Creating a model of a particular landscape -- the Everglades, for example -- still involves a bit of labor. Specific details must be gathered on elevation, soil type, availability of sunlight, and other information needed to map the model onto the real world. The process is a bit like stacking different kinds of maps on top of each other. The SME cell models use the information in these maps to make calculations. That is, to calculate photosynthesis, the model needs to "know" the rainfall it will receive every day depending on its latitude and longitude.

To be sure, SME has made spatial modeling easier. But some pitfalls still await spatial modelers as they make the transition from one to many cells. There are things that STELLA models do behind the scenes that a modeler has to know about when using SME, says Bruce Hannon, a geographer at the University of Illinois at Urbana-Champaign and member of the Alliance Environmental Hydrology team. "There's a learning curve on SME, and it's not trivial."

He ought to know. Hannon teaches classes in ecosystem modeling and is a long-time SME user and collaborator with the Institute. Like Maxwell, he wants to erase some of the seams between STELLA and SME. "The nice thing about STELLA is that it gets the programming out of the way," Hannon says. "But the next step -- going from one to many cells -- ought to be truly transparent." One possible way to do that, Maxwell says, is to incorporate a STELLA-like cell modeling tool into SME itself.


 
 
The Everglades Landscape Model. ELM is designed to track and predict the landscape's response to different water management scenarios in southern Florida. It simulates changes to the vegetation, sediment and water nutrients, and hydrology of the area. The frame above shows southern Florida in 1973, including the residential areas, natural vegetation, and farm land.


  

Villa is also gearing up to take the SME concept to the next level with a tool he has dubbed the Integrating Modeling Architecture (IMA). The IMA would allow spatial and non-spatial models to be linked together seamlessly into a high-level model. This will enable researchers to bridge very different kinds of models. For example, a model of the population dynamics of deer (modeling based on individual organisms) could be run within a model of plant growth in a forest (modeling based on a process). The different models could be defined and linked together by dragging and dropping icons on a computer desktop.

In the meantime, Maxwell continues to improve the SME. The latest user interface, written in Java, incorporates more graphic tools. He and Villa are also collaborating with the Maryland Virtual High School program to modify SME for students. Students in the program, funded by the National Science Foundation, have used STELLA to create models of everything from crab population dynamics to the physics at work in a hot cup of coffee. Maxwell and Villa want to provide a simplified version of SME that students can access via the Web. The Maryland Virtual High School is a partner in the Education Outreach and Training Partnership for Advanced Computational Infrastructure (EOT-PACI).

1 2 3 4