Gaussian Process Emulation for the GR4J Rainfall-Runoff Model
Master's dissertation using Gaussian Process emulation to explore the GR4J rainfall–runoff model efficiently.
My Master’s dissertation focused on using Gaussian Process emulation with the GR4J rainfall–runoff model. The basic idea was to create a statistical approximation of the underlying hydrological model, so that its behaviour could be explored much more efficiently than repeatedly running the full model. The project involved fitting and validating Gaussian Process models and comparing emulator predictions with outputs from GR4J. It was an early opportunity for me to combine statistics, computational modelling and an environmental application, and it later influenced my interest in uncertainty quantification and the use of emulators for larger scientific models.