A neural network (NN) surrogate of the NASA GISS ModelE atmosphere (version E3) is trained on a perturbed parameter ensemble (PPE) spanning 45 physics parameters and 36 outputs. The NN is leveraged in a Markov Chain Monte Carlo (MCMC) Bayesian parameter inference framework to generate a second posterior constrained ensemble coined a "calibrated physics ensemble," or CPE. The CPE members are characterized by diverse parameter combinations and are, by definition, close to top-of-atmosphere radiative balance, and must broadly agree with numerous hydrologic, energy cycle and radiative forcing metrics simultaneously. Global observations of numerous cloud, environment, and radiation properties (provided by global satellite products) are crucial for CPE generation. The inference framework explicitly accounts for discrepancies (or biases) in satellite products during CPE generation. We demonstrate that product discrepancies strongly impact calibration of important model parameter settings (e.g., convective plume entrainment rates; fall speed for cloud ice). Structural improvements new to E3 are retained across CPE members (e.g., stratocumulus simulation). Notably, the framework improved the simulation of shallow cumulus and Amazon rainfall while not degrading radiation fields, an upgrade that neither default parameters nor Latin Hypercube parameter searching achieved. Analyses of the initial PPE suggested several parameters were unimportant for output variation. However, many "unimportant" parameters were needed for CPE generation, a result that brings to the forefront how parameter importance should be determined in PPEs. From the CPE, two diverse 45-dimensional parameter configurations are retained to generate radiatively-balanced, auto-tuned atmospheres that were used in two E3 submissions to CMIP6.

The persistent inter-model spread in the response of global-mean surface temperature to increased CO2 (known as the "Equilibrium Climate Sensitivity," or "ECS") is a crucial problem across model generations. This work examines the influence of the models' present-day atmospheric circulation climatologies, and the accompanying climatological cloud radiative effects, in explaining that spread. We analyze the Coupled Model Intercomparison Project Phase 6 (CMIP6) models and find that they simulate a more poleward, and thus more realistic, edge of the Hadley cell (HC) in the Southern Hemisphere than the CMIP5 models, although the climatological shortwave cloud radiative effects are similar in the two generations of models. A few CMIP5 models with extreme equatorward biases in the HC edge exhibited high ECS due to strong Southern midlatitude shortwave cloud radiative warming in response to climate change, suggesting an ECS dependence on HC position. We find that such constraint no longer holds for the CMIP6 models, due to the absence of models with extreme HC climatologies. In spite of this, however, the CMIP6 models show an increased spread in ECS, with more models in the high ECS range. In addition, an improved representation of the climatological jet dynamics does not lead to a new emergent constraint in the CMIP6 models either.

Recent analyses of Coupled Model Intercomparison Project phase 6 (CMIP6) models have shown higher climate sensitivities than previously reported, and this increase has been preliminary attributed to the simulation of anomalous Shortwave Cloud Radiative Effect (SWCRE) over the southern midlatitude regions. In this work, we further explore how the seasonal and annual SWCRE over different regions of the globe influence the model climate sensitivities. Our study suggests a significant contribution of SWCRE on climate sensitivities in both northern and southern midlatitudes; and the relationship remains robust across the seasons. Additionally, we assess the underlying physics of the inter-model spread to diagnose model biases. The results will contribute to quantify the severity of the Equilibrium Climate Sensitivity, as simulated by the CMIP6 models.

Previous studies have documented the impact of the Arctic sea ice loss and associated warming on the midlatitude weather and climate, especially the influence of sea ice retreat over the Barents-Kara Sea on the North Atlantic and Europe regions. However, less attention has been given to other geographical locations over the Arctic, and to the linear additivity of the circulation response to regional Arctic sea ice loss and temperature anomalies. Using a simplified dry dynamical core model, we demonstrate that responses to regional Arctic temperature anomalies over the Barents-Kara Sea, Baffin Bay-Davis Strait-Labrador Sea, and East Siberia-Chukchi Sea, separately, cause similar equatorward shift of the tropospheric jet, but different stratospheric polar vortex responses. Furthermore, responses to regional Arctic temperature anomalies are not linearly additive, and the residual resembles a positive Northern Annular Mode-like structure. Additional targeted experiments highlight the stratospheric influence in the non-additivity of the midlatitude tropospheric response.

This study investigates the robustness of the stratospheric pathway in linking the sea ice variability over the Barents-Kara Sea in late autumn and early winter to the mid-latitude circulation in the subsequent winter. Two groups of models from the Coupled Model Intercomparison Project phase 5 (CMIP5) archive, one with a well-resolved stratosphere (high-top models) and the other with a poorly-resolved stratosphere (low-top models) are explored to distinguish the role of the stratospheric pathway. The results show that, collectively, high-top models are able to capture the persistent mid-latitude circulation response in the subsequent winter. The response in low-top models is, however, weaker and not as long-lasting most likely due to lack of stratospheric variability. Analysis of eddy heat flux reveals that stronger vertical wave propagation leads to a stronger response in stratospheric polar vortex in high-top models. In particular, it shows that zonal wave-2 eddy heat flux is crucial in leading to a stronger linear constructive interference with the climatological waves in high-top models. The results find that multi-model ensemble of CMIP5 high-top models is able to capture the prolonged impact of sea ice variability on the mid-latitude circulation and out performs the low-top models in this regard. Our study suggests that the representation of the stratosphere in climate models plays an important role in amplifying and extending the mid-latitude circulation response.

Previous studies have extensively investigated the impact of Arctic sea ice anomalies on the midlatitude circulation and associated surface climate in winter. However, there is an ongoing scientific debate regarding whether and how sea ice retreat results in the observed cold anomaly over the adjacent continents. We present a robust "cold Siberia" pattern in the winter following sea ice loss over the Barents-Kara seas in late autumn in an advanced atmospheric general circulation model, with a well-resolved stratosphere. Additional targeted experiments reveal that the stratospheric response to sea ice forcing is crucial in the development of cold conditions over Siberia, indicating the dominant role of the stratospheric pathway compared with the direct response within the troposphere. In particular, the downward influence of the stratospheric circulation anomaly significantly intensifies the ridge near the Ural Mountains and the trough over East Asia. The persistently intensified ridge and trough favor more frequent cold air outbreaks and colder winters over Siberia. This finding has important implications for improving seasonal climate prediction of midlatitude cold events. The results also suggest that the model performance in representing the stratosphere-troposphere coupling could be an important source of the discrepancy between recent studies.

Rapid melting of sea ice and an increased warming have been observed over the Arctic since 1990s and is expected to continue in future climate projections. Possible linkage between the Arctic sea ice and the Northern Hemisphere mid-latitude circulation has been studied previously but is not yet fully understood. This dissertation investigates the influence of the Arctic on the mid-latitudes and the underlying dynamical mechanisms. Specifically, we hypothesize that the stratosphere and its coupling with the troposphere play an important role in amplifying and extending the mid-latitude circulation response to arctic warming. First, we assess the robustness of the stratospheric pathway in linking the sea ice variability, specifically over the Barents-Kara Sea (BKS), in late autumn and early winter to the mid-latitude circulation in the subsequent winter using an ensemble of global climate model simulations. We analyze two groups of models from the Coupled Model Intercomparison Project phase 5 (CMIP5) archive, one with a well-resolved stratosphere (high-top models) and the other with a poorly-resolved stratosphere (low-top models) to distinguish the role of the stratospheric pathway. It has been found that, collectively, high-top models are able to capture the persistent mid-latitude circulation response in the subsequent winter. The response in low-top models is, however, weaker and not as long-lasting most likely due to lack of stratospheric variability. Diagnosis of eddy heat flux reveals that stronger vertical wave propagation leads to a stronger response in stratospheric polar vortex in high- top models. The results robustly demonstrate that multi-model ensemble of CMIP5 high-top models are able to capture the prolonged impact of sea ice variability on the mid-latitude circulation and outperforms the low-top models in this regard. We further explore the dynamical linkage between the BKS sea ice loss and the Siberian cold anomalies using a comprehensive Atmospheric General Circulation Model (AGCM), with a well-resolved stratosphere, with prescribed sea ice loss over BKS region. Decomposition of dynamic and thermodynamic components suggests a dynamically induced warm Arctic cold Siberia pattern in the winter following sea ice loss over the BKS in late autumn. Specifically, the results show that the meridional component of the horizontal temperature advection, from the Arctic into the Siberia, dominates in driving a cold temperature anomaly. Additionally, we conduct targeted experiments in order to quantitatively measure the role of the stratospheric pathway. We find that the stratosphere plays a critical role in the tropospheric circulation anomaly characterized by an intensified ridge-trough pattern that is attributable for the enhanced meridional temperature advection from the Arctic into the Siberia. Next, we extend our study to investigate the sensitivity to geographical location of Arctic sea ice loss and associated warming in modulating the atmospheric circulation. In particular, we assess the linear additivity of the regional Arctic sea ice loss and Arctic Amplification (AA), using a simplified dry dynamical core model. We find that the responses to regional AA over three key regions of the Arctic, i.e. Barents- Kara Sea, East Siberia-Chukchi sea and Baffin Bay-Labrador Sea, separately, show similar equatorward shift of the tropospheric jet but differences in the stratospheric polar vortex. In addition, responses to regional Arctic Amplification are not linearly additive and the residual resembles a positive Northern Annular Mode-like structure. Additional targeted experiments further diagnose the role of the stratosphere in the non-linearity. It is found that the stratosphere-troposphere coupling plays an important role in driving the non-linear circulation response to regional AA. The findings of our research leads to a systematic understanding of the role of the stratospheric pathway in modulating the mid-latitude circulation response to Arctic sea ice loss and accompanied surface warming. Our study suggests that the representation of the stratosphere in climate models plays an important role in correctly simulating the mid-latitude circulation response and could be accountable for the some of the discrepancies among recent studies. Additionally, the result indicates that studying the regional sea ice loss might not provide the full picture of pan-Arctic sea ice melting and caution the use of regional sea ice to explain the recent trend.