Participatory intercomparison strategy for terrestrial carbon cycle models based on a service-oriented architecture2020/05/28
Journal: 《Future Generation Computer Systems》
Cite as: Songshan Yue, Min Chen, Jie Song, Wenping Yuan, Tiexi Chen, Guonian Lü, Chaoran Shen, Zaiyang Ma, Kai Xu, Yongning Wen, Hongquan Song, Participatory intercomparison strategy for terrestrial carbon cycle models based on a service-oriented architecture, Future Generation Computer Systems, Volume 112, 2020, Pages 449-466, ISSN 0167-739X, https://doi.org/10.1016/j.future.2020.05.044.
Abstract: Terrestrial carbon cycle models are important tools for simulating carbon exchange; however, there are still significant uncertainties in the simulation results of different models. Model-data intercomparison has therefore been widely recognized as an effective approach for evaluating model performance and acquiring a more reliable understanding of the terrestrial carbon cycle. Although considerable efforts have been made in establishing model intercomparison projects (MIPs), existing MIPs still experience limitations in supporting teams of researchers working collaboratively online and ensuring the reproducibility of model experiments. This article proposes a participatory intercomparison strategy based on a service-oriented architecture (SOA), which aims to offer a web-based platform for researchers to construct participatory intercomparison (PIC) projects. The three fundamental components of a PIC project are the PIC topic, PIC instance, and PIC task. The PIC topic is used to help participants co-design backgrounds, goals, and comparison protocols. The PIC instance is used to help participants provide models, observations and benchmark data as reusable services. The PIC task is used to help participants formulate comparison workflows and acquire customized comparison results. Using the proposed strategy, a PIC project can be easily created and maintained by a group of geographically distributed participants. Reusability of models and data can be achieved through the proposed service-based wrapping method, and the reproducibility of model-data comparison experiments can be achieved through the workflow-based comparison method. A web platform named “P-MIP” was implemented and tested that helps researchers collaborate online, and a demonstrative PIC project was constructed to verify the feasibility and capability of the proposed strategy.