Authors: Patrick J. Clemins, Noah B. Beckage, Panagiotis D. Oikonomou, Scott Turnbull, Asim Zia – University of Vermont
Title: Computational Workflow Design for a Cyanobacterial Harmful Algal Bloom (CyanoHAB) Forecast Skill Elasticity Experiment
Abstract: Forecast skill elasticity experiments analyze a forecast system’s uncertainty resulting from imperfect initial conditions and future forcings. A typical forecast skill elasticity experiment will explore dozens, or even hundreds, of potential initial conditions and future forcings. This is relatively feasible for a quickly executing forecast system, however, the experiment quickly becomes intractable for a complex forecast system which may take hours to execute without utilizing a large-scale parallel computing environment such as a high-performance computing (HPC) cluster. This work presents a computational workflow for a cyanobacterial harmful algal bloom (CyanoHAB) forecast skill elasticity experiment that utilizes standardized, interchangeable data acquisition modules and automation scripts to configure and submit a large number of initial condition and future forcing scenarios in parallel. This modular, distributed workflow was employed by researchers to simulate five years of daily 7-day CyanoHAB forecasts across fourteen scenarios and has laid the groundwork for seamlessly incorporating additional scenarios as more data acquisition and data preparation modules are developed for new data sources and future forcing scenarios.