The general purpose of Atlas is to forecast how forest management policies will affect timber supply and alter forest landscapes. How much more old growth will there be in 50 years if timber harvesting is decreased by 10%? How much volume of wood can we cut per year if we require 20% of the forest to be older than 200 years? How many large patches of old growth will be around in 150 years if we limit harvest size to less than 40 ha? These are the types of questions Atlas was designed help answer.

Atlas, like any model, has some inputs, some outputs, and an algorithm to create the outputs from the inputs. What is an input? Well, Atlas uses inputs that can be separated into two categories: landscape characteristics, and modeling parameters. The basics of landscape characteristics are: a GIS map, growth and yield curves, and, harvesting units. The GIS map describes the landscape we wish to model, with spatial data on where each individual stand is located, and site data on the characteristics of each stand - specifically, the stand's age, tree species, area, and site index. These values describe the current forest, or at what modelers call year 0. Yield curves describe how the stands will change as they grow. For example, if in ten years (year 10), a stand will have more volume, the yield curve will show an increase in volume over that period. Harvesting units map out where each harvesting treatment will take place. All areas that will eventually be cut are defined before the any modeling is done, whether they will be scheduled for harvest in year 1, or in year 350.

Modeling parameters are values that users enter to control the model; there are two general types of parameters: constraints and objectives. An example of a constraint is the adjacency rule. When adjacency is turned on, harvesting cannot occur next to a harvested block until the harvested block reaches a certain age. The age the block must reach before adjacent cutting is allowed is also a modeling parameter. Three other important modeling parameters are: seral constraints, rotation age, and target harvest. Seral constraints are rules used to enforce a maximum allowable amount of young forest and a minimum required amount of old forest. For each constraint, an age is chosen along with a threshold value, expressed as a percent of the landscape area. For example, 20% and 250 years would be chosen if the user wished to model a landscape that always contains 20% of the landbase in stands older than 250 years. If the constraint is reached, then no more harvesting occurs until at least 20% of the landscape is again 250 years old. Rotation age is the minimum age when a stand can be harvested. The target harvest is the amount of volume that the model will attempt to harvest each year. If the harvest of any of the stands would result in a constraint being violated, then the target harvest cannot be met, and the model will move on to the next year.

Atlas has several harvest priorities available for ranking the order in which stands will be cut. These include "oldest-first", "closest distance to the mill", and a "random sort". (note - it doesn't try successively younger stands to try to break an old seral constraint).

A more technical description is as follows. A complete description is available in the ATLAS Manual {documentation - ATLAS).

Harvest Simulation Algorithm - Without SuperBlocks

FPS-ATLAS uses the following procedure to simulate a harvest scenario when only polygons are cut:

1. Go to the range with the highest harvest priority that is eligible for harvest. If no ranges are eligible, go to step 7, otherwise go to step 2.
2. Within this range go to the access unit that has the highest harvest priority and is eligible for harvest. If no access units are eligible, close this range and go to step 1, otherwise go to step 3.
3. Within this access unit go to the zone that has the highest harvest priority and is eligible for harvest. If no zones are eligible for harvest, close this access unit, and return to step 2, otherwise go to step 4.
4. Within this zone go to the polygon that has the highest harvest priority (can be age, minimum distance, or stand group), and is eligible for harvest (not reserved, meets minimum age requirement, not excluded due to adjacency, etc.). If no polygons are eligible, close this zone, and go to step 3, otherwise go to step 5.
5. Temporarily harvest the polygon and check all constraints that have been applied to relevant cliques, the zone, the access unit, and the range. If successful, go to step 6. If unsuccessful, reject the temporary harvest and pursue the four possibilities: · if the polygon fails a clique constraint, make the polygon ineligible for harvest and return to step 4. · if the polygon fails a zone constraint, close the zone and return to step 3. · if the polygon fails an access unit constraint, close the access unit and return to step 2. · If the polygon fails a range constraint, close the range and return to step 1.
6. Permanently accept the harvest of this polygon. Update attributes and eligibility of polygon, zone, access unit, and range resulting from this harvest. Apply the harvest volume to the periodic harvest target. If the harvest target has been met go to step 7, otherwise return to step 4.
7. Summarize and report harvest statistics for this period. If the planning horizon has been reached, stop, otherwise age the forest by one planning period, update attributes and constraints, and return to step 1.

General Models: Models take inputs, analyze or manipulate them with an algorithm(s), and produce outputs.

Input: An input is any data that is given to the model before the algorithm operates. In ATLAS, inputs are a GIS map, growth and yield curves, harvest units, and modeling parameters.

Output: An output is the data sent to the user by the model. The basic ATLAS outputs include a graph of volume harvested over each year in the planning horizon, GIS maps of where each harvesting event took place, and GIS maps of the age of each stand in each period. More detailed summaries of road locations and lengths, treatment breakdowns, constraint tracking, and other features are also available. See the Technical Documentation and User's Guide for more details.

Algorithm: An algorithm is a method of analyzing or manipulating data. ATLAS employs three algorithms. The first is an "aging" algorithm - as ATLAS moves forward each year into the future, the aging algorithm "grows" each stand's age by one year. The second two are harvesting algorithms. The most common is "oldest first" - this ranks potential harvest units by their age. ATLAS attempts to harvest the oldest stands first, subject to any constraints, when using this algorithm. The other harvesting algorithm is "closest to the mill". This algorithm ranks potential harvest units by their distance from a pre-defined point; the nearest units are listed for harvest first.

Modeling parameters: Modeling parameters are used to control how the algorithm manipulates the inputs. They are used to simulate different levels of control by specifying targets and constraint amounts. ATLAS has many parameters, see the Technical Documentation and User's Guide for details.

GIS: Geographical Information Systems. A computerized mapping system for storing and analyzing spatial data. The GIS used with Atlas has a map comprised of polygons and a database containing information about each polygon. Atlas is not a GIS, but it uses GIS-like tools to analyze harvest policies.

Polygon: a contiguous mapping unit, linked to characteristics about the unit, such as stand age, site index, species type, and volume. The polygon's shape describes a stand's shape and location on the map, while a database contains the stand's characteristics.

Growth and Yield Curve: a graph describing merchantable volume in a stand from regeneration to old growth. There is a different yield curve for each stand type.

Harvest Unit: a contiguous block of land harvested under a single silviculture regime. Harvest units can be groups of polygons, or single stands.

Stand Type: stand types group polygons into categories to reduce the number of yield curves needed. Polygons with similar tree species and site indices are grouped into a stand type and are then "grown" with average values for that stand type.

Site Index: a measure of productivity of a site. Site index is the height of the average of the several largest trees on the site at age 50 years.