Utilizing model interoperability and spatial cloud computing to enable the computability of dust storm forecasting.
详细信息
文摘
Both environmental and human challenges, such as deforestation and desertification, require scientifically sound simulations of physical phenomena to better understand the past and to better predict future trends for improved decision support. However, many scientific problems cannot be processed using a single computer and require computing capability from many distributed computers. The problems should be solved by interdisciplinary efforts instead of by a single science community. Using dust storm forecasting as a case study, I investigate how interoperability technologies can facilitate data access service, model input integration, model coupling, and output utilization and dissemination. Additionally, the research will explore how to utilize spatiotemporal patterns of phenomena, models and computing resources to improve the performance of dust storm forecasting. Finally, I adopt and optimize cloud computing platforms through spatiotemporal patterns to enable the computability of dust storm forecasting over a large area with high resolution to support geospatial decision-making. This research eventually reduce the execution time and communication for two heterogeneous models, Eta-8bin and NMM- dust storm models by enabling the interoperable and loosely-coupling execution of the two models.