Initial performance measures and information related to the
ATLSS Vegetation Succession Model.
by
The Institute for Environmental Modeling
University of Tennessee, Knoxville
Contact:
Introduction
This page contains information describing the Across Trophic Level System
Simulation (ATLSS) Vegetation Succession Model (VSMod), the preliminary runs
of this model, as well as the data produced by these runs.
These data are provided as sample
output of the type available from this model. Our objective in posting these preliminary
results is to provide an opportunity for comments and feedback from those interested in
the model, and to illustrate the types of results obtainable from the model.
The VSMod projects the spatial distribution of vegetation on a yearly
basis. The model can be used to cover any period of time for which
appropriate input and initialization data can be provided. Typically
the model is run for 31 to 36 year time periods for which hydrology
data from the South Florida Water Management Model (SFWMM) are available.
The yearly time step was chosen because we felt that this was the shortest
period of time over which successional changes would occur at the spatial
resolution projected by the model.
The spatial domain of the model includes most of the natural
areas of south Florida including A.R.M. Loxahatchee National Wildlife
Refuge, Water Conservation Areas (WCA) 2 and 3, the Everglades National Park,
most of Big Cypress National Preserve as well as numerous surrounding
natural areas. The spatial domain is divided into a regular grid
where each plot or cell is 500x500 meters in size. This spatial resolution
was chosen to match the resolution of many of the other ATLSS models, including
the ATLSS Spatially Explicit Species Index (SESI) models, that will
eventually use the output of the VSMod as inputs.
The vegetation types used in the VSMod are currently based on the 24 of the
58 natural types present in version 6.6 of the Florida GAP (Fl-GAP) map
(Pearlstine et al. 2002). The 24 types used in the VSMod are those
that have some representation within the spatial domain of the model. The
remaining 34 types have very little ( less than 5 cells ) or no representation within
the spatial domain of the model. Adding any or all of the unused 34 types
is a matter of obtaining appropriate model parameters for each type. Future
model development may include estimating parameters for these types to
allow those types not currently present within the model domain to invade over time.
In addition, the VSMod can be used with any other collection of vegetation
types and vegetation maps for which appropriate model parameters can be
estimated.
To make projections, the model incorporates the effects of
hydrology, fires and nutrients as well as information about how each
of the 24 vegetation types included in the model respond to these
factors. The parameter estimates for the response of vegetation are
documented in two ATLSS reports : Plant Community Parameter Estimates
and Documentation for the Across Trophic Level System Simulation (ATLSS) (Wetzel 2001),
Nutrient and Fire Disturbance and Model Evaluation Documentation for the
Across Trophic Level System Simulation (ATLSS) (Wetzel 2003).
The primary source of hydrology data is the output of the SFWMM. Hydrology
data from this model are post-processed using the ATLSS High Resolution Hydrology (HRH)
model to produce variation at the 500x500 meter resolution within the 2x2 mile
model cells of the SFWMM.
The VSMod output presented here are based on hydrology data from the
Calibration/Verification run of the SFWMM, version 3.7 (CalVer-v3.7).
The CalVer-v3.7 data includes the period from January 1, 1979 to
December 31, 1995.
A description of this data set and the SFWMM are provided at the
South Florida Water Management Model (SFWMM) website.
The primary source for fire data for the vegetation succession model is the ATLSS Fire
model (FMod). This model estimates the spatial pattern of annual area burned by fires within
the region covered by the ATLSS VSMod. The spatial distribution of fires
projected by this model is based on fire history, hydrology and the distribution
of vegetation. A description of the FMod is provided at the ATLSS
FMod website.
Fire data can also be provided to the VSMod in the form of input files. This
feature can be used to run the VSMod using historical fire patterns, or to
evaluate the results of fire management scenarios.
The VSMod output presented here are based on three different fire scenarios,
labeled LowFire, MedFire and HighFire.
Each fire scenario was created using the
ATLSS FMod. The LowFire fire scenario is based on fire model parameters that
represent low probability of fire spread, the MedFire fire scenario is based on
fire model parameters that represent medium probability of fire spread and
the HighFire fire scenario is based on fire model parameters that represent
a high probability of fire spread. The parameterization of the fire model
is incomplete as of this writing, and the parameter values used to represent
low, medium and high probability of spread are best estimates based on the
experience of the ATLSS staff and collaborators.
Nutrient data for the VSMod is produced by the ATLSS Nutrient Model. This is a
spatially explicit model that projects changes in Total Phosphorous (TP) levels
resulting from TP enriched water being discharged into the natural regions
from various water control structures. A basic description of this model is provided
in Wetzel (2003). Nutrient data can also be provided to the VSMod in the form
of input files. As done in the case of input fire data files, nutrient data files can
be used to run the VSMod on historical nutrient patterns or to evaluate
alternative nutrient delivery scenarios.
The vegetation succession model itself is a stochastic cellular automata (SCA) model.
For each of the 500x500 meter cells the model uses a SCA to determine the
transition from one vegetation type to another. The model is a cellular automata
because at any given time a cell is on one of a finite number of states (a state
being a particular vegetation type). The model is stochastic because transitions
between states do not occur deterministically, but occur stochastically
with certain transition probabilities. Typically, models of this sort assume that
the transition probabilities do not change in time and do not vary across space.
The ATLSS VSMod is different in that transition probabilities vary in both
space and time. The transition probabilities depend on the local hydrologic and
fire history, the current level of TP as well as the current vegetation type.
At each time step the effects of these factors are used to update
the transition probabilities and are evaluated at each location across the landscape.
The updated transition probabilities are then used to determine the next state
of each cell across the landscape.
Because the vegetation succession model is stochastic, it must be run several
times to obtain an estimate of the average trend in succession projected
by the model as well as to estimate the variability produced by the
model. For the results presented here 20 replicate runs were used, but the number of replicates can be chosen by the user.
Description of output
For each fire scenario
the results from the replicates are summarized into a set of spatial data layers.
One layer is produced for each vegetation type and each year. For example, for sawgrass
(Fl-GAP type 43), 17 layers are produced, one for each year from 1979 to 1995, inclusive.
Each layer summarizes results across replicates and is composed
of a set of 500x500 meter cells arranged in a raster grid. Each cell contains
a single value that represents the fraction of replicates in which
the particular vegetation type being mapped occupied that cell.
For each fire scenario
the summary layers produced by the VSMod are stored in a set of output files.
These output files provide the same information in three different
formats. Each format is designed to facilitate the analysis of model results
using different tools.
The first file format presents the data as a set of binary sequential (BSQ) files.
In this format, all the yearly layers for a single vegetation type are stored together in
a single file. The filenames for these files have the format,
VSMod{fire scenario}_{index}.bin and
where {fire scenario} is replaced by the fire scenario name,
{index} is replaced by the Fl-GAP index for the vegetation type.
Fire scenario names are LowFire, MedFire and HighFire.
For the low fire scenario, the data file for sawgrass ( which is Fl-GAP type 43) is
VSModLowFire_43.bin.
These data files are for use with the
ATLSS Data Viewer .
The second file format provides the summary layers as a set of standard ASCII Grid files.
In this format, each year and vegetation type is stored as a separate file. The
filenames for these files have the format
VSMod{fire scenario}_{index}_{year}.asc
where {fire scenario},
{index} is replaced by the Fl-GAP index and {year}
is replaced by the year.
For the low fire scenario the data files for sawgrass are :
VSModLowFire_43_1979.asc
VSModLowFire_43_1980.asc
VSModLowFire_43_1981.asc
...
VSModLowFire_43_1994.asc
VSModLowFire_43_1995.asc
These files can be read into any ESRI product for visualization.
The third file format provides the summary layers as a set of standard TIFF image files.
Like the ASCII Grid files, each year and vegetation type is stored as a separate file.
The filenames for these files have format, VSMod{fire scenario}_{index}_{year}.tif, where
{fire scenario} is replaced with the fire scenario name,
{index} is replaced with the vegetation index and {year}
is replaced with the year.
These are graphic image files that can be viewed with a wide range of programs.
Unlike the other two file formats, the frequency data are interpreted as a set of colors. Figure 1. provides the color legend for these images.
The data files have been placed into a set of standard zip files to reduce their
size and gather related files into a single resource.
For the BSQ files each file is placed into its own zip file. The filename
for the zip files is the same as the original BSQ filename, with the .zip file type
extension added to the end of the filename.
For the ASCII Grid and TIFF files, all the files for a vegetation type have
been gathered into a single zip file. For these files the filename format is
VSMod{fire scenario}_{index}.asc.zip for the ASCII Grid files and
VSMod{fire scenario}_{index}.tif.zip for the TIF files, where
the substitution of {fire scenario} and {index}
are the same as those described above.
Rather than displaying all the vegetation types simulated by VSMod,
this page is limited to 4 types: 43 (Sawgrass), 44 (Eleocharis),
45 (Muhly Grass) and 46 (Cat tail). This was done to reduce the
volume of output in this preliminary release of VSMod output.
Even with this reduction in output there are
420 separate files (aggregated into 36 separate zip files).
These four vegetation types were chosen because they
are major constituents of the Everglades ecosystem.
Each type is a defining
member of important natural habitats within the Everglades.
Cat tail is important because of its encroachment into
Sawgrass marshes resulting from the nutrification of parts of the
Everglades. In addition, theses type show
large responses to changes in hydrology and fire over the
1979 to 1995 time frame.
Based on the response to this initial release we
will add additional types to the list provided or
will make the selection user selectable. The capability to
carry out simulations of VSMod using the
ATLSS Model Interface will be provided to authorized
users following a comment period for the preliminary simulations
presented here.
Disclaimer
The version of the vegetation succession model which produced these results has not been evaluated with respect to detailed historical data. Thus the changes in vegetation observed in these files does not represent what the
completed model projections will be when the ATLSS Fire model is parameterized. These sample data files are designed
to provide an example of the output the model will produce. If you have
any questions please feel free to contact:
Figures
Citations
Pearlstine, L. S., S. Smith, et al. (2002). "Assesing State-Wide
Biodiversity in the Florida Gap Analysis Project." Journal of Environmental
Management 66: 127-144.
Wetzel, P. R. (2001). Plant Community Parameter Estimates and Documentation
for the Across Trophic Level System Simulation (ATLSS). Knoxville, TN
37996-1610, The Institute for Environmental Modeling: 58.
Wetzel, P. R. (2002). Nutrient and Fire Disturbance and Model Evalutation
Documentation for the Across Trophc Level System Simulation (ATLSS).
Knoxville, TN 37996-1610, The Institute for Environmental Modeling: 46.
The ATLSS Project at The University of Tennessee is supported by Cooperative
Agreement #1445-CA09-95-0094 with the U.S. Geological Survey and the National Park Service, The University of
Tennessee, and Awards DMS-0010920, DEB-0219269 and IIS-0427471 to the University of Tennessee
from the National Science Foundation.
© 2005 ATLSS TIEM/UTK