SYMP 11-4
Combining heterogeneous data and process understanding using the Predictive Ecosystem Analyzer

Wednesday, August 13, 2014: 9:40 AM
Gardenia, Sheraton Hotel
David S. LeBauer , Energy Biosciences Institute and Department of Plant Biology, University of Illinois, Urbana, IL
Michael Dietze , Earth and Environment, Boston University, Boston, MA
Shawn P. Serbin , Brookhaven National Laboratory, Upton, NY

Ecosystem models express how ecologists understand the mechanisms underlying ecosystem functioning. Using observations to parameterize and assess model performance allows us to test systems-level predictions of interactions between the biosphere and its environment. Although most ecologists can clearly communicate conceptual models as words and images, expressing these concepts as computer code to simulate the processes that generate data is a technical challenge that limits the efficiency with which conceptual models can be tested and revised. 

PEcAn is a suite of modular R functions that supports model-data synthesis. The development of PEcAn demonstrates how the application of 'best practices' in scientific computing has made it easier to generate hypotheses and predictions based on what we already know, and to learn what we don't know by combining simulation of ecological mechanisms with data that we can observe. PEcAn was built to make it easier to evaluate ecological models and data, and has made computationally intensive research more efficient by making it easier for humans to interact with computers through high level R functions and web interfaces. This talk will describe the design elements of PEcAn that allow ecologists to more efficiently conduct research.


A key element of PEcAn's success has been the adoption of best practices of scientific computing, and this talk will describe how collaborative development of modular, interoperable, reproducible, and accessible tools supports integrative ecological research. PEcAn's modularity allows users to execute individual components or the entire workflow from data collection and synthesis through evaluation and development of hypotheses. Modularity also allows multiple ecosystem models, and versions of the same model, to be embedded within an otherwise consistent analysis. Similarly, modularity allows the components of PEcAn to be embedded within other scientific workflows, by providing accessible data management and synthesis tools as well as standardized interface to other software for pre-and post-processing. The collaborative and public software development process documents the exchange and development of ideas, while making assumptions transparent and changeable. Reproducibility and extensibility makes it easier to build on previous research, evaluate our basic understanding of ecosystem functioning as well as make useful ecological predictions. The combined focus on useability and extensibility makes PEcAn effective as a teaching as well as a research tool.