WDIAS: A Microservices-Based Weather Data Integration and Assimilation System

Abstract

Numerical Weather Models (NWMs) utilize data from diverse sources such as automated weather stations, radars, and satellite images. Such multimodal data need to be transcoded into a NWM compatible format before use. Moreover, the data integration system's response time needs to be relatively low to reduce the time to forecast weather events like hurricanes and flash floods. Furthermore, the resulting data need to be accessed by many researchers and third-party applications. Existing weather data integration systems are based on monolithic or client-server architectures, and are proprietary or closed source. Hence, they are not only expensive to operate in an era of cloud computing, but also challenging to customize for regions with different weather patterns. In this paper, we present a Weather Data Integration and Assimilation System (WDIAS) that uses microservices architecture and container orchestration to achieve high scalability, availability, and low-cost operation. WDIAS provides a modular architecture to integrate data from different sources, enforce data quality controls, export data into different formats, and extend the functionality by adding new modules. Using a synthetic workload and an experimental setup on a public cloud, we demonstrate that WDIAS can handle 300 RPS with relatively low latency.