A Bayesian data assimilation framework for lake 3D hydrodynamic models with a physics-preserving particle filtering method using SPUX-MITgcm v1

Abstract. We present a Bayesian inference for three-dimensional hydrodynamic model of Lake Geneva with stochastic weather forcing and high-frequency observational datasets. This is achieved by coupling package, SPUX, hydrodynamics MITgcm, into single framework, SPUX-MITgcm. To mitigate uncertainty in the atmospheric forcing, we use smoothed particle Markov chain Monte Carlo method, where intermediate state posteriors are resampled accordance their respective likelihoods. improve quantification filter, develop bi-directional long short-term memory (BiLSTM) neural network to estimate lake skin temperature from history bulk predictions data. study analyzes benefit costs such state-of-the-art computationally expensive calibration assimilation method lakes.