A primary product of the U.S. Geological Survey Across Trophic Level System Simulation (ATLSS) Program is the set of Spatially Explicit Species Index (SESI) models developed for the Greater Everglades area. These models (eight as of 2002) produce values for habitat suitability ranging from 0.0 to 1.0 for all 111,000 cells of the 500 x 500 m array. These habitat suitability values are typically calculated for every individual year in a 31-year sequence, simulated using inputs from the South Florida Water Management Model (SFWMM) as processed by the ATLSS High Resolution Hydrology Model. Averages can also be computed over any set of years (e.g., wet years, dry years, all 31 years), and over a variety of sub-regions within the total region included. The SESI models are used for relative comparisons between alternative future scenarios, not for producing absolute evaluations of habitat quality.

The current versions of ATLSS SESI models were evaluated and calibrated with historical demographic observations to the maximum extent possible, given data availability and time constraints. Abundance data for many Everglades species are scarce/sporadic, and methods of collection and reporting are often inconsistent over time and space. The degree to which SESI model evaluation was possible has also been limited by delays in availability of calibration water data for the model for recent years, when more complete and consistent species abundance data have been collected. The lack of high-spatial-resolution water data continues to limit evaluation efforts.

As an example of using the best available data in SESI calibration, spatial abundance data were extracted for each group of wading birds from Systematic Reconnaissance Flight (SRF) records as they became available. The 3-level estimates of water depth recorded along with SRF observations (dry, transitional, wet) were used to approximate historical water depths. Abundance counts were summed and demographic trends were graphed over each sub-region. These trend graphs were compared to SESI output, graphed over model sub-regions, for year-to-year trends, and trends in wet, average, and dry periods. These evaluations were all made based upon the SFWMM Calibration/Verification data available through 1995.

There was no opportunity during the rapid cycle of model development to formalize and document these efforts to compare SESI model output to empirical observations and adjust model parameters to reflect historical biological responses to hydrologic parameters such as hydroperiod and water depth. This process is being formalized and improved as additional monitoring data become available and as restoration modifications proceed. The SESI model code is in an object-oriented structure that readily allows for model modification as data collection proceeds.

As with the SESI models, availability of new data has required testing and revisions of the SFWMD Hydrology Model, which produces hydrology data sets used to drive ATLSS scenario evaluations. An expanded set of hydrologic calibration data, providing daily water depths over the model area during the period 1979 to 2000, will become available from the SFWMD. Generation of this data set has required hydrology model modifications and revisions that will result in output that differs from the original calibration hydrology data. Together with recent data documenting species abundances over the model area, this extended and modified set of hydrologic inputs will enable further testing, evaluation, and revision of SESI models. This activity is essential in order to increase the reliability of the relative predictions made by SESI models.

Some approaches to be used in current and further SESI model development, evaluation, and refinement include the following activities.

(1) Extensive model runs have already been performed to evaluate the sensitivity of the models to wet, dry, and typical hydrological patterns as represented in both the F2050 base and the AltD13R scenario. These model runs were accomplished by creating new water data sets by extracting wet, dry, and typical years of water data from the existing water files and recombining them to create multiple water files representing scenarios which were wetter, drier, and more average for each of the F2050 and the AltD13R scenarios than the original F2050 and AltD13R scenarios. The model output for these resultant water files was then summarized for both the entire model region and selected sub-regions and compared to the summarized output for the original water file. Analysis of these results has indicated that the models do indicate the appropriate response to water level differences (index values increase on average across the region for the American Alligator under wetter conditions). These results also indicate a fairly consistent relative pattern of differences for the models (e.g., if the average index value from a model for AltD13R is higher than its value for F2050, this ranking usually holds true for scenarios comprised of wet, dry, and typical water years). An expansion of this methodology is planned to evaluate the sensitivity of scenario rankings to variation of model parameters.

(2) A web-based interface to the SESI models as an extension of the ATLSS Data Viewer will be made available that allows authorized users to make modifications to selected model parameters and execute models on an ATLSS computational server. The resultant output file(s) could then be downloaded to the user's computer and the ATLSS Data Viewer used to compare different model runs and compare abundance data. The ATLSS Data Viewer tool would be invaluable in comparing model output with empirical data, as this tool incorporates many of the spatial summary routines required for such analysis.

The original restoration science concept for south Florida included continuing feedback between modeling, monitoring, and management programs. This must incorporate open lines of communication among modelers, biologists, and managers. Long-range stable funding must be secured for model evaluation, updating, and analysis as new monitoring and calibration data become available. Planned, periodic version updates of SESI models should incorporate information from trends reflected in current monitoring data.

In order for modeling to play an effective role in the restoration process, the following tasks necessary for model development, updating, and effective use over the period of adaptive restoration management in south Florida should be explicitly planned and funded:

Accumulation and timely distribution of monitoring data.

Maintenance of links and dialogues with experts, ensuring that models continue to reflect what is known about modeled species as the knowledge base grows.

Liaison with users of model output, explaining caveats and restrictions, what output represents and how it should be interpreted.

Contact: Louis J. Gross, University of Tennessee, 569 Dabney Hall, Knoxville, TN, 37996-1610; Phone: 865-974-4295, Fax: 865-974-3067, gross@tiem.utk.edu

(This abstract was taken from the Greater Everglades Ecosystem Restoration (GEER) Open File Report 03-54)

Back to Project Homepage