A simple emulator that enables interpretation of parameter-output relationships, applied to two climate model PPEs
Authors:
Qingyuan Yang,
Gregory S Elsaesser,
Marcus Van Lier-Walqui,
Trude Eidhammer
Abstract:
We present a new additive method, nicknamed sage for Simplified Additive Gaussian processes Emulator, to emulate climate model Perturbed Parameter Ensembles (PPEs). It estimates the value of a climate model output as the sum of additive terms. Each additive term is the mean of a Gaussian Process, and corresponds to the impact of a parameter or parameter group on the variable of interest. This desi…
▽ More
We present a new additive method, nicknamed sage for Simplified Additive Gaussian processes Emulator, to emulate climate model Perturbed Parameter Ensembles (PPEs). It estimates the value of a climate model output as the sum of additive terms. Each additive term is the mean of a Gaussian Process, and corresponds to the impact of a parameter or parameter group on the variable of interest. This design caters to the sparsity of PPEs which are characterized by limited ensemble members and high dimensionality of the parameter space. sage quantifies the variability explained by different parameters and parameter groups, providing additional insights on the parameter-climate model output relationship. We apply the method to two climate model PPEs and compare it to a fully connected Neural Network. The two methods have comparable performance with both PPEs, but sage provides insights on parameter and parameter group importance as well as diagnostics useful for optimizing PPE design. Insights gained are valid regardless of the emulator method used, and have not been previously addressed. Our work highlights that analyzing the PPE used to train an emulator is different from analyzing data generated from an emulator trained on the PPE, as the former provides more insights on the data structure in the PPE which could help inform the emulator design.
△ Less
Submitted 8 October, 2024; v1 submitted 30 September, 2024;
originally announced October 2024.
Perturbing parameters to understand cloud contributions to climate change
Authors:
Margaret L. Duffy,
Brian Medeiros,
Andrew Gettelman,
Trude Eidhammer
Abstract:
The sensitivity of cloud feedbacks to atmospheric model parameters is evaluated using a CAM6 perturbed parameter ensemble (PPE). The CAM6 PPE perturbs 45 parameters across 262 simulations, 206 of which are used here. The spread in total cloud feedback and its six components across the CAM6 PPE are comparable to the spread across the CMIP6 and AMIP ensembles, indicating that parametric uncertainty…
▽ More
The sensitivity of cloud feedbacks to atmospheric model parameters is evaluated using a CAM6 perturbed parameter ensemble (PPE). The CAM6 PPE perturbs 45 parameters across 262 simulations, 206 of which are used here. The spread in total cloud feedback and its six components across the CAM6 PPE are comparable to the spread across the CMIP6 and AMIP ensembles, indicating that parametric uncertainty mirrors structural uncertainty. However, the high-cloud altitude feedback is generally larger in the CAM6 PPE than WCRP assessment, CMIP6, and AMIP values. We evaluate the influence of each of the 45 parameters on the total cloud feedback and each of the six cloud feedback components. We also explore whether the CAM6 PPE can be used to constrain the total cloud feedback, with inconclusive results. Further, we find that despite the large parametric sensitivity of cloud feedbacks in CAM6, a substantial increase in cloud feedbacks from CAM5 to CAM6 is not a result of changes in parameter values. Notably, the CAM6 PPE is run with a more recent version of CAM6 (CAM6.3) than was used for AMIP (CAM6.0), and has a smaller total cloud feedback (0.56 W m$^{-2}$ K$^{-1}$) as compared to CAM6.0 (0.81 W m$^{-2}$ K$^{-1}$) owing primarily to reductions in low clouds over the tropics and middle latitudes. The work highlights the large sensitivity of cloud feedbacks to both parameter values and structural details in CAM6.
△ Less
Submitted 12 October, 2023;
originally announced October 2023.
