Explore the vast range of titles available.

In this study, we evaluated the performance of machine learning (ML) models (XGBoost) in predicting low‐cloud fraction (LCF), compared to two generations of the community atmospheric model (CAM5 and CAM6) and ERA5 reanalysis data, each having a different cloud scheme. ML models show a substantial enhancement in predicting LCF regarding root mean squared errors and correlation coefficients. The good performance is consistent across the full spectrums of atmospheric stability and large‐scale vertical velocity. Employing an explainable ML approach, we revealed the importance of including the amount of available moisture in ML models for representing spatiotemporal variations in LCF in the midlatitudes. Also, ML models demonstrated marked improvement in capturing the LCF variations during the stratocumulus‐to‐cumulus transition (SCT). This study suggests ML models' great potential to address the longstanding issues of “too few” low clouds and “too rapid” SCT in global climate models. Plain Language Summary Low clouds impose a strong radiative cooling effect on Earth's climate. Predicting low‐cloud fraction (LCF) is, however, challenging in global climate models (GCMs), partly due to some deficiencies in cloud parameterization schemes. Machine learning (ML) models might fill this gap as it is recognized as an efficient, economical, and accurate method to make predictions. In this study, we find that ML models (XGBoost) exhibit superior proficiency in predicting LCF regarding root mean squared errors and correlation coefficients compared to two generations of the community atmospheric model (CAM5 and CAM6) and ERA5 reanalysis data, each having a different cloud scheme. This improvement helps address one important issue of “too few” low clouds in GCMs. Furthermore, ML models demonstrate marked improvement in representing LCF variations when stratocumulus clouds transition to cumulus clouds, as opposed to too rapid decreases in LCF simulated by two CAMs and ERA5. Such findings testify to the unique role of ML models in refining the parameterization of LCF within GCMs. Key Points Machine learning (ML) models substantially improve “too few” low‐cloud problems in the subtropics compared to traditional cloud schemes They also show marked improvement in representing low‐cloud fraction (LCF) variations during the stratocumulus‐to‐cumulus transition Including the effect of moisture source in ML models is crucial to representing spatiotemporal variations in LCF in the midlatitudes

This study implements the unified parameterization (UP) in the Central Weather Bureau Global Forecast System (CWBGFS) based on the relaxed Arakawa–Schubert scheme (RAS) at a horizontal resolution of 15 km. The new cumulus parameterization that incorporates the UP framework is called URAS. The UP generalizes the representation of moist convection between the parameterized and the explicitly resolved processes according to the process-dependent convective updraft fraction ( σ ). Short-term hindcasts are performed to investigate the impacts of the UP on the simulated precipitation variability and organized convective systems over the Maritime Continent when multiple scales of convection occurred. The result shows that σ is generally larger when convective systems develop, which adaptively reduces the parameterized convection and increases the spatial variation of moisture. In the URAS experiment, the moisture hotspots within organized convective systems contribute to the enhanced local circulation and the more significant variability of precipitation. Consequently, the URAS has a more realistic precipitation spectrum, an improved relationship between the maximum precipitation and the horizontal scale of the convective systems, and an improved column water vapor–precipitation relationship.

In this study, a revised prognostic cloud fraction scheme for atmospheric models is proposed and its performance is evaluated with a diagnostic cloud fraction scheme. A revision is proposed through a direct linkage between hydrometeors in the cumulus parameterization scheme and the amount of predicted cloud fractions. Cloud fractions that are determined via the prognostic cloud fraction scheme appear to be more realistic than those determined via a diagnostic cloud fraction scheme when both are compared with satellite data. In a medium-range forecast test bed, the biases of large-scale features such as temperature, geopotential height, and mean sea level pressure are moderately reduced when the prognostic cloud fraction scheme is used.

The recently developed GFDL Atmospheric Model version 3 (AM3), an atmospheric general circulation model (GCM), incorporates a prognostic treatment of cloud drop number to simulate the aerosol indirect effect. Since cloud drop activation depends on cloud-scale vertical velocities, which are not reproduced in present-day GCMs, additional assumptions on the subgrid variability are required to implement a local activation parameterization into a GCM. This paper describes the subgrid activation assumptions in AM3 and explores sensitivities by constructing alternate configurations. These alternate model configurations exhibit only small differences in their present-day climatology. However, the total anthropogenic radiative flux perturbation (RFP) between present-day and preindustrial conditions varies by ±50% from the reference, because of a large difference in the magnitude of the aerosol indirect effect. The spread in RFP does not originate directly from the subgrid assumptions but indirectly through the cloud retuning necessary to maintain a realistic radiation balance. In particular, the paper shows a linear correlation between the choice of autoconversion threshold radius and the RFP. Climate sensitivity changes only minimally between the reference and alternate configurations. If implemented in a fully coupled model, these alternate configurations would therefore likely produce substantially different warming from preindustrial to present day.

In this study, cycled forecast experiments were performed to assess the ability of different cloud microphysics and cumulus parameterization schemes in the Hurricane Weather Research and Forecasting (HWRF) Model to accurately simulate the evolution of the cloud and moisture fields during the entire life cycle of Hurricane Edouard (2014). The forecast accuracy for each model configuration was evaluated through comparison of observed and simulated Geostationary Operational Environmental Satellite-13 (GOES-13) infrared brightness temperatures and satellite-derived tropical cyclone intensity estimates computed using the advanced Dvorak technique (ADT). Overall, the analysis revealed a large moist bias in the mid- and upper troposphere during the entire forecast period that was at least partially due to a moist bias in the initialization datasets but was also affected by the microphysics and cumulus parameterization schemes. Large differences occurred in the azimuthal brightness temperature distributions, with two of the microphysics schemes producing hurricane eyes that were much larger and clearer than observed, especially for later forecast hours. Comparisons to the forecast 10-m wind speeds showed reasonable agreement (correlations between 0.58 and 0.74) between the surface-based intensities and the ADT intensity estimates inferred via cloud patterns in the upper troposphere. It was also found that model configurations that had the smallest differences between the ADT and surface-based intensities had the most accurate track and intensity forecasts. Last, the cloud microphysics schemes had the largest impact on the forecast accuracy.

Analysis of WRF Model output from experiments using two double-moment microphysics schemes is carried out to demonstrate that there can be an inconsistency between the predicted mass and number concentrations when a single-moment convective parameterization is used together with a double-moment microphysics scheme. This inconsistency may arise because the grid-scale and subgrid-scale cloud schemes generally apply different levels of complexity to the parameterized microphysical processes. In particular, when a multimoment formulation is used in the microphysics scheme and other physical parameterizations modify only the mass-related moment while the values of the second (or higher) moment for individual hydrometeors remain unchanged, an unintended modification of the particle size distribution occurs. Simulated radar reflectivity is shown to be a valuable tool in diagnosing this inconsistency. In addition, potential ways to minimize the problem are explored by including number concentration calculations in the cumulus parameterization that are consistent with the assumptions of hydrometeor sizes in the microphysics parameterization. The results of this study indicate that it is physically preferable to unify microphysical assumptions between the grid-resolved and subgrid cloud parameterization schemes in weather and climate simulation models.

Aerosols influence cloud and precipitation development in complex ways due to myriad feedbacks at a variety of scales from individual clouds through entire storm systems. This paper describes the implementation, testing, and results of a newly modified bulk microphysical parameterization with explicit cloud droplet nucleation and ice activation by aerosols. Idealized tests and a high-resolution, convection-permitting, continental-scale, 72-h simulation with five sensitivity experiments showed that increased aerosol number concentration results in more numerous cloud droplets of overall smaller size and delays precipitation development. Furthermore, the smaller droplet sizes cause the expected increased cloud albedo effect and more subtle longwave radiation effects. Although increased aerosols generally hindered the warm-rain processes, regions of mixed-phase clouds were impacted in slightly unexpected ways with more precipitation falling north of a synoptic-scale warm front. Aerosol impacts to regions of light precipitation, less than approximately 2.5 mm h−1, were far greater than impacts to regions with higher precipitation rates. Comparisons of model forecasts with five different aerosol states versus surface precipitation measurements revealed that even a large-scale storm system with nearly a thousand observing locations did not indicate which experiment produced a more correct final forecast, indicating a need for far longer-duration simulations due to the magnitude of both model forecast error and observational uncertainty. Last, since aerosols affect cloud and precipitation phase and amount, there are resulting implications to a variety of end-user applications such as surface sensible weather and aircraft icing.

Numerical cloud models require estimates of the vapor growth rate for ice crystals. Current bulk and bin microphysical parameterizations generally assume that vapor growth is diffusion limited, though some parameterizations include the influence of surface attachment kinetics through a constant deposition coefficient. A parameterization for variable deposition coefficients is provided herein. The parameterization is an explicit function of the ambient ice supersaturation and temperature, and an implicit function of crystal dimensions and pressure. The parameterization is valid for variable surface types including growth by dislocations and growth by step nucleation. Deposition coefficients are predicted for the two primary growth directions of crystals, allowing for the evolution of the primary habits. Comparisons with benchmark calculations of instantaneous mass growth indicate that the parameterization is accurate to within a relative error of 1%. Parcel model simulations using Lagrangian microphysics as a benchmark indicate that the bulk parameterization captures the evolution of mass mixing ratio and fall speed with typical relative errors of less than 10%, whereas the average axis lengths can have errors of up to 20%. The bin model produces greater accuracy with relative errors often less than 10%. The deposition coefficient parameterization can be used in any bulk and bin scheme, with low error, if an equivalent volume spherical radius is provided.

A method for the parameterization of ice-phase microphysics is proposed and used to develop a new bulk microphysics scheme. All ice-phase particles are represented by several physical properties that evolve freely in time and space. The scheme prognoses four ice mixing ratio variables, total mass, rime mass, rime volume, and number, allowing 4 degrees of freedom for representing the particle properties using a single category. This approach represents a significant departure from traditional microphysics schemes in which ice-phase hydrometeors are partitioned into various predefined categories (e.g., cloud ice, snow, and graupel) with prescribed characteristics. The liquid-phase component of the new scheme uses a standard two-moment, two-category approach. The proposed method and a complete description of the new predicted particle properties (P3) scheme are provided. Results from idealized model simulations of a two-dimensional squall line are presented that illustrate overall behavior of the scheme. Despite its use of a single ice-phase category, the scheme simulates a realistically wide range of particle characteristics in different regions of the squall line, consistent with observed ice particles in real squall lines. Sensitivity tests show that both the prediction of the rime mass fraction and the rime density are important for the simulation of the squall-line structure and precipitation.

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 the 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 midlatitudes. The work highlights the large sensitivity of cloud feedbacks to both parameter values and structural details in CAM6.