VIC-Borg: Multiobjective automatic calibration toolkit for VIC model

Abstract

Calibrating hydrological model parameters is crucial for ensuring the models' predictive accuracy, which is fundamental for informed decision-making. We provide a toolkit that can perform multi-objective automatic calibration for the parameters of a variable infiltration capacity (VIC) model, with the purpose of locating robust calibration parameters while lowering the threshold for beginners as much as possible. We mainly use the C language to establish the internal dependency relationship between the streamflow generation and routing process through direct function calls without iterative edits to soil parameter files. The Fortran language-based streamflow routing program, previously run independently, is encapsulated in the form of a shared library and then coupled to the streamflow generation module, thus simplifying the design of the calibration program. Then, we further encapsulate the coupled VIC module into a multi-objective function and embed it into the Borg multi-objective optimization algorithm to achieve loose coupling between the Borg algorithm and the VIC model. A case study shows that VIC-Borg requires no additional effort for environment configuration in the operating system and greatly simplifies the automatic calibration of multi-objective parameter optimization. At the same time, it follows the established convention of configuring the VIC streamflow generation module, routing module, and Borg algorithm, making it easier for beginners to use.

Keywords

• Automatic calibration
• Multiobjective optimization
• VIC model
• Borg algorithm