Using a surrogate-assisted Bayesian framework to calibrate the runoff-generation scheme in the Energy Exascale Earth System Model (E3SM) v1

Abstract

Abstract. Runoff is a critical component of the terrestrial water cycle, and Earth system models (ESMs) are essential tools to study its spatiotemporal variability. Runoff schemes in ESMs typically include many parameters so that model calibration is necessary to improve the accuracy of simulated runoff. However, runoff calibration at a global scale is challenging because of the high computational cost and the lack of reliable observational datasets. In this study, we calibrated 11 runoff relevant parameters in the Energy Exascale Earth System Model (E3SM) Land Model (ELM) using a surrogate-assisted Bayesian framework. First, the polynomial chaos expansion machinery with Bayesian compressed sensing is used to construct computationally inexpensive surrogate models for ELM-simulated runoff at 0.5∘ × 0.5∘ for 1991–2010. The error metric between the ELM simulations and the benchmark data is selected to construct the surrogates, which facilitates efficient calibration and avoids the more conventional, but challenging, construction of high-dimensional surrogates for the ELM simulated runoff. Second, the Sobol' index sensitivity analysis is performed using the surrogate models to identify the most sensitive parameters, and our results show that, in most regions, ELM-simulated runoff is strongly sensitive to 3 of the 11 uncertain parameters. Third, a Bayesian method is used to infer the optimal values of the most sensitive parameters using an observation-based global runoff dataset as the benchmark. Our results show that model performance is significantly improved with the inferred parameter values. Although the parametric uncertainty of simulated runoff is reduced after the parameter inference, it remains comparable to the multimodel ensemble uncertainty represented by the global hydrological models in ISMIP2a. Additionally, the annual global runoff trend during the simulation period is not well constrained by the inferred parameter values, suggesting the importance of including parametric uncertainty in future runoff projections.

Closing in on Hydrologic Predictive Accuracy: Combining the Strengths of High‐Fidelity and Physics‐Agnostic Models

“…Further, the uncertainty communicated by the ML model in the form of $$\tilde{\boldsymbol{\varepsilon }}$$ and ε ′ distributions offers the potential to eliminate the need for traditional UQ techniques quantifying (most commonly) PBM's epistemic uncertainty. Specifically, while a PBM modeler may always choose to carry out parameter inference using surrogate models (Dwelle et al., 2019; Xu et al., 2022) to ensure physical realism in model QoI simulations, predictions with best‐fit parameters will always remain inexact because of the multi‐variate nature of error sources. The typical Monte Carlo (or “pushed‐forward”) simulations of posterior distributions of perceived uncertainty sources deals only with pre‐contemplated uncertainties: they are in the PBM structure or its inputs (Sargsyan et al., 2019).…”

Water Cycle Acceleration in Czechia: A Water Budget Approach

“…Although the method does not implement any physically detailed hydrological model, the predicted runoff showed better results than the ensemble mean of 13 global hydrological model simulations when compared to observational references. Furthermore, GRUN has been extensively applied in regions with sparse in-situ measurements (Hu et al, 2021;Xiong et al, 2022;Xu et al, 2022;Mei et al, 2023).…”

Gaussian process regression-based Bayesian optimisation (G-BO) of model parameters - a WRF model case study of southeast Australia heat extremes

“…Bayesian optimisation employs probabilistic methods to account for parameter uncertainties in models Xu et al, 2022;. This approach represents input parameters as probability distributions from which multiple samples are drawn.…”