Use of an Uncertainty Quantification Framework to Calibrate the Runoff Generation Scheme in the Energy Exascale Earth System Land Model
Abstract
Land Surface Models (LSMs) are essential tools to study the temporal and spatial variability of runoff, which is critical component in terrestrial water cycle. Runoff schemes in LSMs typically involve many parameters and model calibration is necessary to improve the accuracy of simulated runoff.However, calibration at global scales is challenging because of the high computational cost and the lack of reliable runoff data. In this study, we calibrated the runoff generation scheme of the Energy Exascale Earth System Model (E3SM) Land Model (ELM) using an uncertain quantification framework. Specifically, the Polynomial Chaos Expansion (PCE) machinery is used to construct computationally cheap surrogate models for mimicking the behaviour of ELM in simulating runoff. The surrogate models are then used for sensitivity analysis and parameter inference with an observation-based global runoff benchmark. Our results show the model performance can be improved significantly with the inferred parameters than the default parameters. While the parameter inference process constrains the parametric uncertainty, it is still appreciable, which is comparable to the multi-model ensemble uncertainty. Additionally, the annual global runoff trend during the simulation period is not well constrained with the inferred parameters, suggesting the importance of including parametric uncertainty in future runoff projections.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFM.H55I0840X