Service-oriented interface design for open distributed environmental simulations2020/09/13
Journal: 《Environmental Research》
Cite as: Zhang F, Chen M, Yue S, et al. Service-oriented interface design for open distributed environmental simulations[J]. Environmental Research, 2020, 191: 110225.
Abstract: Modeling and simulations are important methods in environmental research. Currently, massive simulation resources from different domains have been developed to simulate various dynamic phenomena and processes to address different environmental problems. These heterogeneous simulation resources (e.g., models, data, and servers) can be wasted if they are not shared and reused effectively. Recently, experts may exchange resources and conduct simulations in the open web environment via these shared and distributed services. However, some challenges remain, such as the heterogeneity and reusability of simulation resources. The goal of this study was to analyze typical scenarios involved in simulation tasks and design a set of service-oriented interfaces for different simulation resources. These interfaces, including the model description interface, model encapsulation interface, server management interface and sim-task operation interface, can be used to describe, encapsulate, manage and invoke environmental simulation resources, which can further contribute to resource assembly for environmental simulation tasks. This study evaluated the case of PM2.5 concentration distribution simulation by meteorological data, land cover data and a random forest model in 2014. Using the designed interface, this study conducted the simulation and explored the influence of different interpolation methods (inverse distance weighting (IDW) and kriging) for meteorological data in the random forest-based PM2.5 concentration simulation. For this case, the results show that kriging is a more suitable interpolation method than IDW for meteorological data in the simulation, and this interface design can organize simulation resources, configure tasks, and balance task loads in different servers on the open web.