PEcAn
forest and mountain partially covered with fog. Photo: Unsplash
Many of the most pressing questions about global change are not necessarily limited by the need to collect new data as much as by our ability to synthesize existing data. The Predictive Ecosystem Analyzer (PEcAn) seeks to improve this ability.
PRINCIPAL INVESTIGATORS
Michael Dietze (Lead PI)
David LeBauer (Co-PI)
Kenton McHenry (Co-PI)
Ankur Desai (Co-PI)
Staff
Rob Kooper (Applications)
Shawn Serbin (Research Scientist)
Tony Gardella (Project Manager)
students & postdocs
Toni Viskari (Postdoctoral Fellow)
Istem Fer (Postdoctoral Fellow)
Josh Mantooth (Graduate Student)
Betsy Cowdery (Graduate Student)
Alexey Shiklomanov (Graduate Student)
James Simkins (Graduate Student)
Overview:
Develop and promote accessible tools for reproducible ecosystem modeling and forecasting
Feedbacks from the terrestrial biosphere are one of the largest sources of uncertainty in climate change projections (1, 2). Unfortunately, despite a diverse set of modeling approaches, conventional methods to reducing these uncertainties have not progressed rapidly (Figure 1). The slow pace of improvement has occurred despite an unprecedented amount and diversity of data about the terrestrial biosphere, but this data is not being fully utilized to test and improve models. The overarching goal of our project has been to accelerate the pace of model improvement bymaking data and models more accessible. In our earlier Innovation proposal, we noted that a small number of important gaps separate the information we have gathered from the understanding required to improve models and inform policy and management.
The Predictive Ecosystem Analyzer (PEcAn) project has thus far focused on three of these gaps: (1) no single data source provides a complete picture of the terrestrial biosphere, and therefore multiple data sources must be integrated in a sensible manner; (2) current modeling approaches only makes use of a subset of the available data; and (3) this is in large part due to a need for tools to manage the assimilation of data into models. Our Innovation award allowed us to successfully develop and test the database, workflow, and user-friendly application interface that form the PEcAn ecoinformatics toolbox. Here, we seek to build a robust and scalable peer-to-peer network around this toolbox.
Different instances of the PEcAn database operated by different users and modeling teams will sync information across databases and share files. This will allow communication and coordination across the network of models without the need for top-down control over the modeling process, allowing dissemination of tools and data to occur organically and enhancing the long-term sustainability of the code. Users would be able to freely form subnetworks and would control what and when they share with the network. We believe such a network will enable users to spend more time doing science, less time doing redundant data management and building redundant analytical tools, and therefore increase the pace of model improvement and our ability to forecast ecosystem services.
Intellectual Merit
PEcAn is an open-source, modular workflow that manages the flow of information into and out of ecosystem models. By wrapping models in common interface modules that use the same inputs and outputs in a common format, we have created an API that allows all tools to be developed in a general, reusable manner. This greatly reduces the amount of redundant work required by different modeling teams, allowing them to focus more on the science, and allows users to work with multiple models with less of a model-specific learning curve. To date PEcAn has been coupled with four ecosystem models, the Ecosystem Demography (ED) model (7), SipNET (8), BioCro (9), and DALEC (10), but this approach is scalable to the scores of ecosystem models currently used by the community. One of the key goals of our ABI Innovation award was to develop state-variable data assimilation approaches to synthesize different carbon cycle observations. We are currently able to assimilate data from the MODIS satellite and eddy covariance towers (Viskari et al in review) and are presently working on assimilating US Forest Service vegetation inventory data (Viskari et al in prep).
Broader Impacts
To date, funding has supported three postdoctoral fellows and has produced eight publications, two manuscripts in review, sixteen presentations, three undergraduate training workshops, and seven training workshops at the graduate/postdoctoral level. Since the initial public release of the PEcAn system in September 2012, we have released ten software versions and the PEcAn virtual machine has been downloaded 340 times. The code repository is publicly available on Github, where it has been forked (created a separate development repository) by 37 developers who have made over 3000 commits (code revisions). In addition, project information, tutorials, videos, and an interactive demo are all available online. As part of this grant, Dietze had developed and taught a graduate course on Ecological Forecasting and Informatics and is writing a book on this topic under contract with Princeton University Press.
PEcAn is being used successfully in a number of other research projects in order to assimilate data into ecosystem models. These projects create substantial synergy that we will be leveraging in the proposed project. These include development of the database and its web interface (Energy Bioscience Institute, LeBauer), the support for LiDaR and radar remote sensing (NSF EF #1318164, Dietze CoPI), hyperspectral remote sensing (NASA Terrestrial Ecosystems, PI Shawn Serbin [former postdoc on the ABI Innovation award], Dietze CoPI, Desai collaborator), LandSAT, soils, and topography (NSF EF #1241894 Desai and Dietze CoPIs), meteorological downscaling (NSF EF #1241891, Dietze CoPI), and a wide range of uncurated vegetation data sets (NSF DIBBs #1261582, McHenry PI, Dietze CoPI).
Project news & events
Project demo
The PEcAn project is supported by the National Science Foundation (ABI #1062547, ABI #1458021, DIBBS #1261582, ARC #1023477, EF #1318164, EF #1241894, EF #1241891), NASA Terrestrial Ecosystems, the Department of Energy (ARPA-E awards #DE-AR0000594 and DE-AR0000598), the Energy Biosciences Institute, and an Amazon AWS in Education Grant.
Links
recent news
PEcAn 1.6.0 Released
October 2018
This release has a number of bug fixes and added functionality. All can be seen listed here listed here. The VM can be downloaded here).
Major additions include:
A data ingest SHINY app that allows users to upload and register data from their own local machine or from the DataOne.
Preliminary containerized modules of PEcAn using Docker.
Preliminary automated, Daily Forecasting using NOAA GEFS and the Sipnet model.
