed and tested for the forest conditions present in a number of geographic areas in the country. A number of treatment scenarios with trade-offs will be developed for those areas. The modeling approach will be tested and compared with other approaches and the results peer reviewed.

• Six technical approaches for integrating ArcGIS functionality into MAGIS have been investigated and the most promising approaches have been tested by prototyping two GIS interfaces: (1) user selection of exit nodes for the road network, and (2) user specification of the loading nodes - the road locations that tie the treatment unit polygons to the road network for purposes of routine timber haul and other relevant types of traffic. These prototypes have demonstrated the technical feasibility of these interfaces with the database software used in MAGIS. This technology will provide much improved graphical user interfaces.
  
  
• Planning documents have been prepared to combine the development efforts of MAGIS with an effort underway within the Inventory and Monitoring Institute (detached group of the Washington Office in Ft. Collins, CO) to develop replacement software for SNAP (which runs only in DOS). SNAP used a heuristic solver to schedule harvesting and road construction/reconstruction based on approximated least-cost or greatest revenue. There are many similarities in the data requirements and overall objectives between the SNAP replacement and MAGIS. Combining the efforts into developing one software package provides the opportunity for development efficiencies, will ultimately provide a range of options and solvers to users for handling spatial scheduling problems involving land management activities and road access, and provides the Agency the opportunity to combine user and software support efforts in the future. In other words it avoids duplication of software development and support efforts.
  
  
• The computation speed for developing a MAGIS model has been increased by a factor of 20 for large models where stand polygons differ from treatment unit polygons. Logical fuel treatment units do not necessarily follow stand polygon boundaries. The capability was built into MAGIS to compute treatment unit information from cross walking to stand data. However, when this was attempted for a model of the Angeles National Forest, the model building process took nearly three weeks of computer time. This same data set can now be run in 18 hours or less, depending on the processor speed of the computer.
  
  
• Conversion of MAGIS from MicroSoft FoxPro 2.6 to Visual FoxPro 6.0 has continued, as has the debugging process, much of which is connected to this conversion. Bug fixes have involved both screen functionality and computations. It is important to isolate and eliminate these problems to have a reliable decision support system for analyzing fuel treatment options.
  
  
• The recommendations regarding the fire simulation logic that were developed in the contract with Jack Losensky were implemented in SIMPPLLE. Additional logic was added to represent the probability that a fire ignition event will never develop due to weather conditions. The extreme fire spread logic was expanded to represent not only weather front occurrences, but also occurring when a fire event exceeds a given size.
  
  
• Satellite imagery of fire intensity of the year 2000 fires was obtained from the Northern Region. This provides the type of fire that we will try to match the SIMPPLLE simulations against.
  
  
• The user interface for fuel treatments in SIMPPLLE was redesigned. These changes provide a user with not only access to see the logic associated with the treatment, but also the ability to modify the changes that occur as a result of a treatment. These are changes in species composition, size class and structure and density.
  
  
• User editing capabilities for vegetative pathways were added to SIMPPLLE. Prior to these changes any significant modification of the pathway files had to be done outside of SIMPPLLE in a text editor. The files had to be exported, modified and then imported back into the system. With these improvements a user can add/delete states, add disturbance processes to states, and change next states by dragging and dropping connecting lines within the user interface.
  
  
• An efficiency analysis of the Java code that comprises SIMPPLLE was performed. Very significant improvements were made that allow running large landscapes and complex treatment scheduling scenarios on a PC platform. Previously a user had to switch to a Unix platform for simulating areas that approached one million acres in size. Memory, processor speed, and hard disk space needed for executing SIMPPLLE on a PC are now within the range of what is commonly available to resource managers.
  
  
• A validation of SIMPPLLE for realistic modeling of fire events was completed on a 500,000-acre landscape. Three components were addressed: the total acres of fire over a three decade period, the distribution of fires by size class, and the comparison of a single 10,000-acre fire with a simulated fire event started in the same location. These results are very important for the development of user confidence in model predictions.
  
  
• Satellite imagery was used to provide a finer classification of a previously very broad, non-forest map strata on a 500,000-acre landscape. The mapped non-forest strata was reduced by 55 percent, while significant increases were made in shrub and grass map strata as well as low density forested strata. These changes in the non-forest strata also resulted in significant changes in the spatial pattern of plant communities that is very critical to spatially explicit models. Simulations with SIMPPLLE will be made to quantify the impacts on model results.
  
  
• The number of treatment unit polygons that can be handled in MAGIS was increased; it is now feasible to spatially model over 300,000 acres using a standard formulation. This new size-capacity includes an increase in the number of management regime options for treatment unit polygons as well. Spatial analyses of major portions (or all) of a National Forest can now be encompassed by a single MAGIS model.
  
  
• MAGIS has been enhanced to allow the tracking of acres of specific management activities by time period. Previously, MAGIS only tracked broader schedules of activities (formulated as 'management regimes' composed of one or more activities in specified time units). These activity-acre computations allow the user tighter control of solution elements. For example, they can be used to specify target fuel-treatment acreages by period (the solver will select management regimes that meet these fuel-treatment activity-acre targets). In addition, the spatial arrangement of activities selected for a MAGIS solution can be displayed in GIS.
  
  
• A design plus two prototype screens have been developed for the graphical user interfaces for MAGIS using ArcGIS8. This design incorporates two modes for MAGIS: (1) MAGIS Pro, which includes the full formulation and analytical capabilities of MAGIS, and (2) MAGIS Express, a scaled-back version that is much easier for the occasional user to operate. The MAGIS Express design implements the concept of combining the SNAP software replacement and MAGIS (see the second accomplishment above in this report). These mode options will make MAGIS available to a wide variety of potential users for analyzing a variety of spatial aspects of fuel treatments and associated road access questions. It also helps facilitate incorporating other land management planning issues into these analyses.