Illmann, Yannick (2022) Data Flow Analysis of the University of Victoria Earth System Climate Model. (Bachelor thesis), Kiel University, Kiel, 50 pp.

Preview

Text bachelorthesis_YannickIllmann-se.pdf - Published Version Download (2MB) | Preview

Abstract

Most of the earth system model (ESM) simulations implemented in the latest 90s are mainly implemented in the programming language Fortran 90/95. ESMs in general are often aged software applications, containing problematic architecture styles, but they are still evolving over time. To fully understand such data-centric applications every call and data flow relationship needs to be considered. But language characteristics of Fortran make it hard to reconstruct relationships between components or operations.
The OceanDSL project, launched and hosted by the GEOMAR Kiel and Christian-Albrechts-University Kiel, wants to provide solutions addressing the architecture recovery problem, among other challenges. OceanDSL declares a goal, wanting to analyze ESMs and remodulate them to make updates, improvements and maintainability more possible. As a part of OceanDSL this thesis covers an analysis part for identifying interprocedural data flow of an ESM, namely "The University of Victoria Earth System Climate Model" (UVic ESCM), published by the University of Victoria, Canada. The main focus is to implement a tool recovering any flow of data from a former parsed abstract syntax tree (AST) and visualize it by creating different meta-models. Multiple versions of the UVic ESCM are compared and evaluated with metrics such as model complexity. Results show a rising complexity due to added components and data flow connections.

Actions (login required)

View Item