Application of Gaussian process regression to plasma turbulent transport model validation via integrated modelling

Abstract

This paper outlines an approach towards improved rigour in tokamak turbulence\ntransport model validation within integrated modelling. Gaussian process\nregression (GPR) techniques were applied for profile fitting during the\npreparation of integrated modelling simulations allowing for rigourous\nsensitivity tests of prescribed initial and boundary conditions as both fit and\nderivative uncertainties are provided. This was demonstrated by a JETTO\nintegrated modelling simulation of the JET ITER-like-wall H-mode baseline\ndischarge #92436 with the QuaLiKiz quasilinear turbulent transport model, which\nis the subject of extrapolation towards a deuterium-tritium plasma. The\nsimulation simultaneously evaluates the time evolution of heat, particle, and\nmomentum fluxes over $\\sim10$ confinement times, with a simulation boundary\ncondition at $\\rho_{tor} = 0.85$. Routine inclusion of momentum transport\nprediction in multi-channel flux-driven transport modelling is not standard and\nis facilitated here by recent developments within the QuaLiKiz model. Excellent\nagreement was achieved between the fitted and simulated profiles for $n_e$,\n$T_e$, $T_i$, and $\\Omega_{tor}$ within $2\\sigma$, but the simulation\nunderpredicts the mid-radius $T_i$ and overpredicts the core $n_e$ and $T_e$\nprofiles for this discharge. Despite this, it was shown that this approach is\ncapable of deriving reasonable inputs, including derivative quantities, to\ntokamak models from experimental data. Furthermore, multiple figures-of-merit\nwere defined to quantitatively assess the agreement of integrated modelling\npredictions to experimental data within the GPR profile fitting framework.\n