Forest list archive: msg00054

[Prev][Next][Index][Thread]

OOM



>X-Mailer: ELM [version 2.3 PL11]

Here is an abstract of a new model we are working with that uses
object-oriented programming.  It is currently running and development
is continuing. 

Mladenoff, David J.1, George E. Host1, Joel Boeder1 , and Thomas R.
Crow2.  Natural Resources Research Institute, University of
Minnesota, Duluth, MN 55811, and USDA Forest Service, North Central
Forest Experiment Station, Forestry Sciences Laboratory, Rhinelander,
WI.  Modeling forest succession and landscape change at multiple
scales with LANDIS. 

We have developed a computer simulation model of forest landscape
disturbance and succession (LANDIS).  This object-oriented model is
written in C++, is pixel-based, and has a graphical user interface and
analytical capability.  File input and output is in ERDAS GIS format. 
LANDIS is derived very generally from some aspects of LANDSIM (D.
Roberts 199x) a polygon- based fire and succession model for the
central Rocky mts.  that operates on principles of fuzzy set theory
and species vital attributes.  It is also philosophically related to
forest gap models of the JABOWA/FORET type, and attempts to retain the
biological aspects of gap models with the capability to spatially
model larger areas.  The model is driven by a series of tree species
life history and silvical parameters coupled with probability
functions such as disturbance initiation and spread, and species
dispersal and establishment.  The model operates on a ten-year time
step and simulates forest succession semi-quantitatively based on the
presence of species age classes.  Disturbance, management, dispersal,
growth and death subroutines are contained in the model.  Initial
vegetation and landscape site characteristics exist as separate GIS
data coverages.  The model code and parameters are structured to allow
simulation at various scales of resolution depending on landscape
extent and purpose, and relevant temporal scale.  For example, a
landscape could be simulated at 30 m pixel resolution to examine
species dispersal and patch spread on a relatively small (1000s ha)
landscape, or 200 m or 1 km resolution to examine regional, long-term
forest succession and landscape structural change.  We have applied
the model in analyzing a forest landscape in northern Wisconsin.  





[Metla] [Main Index] [Thread Index]

Mail converted by MHonArc 1.1.0