[Paper Review] Perturbing parameters to understand cloud contributions to climate change

Margaret L. Duffy, Brian Medeiros+2 more|arXiv (Cornell University)|Oct 12, 2023

Atmospheric aerosols and clouds4 citations

TL;DR

This study uses a CAM6 perturbed parameter ensemble (PPE) with 206 simulations to assess how 45 atmospheric model parameters influence cloud feedbacks and equilibrium climate sensitivity (ECS). It finds that parametric uncertainty in CAM6 produces cloud feedback spreads comparable to CMIP6 and AMIP ensembles, but the high-cloud altitude feedback is larger than WCRP assessments; parameter tuning between CAM5 and CAM6 does not explain the increased cloud feedback in CAM6, which instead stems from structural model changes, particularly from CAM6.0 to CAM6.3.

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 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.

Motivation & Objective

• To evaluate the sensitivity of cloud feedbacks to 45 atmospheric model parameters in CAM6 using a perturbed parameter ensemble (PPE).
• To assess whether parametric uncertainty in CAM6 mirrors structural uncertainty seen in CMIP6 and AMIP model ensembles.
• To investigate whether the CAM6 PPE can constrain total cloud feedback using mean-state cloud errors and discrepancies from WCRP assessments.
• To determine whether changes in parameter values between CAM5 and CAM6 explain the observed increase in cloud feedback and ECS.
• To isolate the role of structural model changes (e.g., from CAM6.0 to CAM6.3) in altering cloud feedback magnitude.

Proposed method

• Conduct a CAM6 PPE with 206 simulations perturbing 45 parameters across cloud microphysics, convection, and radiation schemes.
• Calculate total and component cloud feedbacks (six components: high-cloud altitude, tropical marine low-cloud, etc.) using top-of-atmosphere (TOA) energy budget and radiative flux anomalies.
• Use the TOA energy balance equation ΔN = ΔF + λΔT to derive feedback parameters, with λ = λ_cloud + λ_noncloud.
• Apply regression analysis to identify parameters most correlated with total and component cloud feedbacks.
• Compare CAM6 PPE results with CMIP6 and AMIP model ensembles and WCRP assessments to evaluate constraint potential.
• Use a simple regression model to quantify the contribution of parameter changes between CAM5 and CAM6 to the observed increase in cloud feedback.

Figure 1: Top-of-atmosphere radiative imbalance in each of the 262 PPE simulations for both the PD (left) and SST4K (right) experiments. The dashed horizontal line goes through the default value and the black lines are 15 W m -2 away from the default. Dot colors correspond to the E ${}_{\textrm{NET}

Experimental results

Research questions

• RQ1 How do individual model parameters influence the total and component cloud feedbacks in CAM6?
• RQ2 To what extent does parametric uncertainty in CAM6 reproduce the spread in cloud feedbacks seen in CMIP6 and AMIP ensembles?
• RQ3 Can the CAM6 PPE be used to constrain total cloud feedback using mean-state cloud errors or discrepancies from WCRP assessments?
• RQ4 Is the increase in cloud feedback from CAM5 to CAM6 attributable to changes in parameter values or structural model changes?
• RQ5 What explains the reduced total cloud feedback in CAM6.3 (PPE) compared to CAM6.0 (used in AMIP)?

Key findings

• The spread in total cloud feedback and its six components across the CAM6 PPE is comparable in magnitude to that observed in CMIP6 and AMIP ensembles, indicating that parametric uncertainty captures a significant portion of structural uncertainty.
• The high-cloud altitude feedback in the CAM6 PPE is substantially larger than the WCRP assessment and values from CMIP6 and AMIP, suggesting potential overestimation in this component.
• The parameters most strongly correlated with the high-cloud altitude feedback are the convective parcel temperature perturbation (ZM_tiedke_add) and the subgrid ice activation scaling (microp_aero_wsubi_scale).
• Despite strong parametric sensitivity, changes in parameter values between CAM5 and CAM6 explain little of the increase in total cloud feedback; structural changes in the model are the primary driver.
• The reduction in total cloud feedback from CAM6.0 to CAM6.3 (from 0.81 to 0.56 W m⁻² K⁻¹) is primarily due to decreased low clouds in the tropics and middle latitudes, not parameter tuning.
• Mean-state cloud errors and discrepancies from WCRP assessments are uncorrelated with total cloud feedback in the CAM6 PPE, unlike in CMIP models, leading to inconclusive results for feedback constraint using these metrics.

Figure 2: Spread in global-mean radiative feedback across CMIP6 models (yellow) and across PPE (blue). From bottom to top are the total radiative feedbacks, the cloud-radiative feedbacks, SW cloud-radiative feedbacks, and LW cloud-radiative feedbacks. The black square over the AMIP models denotes th