Applegate River Watershed Forest Simulation Project
APPLEGATE RIVER WATERSHED FOREST SIMULATION PROJECT
Meeting w/ Norm Johnson, Jim Agee, John Sessions, Chris Maguire, Bernhard Bahro, Heidi Wedin
OSU College of Forestry, November 5 1997, 9:00 AM-Noon
Notes taken by Todd Kesterson, Applegate River Watershed Council ph: 899-9982 (Watershed Council Office),
fax: 899-1256, E-mail: arwc@mind.net
MEETING AGENDA
0. Website present
1. Objectives of conference call
2. Wildlife
3. Conference call (Bahro and Agee only)
4. John Sessions presentation: how wildlife is handled in Umpqua basin project
5. Summarize
6. Assign tasks
7. Lunch w/ Agee, Johnson, Kesterson
0. WEBSITE
changes were presented by Yeon-Su. Changes made to public comment section include: nested comments,
tagged comments (angry, sad, etc.), comments stored under appropriate working group.
Working groups are open to anyone who is interested. Norm emphasized the need to get community input on web
since many people do not have computer access. He expects that primary input will be from agency personnel
and key (outspoken) community members.
*Norm feels that there has been enough public review on goals, but he still wants input from scientists.
1. OBJECTIVES OF CONFERENCE CALL
Much discussion of modeling systems. Farsight (state of the art , GIS based fire modeling system) will be
embedded into project. This is written in C++ and source code is available for modification. This is a rapidly
changing program, so concern over being not being compatible w/ later revs.
Assumptions in Sierras was that fire would not be suppressed. Suppression module can be invoked on Applegate
project to show results is fire is suppressed midway.
Emissions can also be modeled (based on EPA air quality standards). This was based on conversation w/ Mark Coen?
(fire behavior researcher) and another researcher (didn't get name). Norm said he could probably come up w/ $$
to pay Mark, but John wants to see code first.
*Issues of emission and suppression modules were apparently big breakthroughs that impressed Norm..
INSECT AND STAND DISTURBANCE
Possibly simulate forest growth, play disturbance scenario, then simulate forest growth again w/ the results
of disturbance incorporated.
*It is essential to determine the number of forest stands available for simulation.
Climate modeling plays a significant role in insect modeling
2. WILDLIFE-presented by Chris Maguire
Books referenced: 1) "Brown's book", 2) Atlas of Oregon Wildlife Brown divides plant communities into 11 groups
and 6 stand conditions (see handouts).
Breeding, feeding, and living environment are considered for each animal species.
Ranking of stand conditions is as follows: 0= plant comm. doesn't exist for that stand condition. 1= primary site
for any or all requirements of animal (breeding, feeding, and living environment). 2= secondary source for animal requirements.
This ranking was done for all animal species found in Applegate (44 pages total @ 6 animals/ page).
Atlas of Oregon Wildlife was used to predict distribution of animals. Concerns: Can we show special habitat
features such as caves or snags? Is this data available?
The starting point (original data) is most critical, since accuracy of computations depends on it.
How do we determine if species is actually in area rather than just predicted?
Habitat should possibly be represented as POTENTIAL habitat. Model doesn't look at such factors as habitat connectivity.
Should pixels in riparian area have info on moisture, or should this be handled in polys?
How well can animals be modeled?
How small an area can be modeled? John felt that 10 acres was probably workable in model, but Chris didn't feel
that source maps were nearly that accurate.
Jim felt that originally, hardwoods were less dominant in Applegate. After big fire, hardwoods would likely sprout up
to replace conifers and to replace themselves.
*This issue of hardwoods vs. conifers in desired conditions is important
General discussion of data storage:
Slope, aspect and thinning prescription could likely be stored w/ each pixel.
Aprox 200-300 data elements could be stored w/ each polygon
20-40 million pieces of info. can be handled at once, so this has to be divided among pixels and polygons.
3. CONFERENCE CALL (BAHRO AND AGEE ONLY) No discussion followed this call.
4. JOHN SESSIONS PRESENTATION
This model looks at ecological health and economic viability of Umpqua basin. Question: Is there a way to rearrange
checkerboard ownership so that it improves ecological health while improving timber harvest rates? Program looks
at range of polygons based on known home range of individual species. Each adjacent polygon is also considered.
We know very little about many animals, so they may not be able to be modeled accurately. Habitat suitability of
each polygon is measured over time, given existing land pattern vs. other options. Goals of private landowners are incorporated.
Differences between public and private management strategies are also included (harvest cycles, width of riparian zones, etc.)
Results show how simulated solutions deviate from desired condition. Optimum results did not clump LSR areas as many
people expected, but riparian buffers on streams were shifted to federal ownership.
*Applegate model is NOT intended to look at land exchanges. The Umpqua model was only shown as an example of how various
land management strategies could be considered in simulation. John feels that such a modeling process makes options
more visible to the public than would otherwise be possible.
Computer used for presentation: 200 MHz Pentium w/ 64 megs RAM. This model used 6000 polygons and could complete
computation in aprox 6 secs. Applegate will have 50000 polygons and would compute in aprox 1 min. on this machine.
5. SUMMARY
In this model, computation will stop after each iteration, rather than waiting until all computations are complete.
This allows users to see results along the way (ex: every 5 yrs) rather than waiting until a longer cycle (ex: 100 yrs)
has been calculated.
Time scale for calculations will likely be 5 yr increments.
There are some limitations with size of polygons in this model. Goal is to have 50,000 polygons representing 10 acres each.
When calculating, disturbances (fire) will be modeled using these 10 acre blocks, which does not accurately represent
the shape of the disturbance. But it's the best they can do currently and is a vast improvement over any current
modeling method in the Applegate. (By end of year, 1 meter satellite data will be commercially available to replace
current 30 m data, but there was no indication that this could be used on Applegate model.
* Norm wants feedback on possible problems w/ grouping pixels into such large polygons.
6. ASSIGNMENTS
Land Classification-- Norm Johnson
Farsite Learning-- John Sessions and others Growth/ Yield Modeling-- Heidi Wedin, Jim Agee Measures of Goal Attainment-- All
Fuelbreak Paper-- Jim Agee (hopefully available for public review by end of December)