Multi-objective calibration of the Community Land Model Version 5.0 using in-situ observations of water and energy fluxes and variables

Abstract

Abstract. This study evaluate water and energy fluxes and variables in combination with parameter optimization of the state-of-the-art land surface model Community Land Model version 5 (CLM5), using six years of hourly observations of latent heat flux, sensible heat flux, groundwater recharge and soil moisture. The results show that multi-objective calibration in combination with truncated singular value decomposition and Tikhonov regularization is a powerful method to improve the current practice of using look-up tables to define parameter values in land surface models. Furthermore, reliability of the optimized model parameters can be estimated by statistical measures such as identifiability and relative error variance reduction. As in most other eddy covariance studies, closure of the land surface energy balance is not achieved on observation data. However, using direct measurement of turbulent fluxes as target variable, the parameter optimization is capable of matching simulations and observations of latent heat, especially during the summer period, while simulated sensible heat is clearly biased. The fact that CLM5 is not capable of matching sensible heat, not even with advanced parameter optimization of model parameter values, suggests that the lack of energy closure is due to biases in the sensible heat flux. The results from this study contribute to improvements in model characterization of water and energy fluxes. It is underlined that parameter calibration using available observations of hydrologic and energy fluxes and variables is necessary to obtain the optimal parameter set of a land surface model.

Benefit of Multivariate Model Calibration for Different Climatic Regions

“…Multi‐variate calibration evaluates model fluxes and stores more directly. Various studies have highlighted the value of detailed in‐situ observations of snow (Corbari et al., 2022; Kelleher et al., 2017; Sleziak et al., 2020), actual evapotranspiration (Stisen et al., 2018; Széles et al., 2020), soil moisture (Denager et al., 2023; Pleasants et al., 2023; Stisen et al., 2018), overland flow (Széles et al., 2020), or groundwater (Fowler et al., 2020; Kelleher et al., 2017; Pleasants et al., 2023) for improving model internal consistency in experimental catchments. As many catchments lack ground‐based measurements beyond streamflow, remote sensing‐based observations may provide an important and valuable source of information for hydrological data (Ali et al., 2023; Jiang & Wang, 2019) despite the challenges related to their observation uncertainty or spatiotemporal resolution (Beck et al., 2021; Chen et al., 2021; Mortimer et al., 2020; Sun et al., 2018; Zhang et al., 2016).…”

Disentangling the Hydrological and Hydraulic Controls on Streamflow Variability in E3SM V2 – A Case Study in the Pantanal Region

“…TOPMODEL (Beven and Kirkby, 1979;Niu et al, 2005), and Variable Infiltration Capacity (VIC; Liang et al, 1994) are the two most widely used runoff generation parameterizations (Sheng et al, 2017). It has been demonstrated that calibrating relevant parameters in LSMs leads to improved performance in the simulated runoff at site level (Denager et al, 2022), at the watershed scale (Hou et al, 2012;Huang et al, 2013;Liao and Zhuang, 2017), continent scale (Troy et al, 2008;Yang et al, 2019), and global scale (Yang et al, 2021;Xu et al, 2022a). The simulated runoff is then routed as streamflow to the outlet through the river network in RTMs (or river component in ESM).…”