Automated Water Supply Model (AWSM): Streamlining and standardizing application of a physically based snow model for water resources and reproducible science. (November 2020)

• Record Type: Journal Article
• Authors: Havens, Scott; Marks, Danny; Sandusky, Micah; Hedrick, Andrew; Johnson, Micah; Robertson, Mark; Trujillo, Ernesto
• Is Part Of: Computers & geosciences. Volume 144(2020)
• Publication Date: 2020-11
• Subjects: Hydrology -- Computational method -- Software engineering -- Data assimilation; Environmental policy -- Periodicals; 550.5
• DOI: 10.1016/j.cageo.2020.104571
• Languages: English
• ISSNs: 0098-3004

Abstract

Abstract: Reproducible science requires a shift in thinking and application for how data, code and analysis are shared. Now, scientists must act more like software engineers to design models and perform analysis that use principles and techniques pioneered by software developers. Creating reproducible models that are easy to use and understand is in the best interest for the snow and hydrology community, enabling studies by other researchers and facilitating technology transfer to operational applications. Here, we present the Automated Water Supply Model (AWSM) that streamlines and standardizes the workflow of a physically based snow model to create fully reproducible model simulations that can be utilized by researchers and operational water resource managers. AWSM orchestrates four core components that historically required significant, ad-hoc modeler interaction to load the input data, spatially interpolate to the modeling domain, run the models and process the outputs. Because AWSM was developed using principles and techniques from software engineering, users can quickly perform reproducible simulations on any operating system, from a laptop to the cloud. The three fully reproducible example case studies showcase the simplicity and flexibility of using AWSM to perform simulations from small research catchments to simulations that aid in real time water management decisions. Highlights: A reproducible scientific framework for a physically based snow model. The Python framework can be used for research or operational applications. Follows software engineering best practices for code versioning and packaging. Three reproducible example case studs providing readers insight to modeling system.