Explore the vast range of titles available.

Physics‐informed machine learning (ML) applied to geophysical simulation is developing explosively. Recently, graph neural net and vision transformer architectures have shown 1–7 days global weather forecast skill superior to any conventional model with integration times over 1,000 times faster, but longer simulations rapidly degrade. ML that achieves high skill in both weather and climate applications is a tougher goal. This Commentary was inspired by Arcomano et al. (2023, https://doi.org/10.1029/2022GL102649), who show impressive progress toward that goal using hybrid ML, combining reservoir computing (RC) to a coarse‐grid climate model and coupling to a separate data‐driven RC model that interactively predicts sea‐surface temperature. This opens new horizons; where will the next ML breakthrough come from, and is conventional climate modeling about to be disrupted? Plain Language Summary Many new research groups are making rapid progress in applying diverse machine learning (ML) methodologies to weather forecasting and climate modeling. These new approaches could make simulations that are 1,000x faster than conventional approaches for the same fidelity. One successful approach for weather forecasting has been replacing an entire conventional global atmospheric model with a ML emulator, but so far the spatial patterns of long‐term average temperature and precipitation simulated by this approach develop large biases. An alternate new approach, hybrid reservoir computing, combines the conventional model with a form of ML that remembers the recent atmospheric evolution. It produces better climate simulations, including realistic El‐Nino/La Nina variability, but with much less spatial detail. This opens new horizons; where will the next ML breakthrough come from, and is conventional climate modeling about to be disrupted? Key Points Arcomano et al. (2023, https://doi.org/10.1029/2022GL102649) combined reservoir computing (RC) with a coarse‐grid climate model for data‐driven ocean‐coupled simulations By building long‐term memory into predictions, RC nearly removes climate bias Challenges remain with interpretability and scalability to fine‐scale prediction that new machine learning approaches may soon surmount

A new moist turbulence parameterization is presented and implemented in the Community Atmosphere Model (CAM). It is derived from Grenier and Bretherton but has been heavily modified to improve its numerical stability and efficiency with the long time steps used in climate models. A goal was to provide a more physically realistic treatment of marine stratocumulus-topped boundary layers than in the current CAM. Key features of the scheme include use of moist-conserved variables, an explicit entrainment closure for convective layers, diagnosis of turbulent kinetic energy (TKE) for computation of turbulent diffusivities, an efficient new formulation of TKE transport as a relaxation to layer-mean TKE, and unified treatment of all turbulent layers in each atmospheric column. The scheme is compared with the default turbulence parameterizations in the CAM using three single-column modeling cases, using both operational and high vertical and time resolution. Both schemes performed comparably well on the dry convective boundary layer case. For a stable boundary layer case, the default CAM overdeepens the boundary layer unless its free-tropospheric mixing length is greatly reduced, whereupon the new scheme and default CAM again both perform well at both tested resolutions. A nocturnal stratocumulus case was much better simulated by the new scheme than the default CAM, with much less resolution sensitivity. Global climate simulations with the new scheme in tandem with a new shallow cumulus parameterization are presented in a companion paper.

General circulation models (GCMs) typically have a grid size of 25–200 km. Parametrizations are used to represent diabatic processes such as radiative transfer and cloud microphysics and account for subgrid‐scale motions and variability. Unlike traditional approaches, neural networks (NNs) can readily exploit recent observational data sets and global cloud‐system resolving model (CRM) simulations to learn subgrid variability. This article describes an NN parametrization trained by coarse‐graining a near‐global CRM simulation with a 4‐km horizontal grid spacing. The NN predicts the residual heating and moistening averaged over (160 km)2 grid boxes as a function of the coarse‐resolution fields within the same atmospheric column. This NN is coupled to the dynamical core of a GCM with the same 160‐km resolution. A recent study described how to train such an NN to be stable when coupled to specified time‐evolving advective forcings in a single‐column model, but feedbacks between NN and GCM components cause spatially extended simulations to crash within a few days. Analyzing the linearized response of such an NN reveals that it learns to exploit a strong synchrony between precipitation and the atmospheric state above 10 km. Removing these variables from the NN's inputs stabilizes the coupled simulations, which predict the future state more accurately than a coarse‐resolution simulation without any parametrizations of subgrid‐scale variability, although the mean state slowly drifts. Key Points A neural‐network parametrization is coupled to a general circulation model The neural network is trained by coarse‐graining a near‐global cloud‐system resolving simulation The coupled simulation doubles the forecast accuracy of the general circulation model

This paper provides a description of the integrated representation for the cloud processes in the Community Atmosphere Model, version 5 (CAM5). CAM5 cloud parameterizations add the following unique characteristics to previous versions: 1) a cloud macrophysical structure with horizontally nonoverlapped deep cumulus, shallow cumulus, and stratus in each grid layer, where each of which has its own cloud fraction, and mass and number concentrations for cloud liquid droplets and ice crystals; 2) stratus–radiation–turbulence interactions that allow CAM5 to simulate marine stratocumulus solely from grid-mean relative humidity without relying on a stability-based empirical formula; 3) prognostic treatment of the number concentrations of stratus liquid droplets and ice crystals, with activated aerosols and detrained in-cumulus condensates as the main sources and with evaporation, sedimentation, and precipitation of stratus condensate as the main sinks; and 4) radiatively active cumulus and snow. By imposing consistency between diagnosed stratus fraction and prognosed stratus condensate, unrealistically empty or highly dense stratus is avoided in CAM5. Because of the activation of the prognostic aerosols and the parameterizations of the radiation and stratiform precipitation production as a function of the cloud droplet size, CAM5 simulates various aerosol indirect effects as well as the direct effects: that is, aerosols affect both the radiation budget and the hydrological cycle. Detailed analysis of various simulations indicates that CAM5 improves upon CAM3/CAM4 in global performance as well as in physical formulation. However, several problems are also identified in CAM5, which can be attributed to deficient regional tuning, inconsistency between various physics parameterizations, and incomplete treatment of physics. Efforts are continuing to further improve CAM5.

How clouds respond to warming remains the greatest source of uncertainty in climate projections. Improved computational and observational tools can reduce this uncertainty. Here we discuss the need for research focusing on high-resolution atmosphere models and the representation of clouds and turbulence within them.

Observations in subtropical regions show that stratiform low cloud cover is well correlated with the lower-troposphere stability (LTS), defined as the difference in potential temperatureθbetween the 700-hPa level and the surface. The LTS can be regarded as a measure of the strength of the inversion that caps the planetary boundary layer (PBL). A stronger inversion is more effective at trapping moisture within the marine boundary layer (MBL), permitting greater cloud cover. This paper presents a new formulation, called theestimated inversion strength(EIS), to estimate the strength of the PBL inversion given the temperatures at 700 hPa and at the surface. The EIS accounts for the general observation that the free-tropospheric temperature profile is often close to a moist adiabat and its lapse rate is strongly temperature dependent. Therefore, for a given LTS, the EIS is greater at colder temperatures. It is demonstrated that while the seasonal cycles of LTS and low cloud cover fraction (CF) are strongly correlated in many regions, no single relationship between LTS and CF can be found that encompasses the wide range of temperatures occurring in the Tropics, subtropics, and midlatitudes. However, a single linear relationship between CF and EIS explains 83% of the regional/seasonal variance in stratus cloud amount, suggesting that EIS is a more regime-independent predictor of stratus cloud amount than is LTS under a wide range of climatological conditions. The result has some potentially important implications for how low clouds might behave in a changed climate. In contrast to Miller’s thermostat hypothesis that a reduction in the lapse rate (Clausius-Clapeyron) will lead to increased LTS and increased tropical low cloud cover in a warmer climate, the results here suggest that low clouds may be much less sensitive to changes in the temperature profile if the vertical profile of tropospheric warming follows a moist adiabat.

A method is proposed for combining information from several emergent constraints into a probabilistic estimate for a climate sensitivity proxy Y such as equilibrium climate sensitivity (ECS). The method is based on fitting a multivariate Gaussian PDF for Y and the emergent constraints using an ensemble of global climate models (GCMs); it can be viewed as a form of multiple linear regression of Y on the constraints. The method accounts for uncertainties in sampling this multidimensional PDF with a small number of models, for observational uncertainties in the constraints, and for overconfidence about the correlation of the constraints with the climate sensitivity. Its general form (Method C) accounts for correlations between the constraints. Method C becomes less robust when some constraints are too strongly related to each other; this can be mitigated using regularization approaches such as ridge regression. An illuminating special case, Method U, neglects any correlations between constraints except through their mutual relationship to the climate proxy; it is more robust to small GCMsample size and is appealingly interpretable. These methods are applied to ECS and the climate feedback parameter using a previously published set of 11 possible emergent constraints derived from climate models in the Coupled Model Intercomparison Project (CMIP). The ±2σ posterior range of ECS for Method C with no overconfidence adjustment is 4.3 ± 0.7 K. For Method U with a large overconfidence adjustment, it is 4.0 ± 1.3 K. This study adds confidence to past findings that most constraints predict higher climate sensitivity than the CMIP mean.

Tropical convection that overshoots the cold point tropopause can impact the climate by directly influencing water vapor, temperatures, and thin cirrus in the upper troposphere‐lower stratosphere (UTLS) region. The distribution of cold point overshoots between land and ocean may help determine how the overshoots will affect the UTLS in a changing climate. Using 4 years of satellite and reanalysis data, we test a brightness temperature proxy calibrated by radar/lidar data to identify cold point‐overshooting convection across the global tropics. We find evidence of cold point‐overshooting convection throughout the tropics, though other cirrus above the cold point cover an area 100 times larger than overshooting tops. Cold point‐overshooting convection occurs 30%–40% more often over convectively active land areas than over the warmest oceans. This proxy can be generalized to evaluate the fidelity of cold point overshoots simulated by storm‐resolving models. Plain Language Summary Extremely deep convection in the tropics that overshoots the cold point, the coldest temperature level between the upper troposphere and lower stratosphere, influences the vertical temperature structure of this region and water vapor in the lower stratosphere, where it acts as a greenhouse gas. Overshooting cloud tops appear “cold” in infrared satellite imagery, so they can be identified from the difference between their brightness temperature and the nearby cold point temperature. We calibrate this brightness temperature proxy using satellite measurements of cloud ice. Cold point overshoots occur almost as often over the warmest oceans as over moist tropical land areas. Overshooting tops comprise only 1% of satellite‐detectable cloud above the cold point, most of which is very thin ice cloud. Our proxy can be used as a real‐world observational test of cold point overshoots simulated by the most realistic global atmospheric models, which resolve individual thunderstorm systems. Key Points We identify likely tropical cold point overshoots using a radar/lidar calibrated cold point‐relative brightness temperature proxy In a 4‐year climatology, cold point overshoots only modestly favor convectively active land areas over the Indo‐Pacific warm pool Thin cirrus above the cold point covers over 100‐fold more tropical area than cold point overshoots

A linear mixed layer model that skillfully reproduces observed surface winds and convergence over the tropical oceans is used to examine the relative influence of boundary layer and free-tropospheric processes on the distribution of climatological surface winds and convergence. The semiempirical model assumes a subcloud-layer momentum force balance between pressure gradients, Coriolis acceleration, linearized friction, and downward momentum mixing, and it utilizes boundary conditions from the 40-yr ECMWF Re-Analysis (ERA-40). Observed pressure gradients are linearly decomposed into boundary layer (defined as the region below 850 hPa) and free-tropospheric components, and the surface winds and convergence associated with these components are computed. Results show that surface zonal winds are predominantly associated with a combination of free-tropospheric pressure gradients and downward momentum mixing, whereas the distribution of convergence is primarily due to boundary layer temperature gradients, which are closely related to SST gradients. The authors conclude that the climatological distribution of boundary layer convergence is primarily a function of the pattern of SST gradients and is better regarded as a cause rather than a consequence of deep convection.

A review of the experimental protocol and motivation for DYAMOND, the first intercomparison project of global storm-resolving models, is presented. Nine models submitted simulation output for a 40-day (1 August–10 September 2016) intercomparison period. Eight of these employed a tiling of the sphere that was uniformly less than 5 km. By resolving the transient dynamics of convective storms in the tropics, global storm-resolving models remove the need to parameterize tropical deep convection, providing a fundamentally more sound representation of the climate system and a more natural link to commensurately high-resolution data from satellite-borne sensors. The models and some basic characteristics of their output are described in more detail, as is the availability and planned use of this output for future scientific study. Tropically and zonally averaged energy budgets, precipitable water distributions, and precipitation from the model ensemble are evaluated, as is their representation of tropical cyclones and the predictability of column water vapor, the latter being important for tropical weather.