The Impact of Parametric Uncertainties on Biogeochemistry in the E3SM Land Model
A global sensitivity analysis (GSA) was performed using a new methodology on the Energy Exascale Earth System (E3SM) v1 land model that indicates the key parameters driving uncertainty in model projections of carbon and energy-related outputs. The GSA was performed across 96 sites representing a wide range of climatic conditions and vegetation types. A small subset of the parameters analyzed was identified as sensitive, and parameter sensitivities were found to be consistent within a vegetation type.
Understanding parametric sensitivities in land models can aid in prioritization of model development and observations targeted to reduce prediction uncertainties. Uncertainty quantification techniques, which include parameter sensitivity analysis, are an important tool to evaluate uncertainties associated with specific processes (for example, photosynthesis or plant respiration). The methods developed here are also a first step towards formal model calibration at a global scale.
Global sensitivity analysis (GSA) of high-dimensional, nonlinear models such as E3SM using traditional methods requires a large number of simulations and is prohibitively expensive from a computational standpoint. Here, a GSA is conducted by first constructing a surrogate representation of the full E3SM land model using a new method to handle the high-dimensional parameter space with a relatively small number of E3SM land model evaluations. This surrogate model allows for efficient extraction sensitivity information, leading to the identification of insensitive parameters and reducing the dimensionality of the problem. The GSA is performed at 96 FLUXNET sites covering multiple plant functional types (PFTs) and climate conditions for five carbon and energy-related model outputs. About 20 of the model parameters are identified as sensitive with the rest being relatively insensitive across all outputs and PFTs. These sensitivities are dependent on PFT, and are relatively consistent among sites within the same PFT. The five model outputs have a majority of their highly sensitive parameters in common. A common subset of sensitive parameters is also shared among PFTs, but some parameters are specific to certain types (e.g., deciduous phenology).The relative importance of these parameters shifts significantly among PFTs and with climatic variables such as mean annual temperature.