Cloud Budget Research Articles

Limousine: Blending Learned and Classical Indexes to Self-Design Larger-than-Memory Cloud Storage Engines

We present Limousine, a self-designing key-value storage engine, that can automatically morph to the near-optimal storage engine architecture shape given a workload, a cloud budget, and target performance. At its core, Limousine identifies the fundamental design principles of storage engines as combinations of learned and classical data structures that collaborate through algorithms for data storage and access. By unifying these principles over diverse hardware and three major cloud providers (AWS, GCP, and Azure), Limousine creates a massive design space of quindecillion (1048) storage engine designs the vast majority of which do not exist in literature or industry. Limousine contains a distribution-aware IO model to accurately evaluate any candidate design. Using these models, Limousine searches within the exhaustive design space to construct a navigable continuum of designs connected along a Pareto frontier of cloud cost and performance. If storage engines contain learned components, Limousine also introduces efficient lazy write algorithms to optimize the holistic read-write performance. Once the near-optimal design is decided for the given context, Limousine automatically materializes the corresponding design in Rust code. Using the YCSB benchmark, we demonstrate that storage engines automatically designed and generated by Limousine scale better by up to 3 orders of magnitude when compared with state-of-the-art industry-leading engines such as RocksDB, WiredTiger, FASTER, and Cosine, over diverse workloads, data sets, and cloud budgets.

Steady-State Supersaturation Distributions for Clouds under Turbulent Forcing

Abstract The supersaturation equation for a vertically moving adiabatic cloud parcel is analyzed. The effects of turbulent updrafts are incorporated in the shape of a stochastic Lagrangian model, with spatial and time correlations expressed in terms of turbulent kinetic energy. Using the Fokker–Planck equation, the steady-state probability distributions of supersaturation are analytically computed for a number of approximations involving the time-scale separation between updraft fluctuations and phase relaxation, and droplet or ice particle size fluctuations. While the analytical results are presented in general for single-phase clouds, the calculated distributions are used to compute mixed-phase cloud properties—mixed fraction and mean liquid water content in an initially icy cloud—and are argued to be useful for generalizing and constructing new parameterization schemes. Significance Statement Supersaturation is the fuel for the development of clouds in the atmosphere. In this paper, our goal is to better understand the supersaturation budget of clouds embedded in a turbulent environment by analyzing the basic equations of cloud microphysics. It is found that the turbulent characteristics of an air parcel substantially affect the cloud’s supersaturation budget and hence its life cycle. This is also shown in the context of mixed-phase clouds where, depending on the turbulent regime, different liquid-to-ice ratios are found. Consequently, the theoretical approach of this paper is crucial to develop tools to parameterize small-scale atmospheric features, like clouds, into global circulation models to improve climate projections for the future.

Low-Level Cloud Budgets across Sea Ice Edges

Abstract Interpreting behaviors of low-level clouds (LLCs) in a climate model is often not straightforward. This is particularly so over polar oceans where frozen and unfrozen surfaces coexist, with horizontal winds streaming across them, shaping LLCs. To add clarity to this interpretation issue, we conduct budget analyses of LLCs using a global atmosphere model with a fully prognostic cloud scheme. After substantiating the model’s skill in reproducing observed LLCs, we use the modeled budgets of cloud fraction and water content to elucidate physics governing changes of LLCs across sea ice edges. Contrasting LLC regimes between open water and sea ice are found. LLCs over sea ice are primarily maintained by large-scale condensation: intermittent intrusions of maritime humid air and surface radiative cooling jointly sustain high relative humidity near the surface, forming extensive but tenuous stratus. This contrasts with the LLCs over open water where the convection and boundary layer condensation sustain the LLCs on top of deeper boundary layers. Such contrasting LLC regimes are influenced by the direction of horizontal advection. During on-ice flow, large-scale condensation dominates the regions, both open water and sea ice regions, forming clouds throughout the lowest several kilometers of the troposphere. During off-ice flow, as cold air masses travel over the open water, the cloud layer lifts and becomes denser, driven by increased surface fluxes that generate LLCs through boundary layer condensation and convective detrainment. These results hold in all seasons except summer when the atmosphere–surface decoupling substantially reduces the footprints of surface type changes.

Cosine

We present a self-designing key-value storage engine, Cosine, which can always take the shape of the close to "perfect" engine architecture given an input workload, a cloud budget, a target performance, and required cloud SLAs. By identifying and formalizing the first principles of storage engine layouts and core key-value algorithms, Cosine constructs a massive design space comprising of sextillion (10 36 ) possible storage engine designs over a diverse space of hardware and cloud pricing policies for three cloud providers - AWS, GCP, and Azure. Cosine spans across diverse designs such as Log-Structured Merge-trees, B-trees, Log-Structured Hash-tables, in-memory accelerators for filters and indexes as well as trillions of hybrid designs that do not appear in the literature or industry but emerge as valid combinations of the above. Cosine includes a unified distribution-aware I/O model and a learned concurrency-aware CPU model that with high accuracy can calculate the performance and cloud cost of any possible design on any workload and virtual machines. Cosine can then search through that space in a matter of seconds to find the best design and materializes the actual code of the resulting storage engine design using a templated Rust implementation. We demonstrate that on average Cosine outperforms state-of-the-art storage engines such as write-optimized RocksDB, read-optimized WiredTiger, and very write-optimized FASTER by 53x, 25x, and 20x, respectively, for diverse workloads, data sizes, and cloud budgets across all YCSB core workloads and many variants.

Improving the representation of supercooled liquid water in the HARMONIE-AROME weather forecast model

A realistic representation of mixed-phase clouds in weather and climate models is essential to accurately simulate the model’s radiative balance and water cycle. In addition, it is important for providing downstream applications with physically realistic model data for computation of, for instance, atmospheric icing on societal infrastructure and aircraft. An important quantity for forecasts of atmospheric icing is to model accurately supercooled liquid water (SLW). In this study, we implement elements from the Thompson cloud microphysics scheme into the numerical weather prediction model HARMONIE-AROME, with the aim to improve its ability to predict SLW. We conduct an idealised process-level evaluation of microphysical processes, and analyse the water phase budget of clouds and precipitation to compare the modified and original schemes, and also identify the processes with the most impact to form SLW. Two idealised cases representing orographic lift and freezing drizzle, both known to generate significant amounts of SLW, are setup in a 1 D column version of HARMONIE-AROME. The experiments show that the amount of SLW is largely sensitive to the ice initiation processes, snow and graupel collection of cloud water, and the rain size distribution. There is a doubling of the cloud water maximum mixing ratio, in addition to a prolonged existence of SLW, with the modified scheme compared with the original scheme. The spatial and temporal extent of cloud ice and snow are reduced, due to stricter conditions for ice nucleation. The findings are important as the HARMONIE-AROME models is used for operational forecasting in many countries in northern Europe having a colder climate, as well as for climate assessments over the Arctic region.

Estimate of rain evaporation rates from dual-wavelength lidar measurements: comparison against a model analytical solution

Rain evaporation significantly contributes to moisture and heat cloud budgets. In this paper, we illustrate an approach to estimate the median volume raindrop diameter and the rain evaporation rate profiles from dual-wavelength lidar measurements. These observational results are compared with those provided by a model analytical solution. We made use of measurements from the multi-wavelength Raman lidar BASIL.

Rain Evaporation Rate Estimates from Dual-Wavelength Lidar Measurements and Intercomparison against a Model Analytical Solution

AbstractRain evaporation, while significantly contributing to moisture and heat cloud budgets, is a still poorly understood process with few measurements presently available. Multiwavelength lidars, widely employed in aerosols and clouds studies, can also provide useful information on the microphysical characteristics of light precipitation, for example, drizzle and virga. In this paper, lidar measurements of the median volume raindrop diameter and rain evaporation rate profiles are compared with a model analytical solution. The intercomparison reveals good agreement between the model and observations, with a correlation between the profiles up to 65% and a root-mean-square error up to 22% with a 5% bias. Larger discrepancies are due to radiosonde soundings different air masses and model assumptions no more valid along the profile as nonsteady atmosphere and/or appearance of collision–coalescence processes. Nevertheless, this study shares valuable information to better characterize the rain evaporation processes.

Surface rainfall and cloud budgets associated with mei-yu torrential rainfall over eastern China during June 2011

Surface rainfall and cloud budgets associated with three heavy rainfall events that occurred over eastern China during the mei-yu season in June 2011 were analyzed using 2D cumulus ensemble model simulation data. Model domain mean rainfall showed three peaks in response to three prescribed ascending motion maxima, primarily through the mean moisture convergence during the torrential rainfall period. Prescribed ascending motion throughout the troposphere produced strong convective rainfall during the first (9 June) and third (17–18 June) rainfall events, whereas strong prescribed ascending motion in the mid and upper troposphere and weak subsidence near the surface generated equally important stratiform and convective rainfall during the second rainfall event (14 June). The analysis of surface rainfall budgets reveals that convective rainfall was associated with atmospheric drying during the first event and moisture convergence during the third event. Both stratiform and convective rainfall responded primarily to moisture convergence during the second event. An analysis of grid data shows that the first and third mean rainfall maxima had smaller horizontal scales of the precipitation system than the second.

Temporal and spatial scale dependence of precipitation analysis over the tropical deep convective regime

Data from Goddard cumulus ensemble model experiment are used to study temporal and spatial scale dependence of tropical rainfall separation analysis based on cloud budget during Tropical Ocean Global Atmosphere Coupled Ocean Atmosphere Response Experiment (TOGA COARE). The analysis shows that the calculations of model domain mean or time-mean grid-scale mean simulation data overestimate the rain rates of the two rainfall types associated with net condensation but they severely underestimate the rain rate of the rainfall type associated with net evaporation and hydrometeor convergence.

A partitioning analysis of tropical rainfall based on cloud budget

Precipitation in the deep tropical convective regime is investigated through the partitioning analysis of grid-scale data from a two-dimensional cloud-resolving model simulation based on cloud budget. The model is forced by large-scale vertical velocity, zonal wind, horizontal advection, and sea surface temperature observed and derived from TOGA COARE. In cloud budget, rainfall is associated with cloud microphysical processes denoted by the net condensation and dynamical processes denoted by hydrometeor change/convergence. Thus, three rainfall types are separated in this scheme. Time-mean analysis shows that the rainfall type associated with hydrometeor loss/convergence only accounts for 27% of total rainfall, which corresponds to downward motions and downward water vapor and hydrometeor mass flux in the mid and lower troposphere. The two other rainfall types associated with the net condensation contribute equally to total rainfall (36%), which are generally related to upward motions and upward water vapor and hydrometeor mass flux throughout the troposphere.

Thermal aspect of the diurnal variation of tropical convective and stratiform rainfall

The diurnal variation of radiation plays a key role in determining the diurnal variations of tropical oceanic convective and stratiform rainfall, and the examination of such a relationship requires a direct link between the radiation term in a heat budget and the surface rain rate in a cloud budget. Thus, the thermally related surface rainfall budgets derived from the combination of cloud and heat budgets are analysed with two-dimensional equilibrium cloud-resolving model simulation data to study the effects of sea surface temperature (SST) and cloud radiative, and microphysical processes on the diurnal variations of convective and stratiform rainfall. The results show that the increase in SST, the inclusion of diurnal variation of SST and the exclusion of cloud radiative processes increase negative diurnal anomalies of heat divergence over rainfall-free regions during the nighttime through changing the vertical structures of diurnal anomaly of radiation in the troposphere. The strengthened negative diurnal anomalies of heat divergence over rainfall-free regions enhance positive diurnal anomalies of heat divergence over convective regions, which intensifies the positive diurnal anomaly of convective rainfall. The exclusion of microphysical effects of ice clouds increases the negative diurnal anomaly of heat divergence over rainfall-free regions during the nighttime through reducing latent heat; this appears to enhance the positive diurnal anomaly of heat divergence over raining stratiform regions, and thus stratiform rainfall.

Precipitation equations and their applications to the analysis of diurnal variation of tropical oceanic rainfall

The cloud, water, and heat budgets lay down a foundation for studying relations between precipitation, clouds, and environmental water vapor and thermal conditions. The water vapor constrained precipitation equation is derived by combining cloud budget with water vapor budget, whereas the thermally constrained precipitation equation is derived by combining cloud budget with heat budget. The precipitation equations are applied to analyze the diurnal variations of tropical oceanic rainfall using the 21 day two‐dimensional cloud‐resolving model simulation during Tropical Ocean–Global Atmosphere Coupled Ocean‐Atmosphere Response Experiment and two additional sensitivity experiments. One sensitivity experiment excludes the diurnal variation of large‐scale forcing, and the other experiment excludes the diurnal variation of solar zenith angle. The analysis shows that the diurnal variations of water vapor convergence and heat divergence associated with diurnally varying imposed large‐scale upward motions play a primary role in the development of rainfall peaks in both afternoon and nighttime, whereas the diurnal variation of radiation is secondary in the formation of nocturnal rainfall peak. The diurnal variation of radiation associated with diurnally varying solar zenith angle determines the diurnal variations of tropical oceanic rainfall when the diurnal variation of large‐scale circulation is absent. The diurnal variations of rainfall can be concisely described by simplified diurnally perturbed surface rainfall equations.

Water vapor, cloud, and surface rainfall budgets associated with the landfall of Typhoon Krosa (2007): A two-dimensional cloud-resolving modeling study

Water vapor, cloud, and surface rainfall budgets associated with the landfall of Typhoon Krosa on 6–8 October 2007 are analyzed based on a two-dimensional cloud-resolving model simulation. The model is integrated with imposed zonally-uniform vertical velocity, zonal wind, horizontal temperature, and vapor advection from NCEP/Global Data Assimilation System (GDAS) data. The simulation data that are validated with observations are examined to study physical causes associated with surface rainfall processes during the landfall. The time- and domain-mean analysis shows that when Krosa approached the eastern coast of China on 6 October, the water vapor convergence over land caused a local atmospheric moistening and a net condensation that further produced surface rainfall and an increase of cloud hydrometeor concentration. Meanwhile, latent heating was balanced by advective cooling and a local atmospheric warming. One day later, the enhancement of net condensation led to an increase of surface rainfall and a local atmospheric drying, while the water vapor convergence weakened as a result of the landfall-induced deprivation of water vapor flux. At the same time, the latent heating is mainly compensated the advective cooling. Further weakening of vapor convergence on 8 October enhanced the local atmospheric drying while the net condensation and associated surface rainfall was maintained. The latent heating is balanced by advective cooling and a local atmospheric cooling.

Dependence of cloud properties derived from spectrally resolved visible satellite observations on surface temperature

Abstract. Cloud climate feedback constitutes the most important uncertainty in climate modelling, and currently even its sign is still unknown. In the recently published report of the intergovernmental panel on climate change (IPCC), 6 out of 20 climate models showed a positive and 14 a negative cloud radiative feedback in a doubled CO2 scenario. The radiative budget of clouds has also been investigated by experimental methods, especially by studying the relation of satellite observed broad band shortwave and longwave radiation to sea surface temperature. Here we present a new method for the investigation of the dependence of cloud properties on temperature changes, derived from spectrally resolved satellite observations in the visible spectral range. Our study differs from previous investigations in three important ways: first, we directly extract cloud properties (effective cloud fraction and effective cloud top height) and relate them to surface temperature. Second, we retrieve the cloud altitude from the atmospheric O2 absorption instead from thermal IR radiation. Third, our correlation analysis is performed using 7.5 years of global monthly anomalies (with respect to the average of the same month for all years). For most parts of the globe (except the tropics) we find a negative correlation of effective cloud fraction versus surface-near temperature. In contrast, for the effective cloud top height a positive correlation is found for almost the whole globe. Both findings might serve as an indicator for an overall positive cloud radiative feedback. Another peculiarity of our study is that the cloud-temperature relationships are determined for fixed locations (instead to spatial variations over selected areas) and are based on the "natural" variability over several years (instead the anomaly for a strong El-Nino event). From a detailed comparison to cloud properties from the International Satellite Cloud Climatology Project (ISCCP), in general good agreement is found. However, also systematic differences occurred indicating that our results provide independent and complementary information on cloud properties. Climate models should thus aim to reproduce our findings. Recommendations for the development of a "processor" to convert model results into the cloud sensitive quantities observed by the satellite are given.

Solar radiation budget and radiative forcing due to aerosols and clouds

This study integrates global data sets for aerosols, cloud physical properties, and shortwave radiation fluxes with a Monte Carlo Aerosol‐Cloud‐Radiation (MACR) model to estimate both the surface and the top‐of‐atmosphere (TOA) solar radiation budget as well as atmospheric column solar absorption. The study also quantifies the radiative forcing of aerosols and that of clouds. The observational input to MACR includes data from the Multiangle Imaging Spectroradiometer (MISR) for aerosol optical depths, single scattering albedos, and asymmetry factors; satellite retrieved column water vapor amount; the Total Ozone Mapping Spectrometer (TOMS) total ozone amount; and cloud fraction and cloud optical depth from the Cloud and Earth's Radiant Energy System (CERES) cloud data. The present radiation budget estimates account for the diurnal variation in cloud properties. The model was validated against instantaneous, daily and monthly solar fluxes from the ground‐based Baseline Surface Radiation Network (BSRN) network, the Global Energy Balance Archive (GEBA) surface solar flux data, and CERES TOA measurements. The agreement between simulated and observed values are within experimental errors, for all of the cases considered here: instantaneous fluxes and monthly mean fluxes at stations around the world and TOA fluxes and cloud forcing for global annual mean and zonal mean fluxes; in addition the estimated aerosol forcing at TOA also agrees with other observationally derived estimates. Overall, such agreements suggest that global data sets of aerosols and cloud parameters released by recent satellite experiments (MISR, MODIS and CERES) meet the required accuracy to use them as input to simulate the radiative fluxes within instrumental errors. Last, the atmospheric solar absorption derived in this study should be treated as an improved estimate when compared with earlier published studies. The main source of improvement in the present estimate is the use of global distribution of observed parameters for model input such as aerosols and clouds. The agreement between simulated and observed solar fluxes at the surface supports our conclusion that the present estimate is an improvement over previous studies. MACR with the global input data was used to simulate the global and regional solar radiation budget, aerosol radiative forcing and cloud radiative forcing for a 3‐year period from 2000 to 2002. We estimate the planetary albedo for a 3‐year average to be 28.9 ± 1.2% to be compared with CERES estimate of 28.6% and ERBE's estimate of 29.6%. Without clouds (including aerosols) the planetary albedo is only 15.0 ± 0.6%. The global mean TOA shortwave cloud forcing is −47.5 ± 4 W m−2, comparing well with the CERES and ERBE estimates of −46.5 and −48 W m−2, respectively. The clear‐sky atmospheric absorption is 72 ± 3 W m−2, and the surface absorption is 218 ± 4 W m−2. Clouds in all‐sky conditions enhance atmospheric absorption from 72 ± 3 W m−2 to 79 ± 5 W m−2 and decrease surface solar absorption from 218 ± 4 W m−2 to 164 ± 6 W m−2. The present estimate of 79 W m−2 for all‐sky solar absorption is much larger than the Intergovernmental Panel on Climate Change (2001) values of about 67 W m−2. Most of the increased atmospheric solar absorption is due to improved treatment of aerosol absorption (backed by surface based aerosol network and chemical transport models) and water vapor spectroscopic data. The global mean clear‐sky aerosol (both natural and anthropogenic) radiative forcing at the TOA and the surface are −6.0 ± 1 W m−2 and −11.0 ± 2 W m−2, respectively. In the presence of clouds the aerosol radiative forcing is −3.0 ± 1 W m−2 (at TOA) and −7.0 ± 2 W m−2 (at the surface). The study also documents the significant regional variations in the solar radiation budget and radiative forcing of aerosols and clouds.

Estimation of Oceanic Precipitation Efficiency in Cloud Models

Abstract Precipitation efficiency is estimated based on vertically integrated budgets of water vapor and clouds using hourly data from both two-dimensional (2D) and three-dimensional (3D) cloud-resolving simulations. The 2D cloud-resolving model is forced by the vertical velocity derived from the Tropical Ocean Global Atmosphere Coupled Ocean–Atmosphere Response Experiment (TOGA COARE). The 3D cloud-resolving modeling is based on the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) simulation of Typhoon Nari (in 2001). The analysis of the hourly moisture and cloud budgets of the 2D simulation shows that the total moisture source (surface evaporation and vertically integrated moisture convergence) is converted into hydrometeors through vapor condensation and deposition rates regardless of the area size where the average is taken. This leads to the conclusion that the large-scale and cloud-microphysics precipitation efficiencies are statistically equivalent. Results further show that convergence (divergence) of hydrometeors would make precipitation efficiency larger (smaller). The precipitation efficiency tends to be larger (even >100%) in light rain conditions as a result of hydrometeor convergence from the neighboring atmospheric columns. Analysis of the hourly moisture and cloud budgets of the 3D results from the simulation of a typhoon system with heavy rainfall generally supports that of 2D results from the simulation of the tropical convective system with moderate rainfall intensity.

Entrainment and Detrainment in Numerically Simulated Cumulus Congestus Clouds. Part II: Cloud Budgets

Abstract This paper is the second in a three-part series in which a three-dimensional numerical cloud model is used to simulate cumulus congestus clouds at high resolution in an effort to better understand the mechanisms associated with entrainment and detrainment. The prescribed environment is that associated with nonprecipitating summertime New Mexican cumulus clouds that formed on consecutive days. Using budgets of mass and moisture, the effects of the clouds and their environment are examined here with an emphasis on understanding the life cycle of the clouds and the production of narrow detrainment layers aloft. Results are compared with measurements obtained in similar environments. The mass flux profiles indicate the presence of a strong, persistent thermally driven circulation within the boundary layer, with the cloud circulation being secondary. Collapsing turrets appear to be responsible for the significant detrainment that occurs at mid levels within one simulated cloud. Transport by downdrafts...

Radiative/Turbulent Transfer Interactions in Layer Clouds

Abstract The differential absorption and emission of radiation with height inside clouds creates sources and sinks of buoyancy and thus can be an important factor in the turbulence-maintaining and dissipating processes of the clouds. This paper is concerned with the roles that solar and infrared radiation play in the turbulence budget of layer clouds, with primary emphasis on marine stratocumulus and inferential discussion of other layer cloud systems. Physically realistic parameterizations of solar and infrared (IR) fluxes are used to show how the turbulence generation by cloud-top IR cooling can be more than offset by stabilization due to absorption of sunlight, and how the role of cloud-base IR warming depends crucially on the height of the cloud base. In the context of a mixed-layer model, these effects can be cast entirely in terms of the height of the layer's center of mass relative to the net heating and/or cooling due to the radiative transfer. Implications for the diurnal cycle and for a thin-clo...