|PORTER, CHERYL - University Of Florida|
|VILLALOBOS, CHRIS - University Of Florida|
|HALZWORTH, DEAN - Commonwealth Scientific And Industrial Research Organisation (CSIRO)|
|NELSON, ROGER - Washington State University|
|ATHANASIADIS, IOANNIS - Democritus University Of Thrace|
|SANDER, JANSSEN - Alterra-Ilri, Netherlands|
|RIPOCHE, DOMNINIQUE - Institut National De La Recherche Agronomique (INRA)|
|CUFI, JULIEN - Institut National De La Recherche Agronomique (INRA)|
|RAES, DIRK - Leuven University|
|ZHANG, MENG - University Of Florida|
|KNAPEN, ROB - Alterra-Ilri, Netherlands|
|SAHAJPAL, RITVIK - University Of Maryland|
|BOOTE, KENNETH - University Of Florida|
|JONES, JAMES - University Of Florida|
Submitted to: Environmental Modelling & Software
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/4/2014
Publication Date: 10/30/2014
Citation: Porter, C.H., Villalobos, C., Halzworth, D., Nelson, R., White, J.W., Athanasiadis, I.N., Sander, J., Ripoche, D., Cufi, J., Raes, D., Zhang, M., Knapen, R., Sahajpal, R., Boote, K., Jones, J.W. 2014. Harmonization and translation of crop modeling data to ensure interoperability. Environmental Modelling & Software. 62:495-508.
Interpretive Summary: To understand how climate uncertainty affects agricultural production, researchers often employ computer simulation models that attempt to describe the physiological process of crop growth. The accuracy of such predictions depends on the quality of the simulation models. The Agricultural Model Intercomparison and Improvement Project (AgMI, www.agmip.org) seeks to improve the capability of ecophysiological and economic models used to describe the potential impacts of climate uncertainty,especially climate change, on agriculture. The project emphasizes comparing results from multiple models, so it is essential that data be managed efficiently across the different models and subsequent analyzes. Achieving such “interoperability” potentially increases research efficiency, allows models to use a greater variety of datasets, and facilitates. AgMIP has achieved this interoperability by establishing a data exchange mechanism with variables defined in accordance with international standards, implementing a flexibly structured data schema to store experimental data, and designing a method to fill gaps in model-required input data. The software tools provide a mechanism to translate harmonized data to the diverse formats required by different models, thus ensuring that all participating models receive equivalent information. Tools were also developed to estimate model inputs not present in a given dataset and to create hypothetical farming or climate scenarios for subsequent simulation. For the first time, researchers are able to use these tools to run a range of different models on a single, harmonized dataset. This allows them to compare the models directly, ultimately leading to improved models and hence better information for stakeholders ranging from producers to national policy makers.
Technical Abstract: The Agricultural Model Intercomparison and Improvement Project (AgMIP, www.agmip.org) seeks to improve the capability of ecophysiological and economic models to describe the potential impacts of climate change on agricultural systems. AgMIP protocols emphasize the use of multiple models; consequently, data harmonization is essential to facilitate interpretation, storage, access, and interoperability of data products. Achieving interoperability and exchange of data over multiple models potentially increases research efficiency, allows models to use a greater variety of datasets, and facilitates comparability and ensembles. This interoperability was achieved by establishing a standardized data exchange mechanism with variables defined in accordance with international standards; implementing a flexibly structured data schema to store experimental data; and designing a method to fill gaps in model-required input data. The harmonized schema and model-specific data translation tools were developed through cooperation and discussions between crop and economics modeling groups participating in AgMIP activities. The tools provide a mechanism to translate harmonized data to various model-ready input formats ensuring that all participating models are provided equivalent information. Tools were also developed which allow researchers to uniformly provide to multiple models any required model inputs not present in the data and to apply hypothetical management regimens and climate scenarios to existing field data. For the first time ever, researchers and modelers are able to use these tools to run an ensemble of models on a single, harmonized dataset. This allows them to compare models directly, leading ultimately to model improvements. Perhaps the most important outcome is the development of a platform that facilitates researcher collaboration from many organizations, across many countries. This would have been very difficult to achieve without the AgMIP data interoperability standards described in this paper.