Search Results
You are looking at 101 - 110 of 115 items for
- Author or Editor: William R. Cotton x
- Refine by Access: All Content x
Abstract
An existing two-way interactive grid-nesting technique is generalized to accommodate stretched grids and a spatially variable grid-nesting ratio. The new scheme applies the same reversibility constraint and possesses the same scalar and momentum conservation properties as its predecessor. The scheme can be applied in any coordinate direction but was motivated primarily by the common use of vertical grid stretching. The scheme is successfully toted in a simulation of a collapsing cold pool.
Abstract
An existing two-way interactive grid-nesting technique is generalized to accommodate stretched grids and a spatially variable grid-nesting ratio. The new scheme applies the same reversibility constraint and possesses the same scalar and momentum conservation properties as its predecessor. The scheme can be applied in any coordinate direction but was motivated primarily by the common use of vertical grid stretching. The scheme is successfully toted in a simulation of a collapsing cold pool.
Abstract
The Colorado State University cloud model is applied to the simulation of orogrophic cloud snowfall. A model of ice crystal aggregation processes and primary nucleation and secondary ice particle production of crystals is described. Sensitivity experiments demonstrated that aggregation plays an important role in controlling the fields of cloud liquid water content, ice crystal concentrations, and surface precipitation amounts.
The sensitivity experiments also support observations that the air mass is often quite clean in upper levels of stable orographic clouds. Introducing a reduction of available nuclei that can be activated by deposition/sorption processes brought concentrations to within observed values.
This study clearly emphasizes the need for a great deal more fundamental research in the physics of aggregation processes and primary and secondary nucleation of ice crystals.
Abstract
The Colorado State University cloud model is applied to the simulation of orogrophic cloud snowfall. A model of ice crystal aggregation processes and primary nucleation and secondary ice particle production of crystals is described. Sensitivity experiments demonstrated that aggregation plays an important role in controlling the fields of cloud liquid water content, ice crystal concentrations, and surface precipitation amounts.
The sensitivity experiments also support observations that the air mass is often quite clean in upper levels of stable orographic clouds. Introducing a reduction of available nuclei that can be activated by deposition/sorption processes brought concentrations to within observed values.
This study clearly emphasizes the need for a great deal more fundamental research in the physics of aggregation processes and primary and secondary nucleation of ice crystals.
Abstract
Pollution aerosols acting as cloud condensation nuclei (CCN) have the potential to alter warm rain clouds via the aerosol first and second indirect effects in which they modify the cloud droplet population, cloud lifetime and size, rainfall efficiency, and radiation balance from increased albedo. For constant liquid water content, an increase in CCN concentration (N CCN) tends to produce an increased concentration of droplets with smaller diameters. This reduces the collision and coalescence rate, and thus there is a local reduction in rainfall. While this process applies to warm clouds, it does not identically carry over to mixed-phase clouds in which crystal nucleation, crystal riming, crystal versus droplet fall speed, and collection efficiency play active roles in determining precipitation amount. Sulfate-based aerosols serve as very efficient cloud nuclei but are not effective as ice-forming nuclei. In clouds where precipitation formation is dominated by the ice phase, N CCN influences precipitation growth by altering the efficiency of droplet collection by ice crystals and the fall trajectories of both droplet and crystal hydrometeors. The temporal and spatial variation in both crystal and droplet populations determines the resultant snowfall efficiency and distribution. Results of numerical simulations in this study suggest that CCN can play a significant role in snowfall production by winter, mixed-phase, cloud systems when liquid and ice hydrometeors coexist. In subfreezing conditions, a precipitating ice cloud overlaying a supercooled liquid water cloud allows growth of precipitation particles via the seeder–feeder process, in which nucleated ice crystals fall through the supercooled liquid water cloud and collect droplets. Enhanced N CCN from sulfate pollution by fossil fuel emissions modifies the droplet distribution and reduces crystal riming efficiency. Reduced riming efficiency inhibits the rate of snow growth, producing lightly rimed snow crystals that fall slowly and advect farther downstream prior to surface deposition. Simulations indicate that increasing N CCN along the orographic barrier of the Park Range in north-central Colorado results in a modification of the orographic cloud such that the surface snow water equivalent amounts are reduced on the windward slopes and enhanced on the leeward slopes. The inhibition of snowfall by pollution aerosols (ISPA) effect has significant implications for water resource distribution in mountainous terrain.
Abstract
Pollution aerosols acting as cloud condensation nuclei (CCN) have the potential to alter warm rain clouds via the aerosol first and second indirect effects in which they modify the cloud droplet population, cloud lifetime and size, rainfall efficiency, and radiation balance from increased albedo. For constant liquid water content, an increase in CCN concentration (N CCN) tends to produce an increased concentration of droplets with smaller diameters. This reduces the collision and coalescence rate, and thus there is a local reduction in rainfall. While this process applies to warm clouds, it does not identically carry over to mixed-phase clouds in which crystal nucleation, crystal riming, crystal versus droplet fall speed, and collection efficiency play active roles in determining precipitation amount. Sulfate-based aerosols serve as very efficient cloud nuclei but are not effective as ice-forming nuclei. In clouds where precipitation formation is dominated by the ice phase, N CCN influences precipitation growth by altering the efficiency of droplet collection by ice crystals and the fall trajectories of both droplet and crystal hydrometeors. The temporal and spatial variation in both crystal and droplet populations determines the resultant snowfall efficiency and distribution. Results of numerical simulations in this study suggest that CCN can play a significant role in snowfall production by winter, mixed-phase, cloud systems when liquid and ice hydrometeors coexist. In subfreezing conditions, a precipitating ice cloud overlaying a supercooled liquid water cloud allows growth of precipitation particles via the seeder–feeder process, in which nucleated ice crystals fall through the supercooled liquid water cloud and collect droplets. Enhanced N CCN from sulfate pollution by fossil fuel emissions modifies the droplet distribution and reduces crystal riming efficiency. Reduced riming efficiency inhibits the rate of snow growth, producing lightly rimed snow crystals that fall slowly and advect farther downstream prior to surface deposition. Simulations indicate that increasing N CCN along the orographic barrier of the Park Range in north-central Colorado results in a modification of the orographic cloud such that the surface snow water equivalent amounts are reduced on the windward slopes and enhanced on the leeward slopes. The inhibition of snowfall by pollution aerosols (ISPA) effect has significant implications for water resource distribution in mountainous terrain.
Abstract
Observational data collected during the FIRE II experiment showing the existence of bimodal ice spectra along with experimental evidence of the size dependence of riming are utilized in the development of a bimodal ice spectrum parameterization for use in the RAMS model. Two ice classes are defined: pristine ice and snow, each described by a separate, complete gamma distribution function. Pristine ice is small ice consisting of particles with mean sizes less than 125 µm, while snow is the large class consisting of particles greater than 125 µm. Analytical equations are formulated for the conversion between the ice classes by vapor depositional growth (sublimation). During ice subsaturated conditions, a number concentration sink is parameterized for all ice species. The performance of the parameterizations in a simple parcel model is discussed and evaluated against an explicit Lagrangian parcel microphysical model.
Abstract
Observational data collected during the FIRE II experiment showing the existence of bimodal ice spectra along with experimental evidence of the size dependence of riming are utilized in the development of a bimodal ice spectrum parameterization for use in the RAMS model. Two ice classes are defined: pristine ice and snow, each described by a separate, complete gamma distribution function. Pristine ice is small ice consisting of particles with mean sizes less than 125 µm, while snow is the large class consisting of particles greater than 125 µm. Analytical equations are formulated for the conversion between the ice classes by vapor depositional growth (sublimation). During ice subsaturated conditions, a number concentration sink is parameterized for all ice species. The performance of the parameterizations in a simple parcel model is discussed and evaluated against an explicit Lagrangian parcel microphysical model.
Abstract
The impact of giant and ultragiant cloud condensation nuclei (>5-μm radius) on drizzle formation in stratocumuli is investigated within a number of modeling frameworks. These include a simple box model of collection, a trajectory ensemble model (comprising an ensemble of Lagrangian parcel models), a 2D eddy-resolving model, and a 3D large-eddy simulation model. Observed concentrations of giant cloud condensation nuclei (GCCN) over the ocean at ambient conditions indicate that 20-μm radius haze particles exist in concentrations of between 10−4 and 10−2 cm−3, depending on ambient wind speed and seastate. It is shown that these concentrations are sufficient to move a nonprecipitating stratocumulus into a precipitating state at typical cloud condensation nucleus (CCN) concentrations of 50 to 250 cm−3, with higher concentrations of GCCN being required at higher CCN concentrations. However, at lower CCN concentrations, drizzle is often active anyway and the addition of GCCN has little impact. At high CCN concentrations, drizzle development is slow and GCCN have the greatest potential for enhancing the collection process. Thus, although drizzle production decreases with increasing CCN concentration, the relative impact of GCCN increases with increasing CCN concentration. It is also shown that in the absence of GCCN, a shift in the modal radius of the CCN distribution to larger sizes suppresses drizzle because larger modal radii enable the activation of larger droplet number concentrations. Finally, calculations of the impact of GCCN on cloud optical properties are performed over a range of parameter space. Results indicate that the presence of GCCN moderates the effect of CCN on optical properties quite significantly. In the absence of GCCN, an increase in CCN from 50 to 150 cm−3 results in a threefold increase in albedo; when GCCN exist at a concentration of 10−3 cm−3, the increase in albedo is only twofold. Thus the variable presence of GCCN represents yet another uncertainty in estimating the influence of anthropogenic activity on climate.
Abstract
The impact of giant and ultragiant cloud condensation nuclei (>5-μm radius) on drizzle formation in stratocumuli is investigated within a number of modeling frameworks. These include a simple box model of collection, a trajectory ensemble model (comprising an ensemble of Lagrangian parcel models), a 2D eddy-resolving model, and a 3D large-eddy simulation model. Observed concentrations of giant cloud condensation nuclei (GCCN) over the ocean at ambient conditions indicate that 20-μm radius haze particles exist in concentrations of between 10−4 and 10−2 cm−3, depending on ambient wind speed and seastate. It is shown that these concentrations are sufficient to move a nonprecipitating stratocumulus into a precipitating state at typical cloud condensation nucleus (CCN) concentrations of 50 to 250 cm−3, with higher concentrations of GCCN being required at higher CCN concentrations. However, at lower CCN concentrations, drizzle is often active anyway and the addition of GCCN has little impact. At high CCN concentrations, drizzle development is slow and GCCN have the greatest potential for enhancing the collection process. Thus, although drizzle production decreases with increasing CCN concentration, the relative impact of GCCN increases with increasing CCN concentration. It is also shown that in the absence of GCCN, a shift in the modal radius of the CCN distribution to larger sizes suppresses drizzle because larger modal radii enable the activation of larger droplet number concentrations. Finally, calculations of the impact of GCCN on cloud optical properties are performed over a range of parameter space. Results indicate that the presence of GCCN moderates the effect of CCN on optical properties quite significantly. In the absence of GCCN, an increase in CCN from 50 to 150 cm−3 results in a threefold increase in albedo; when GCCN exist at a concentration of 10−3 cm−3, the increase in albedo is only twofold. Thus the variable presence of GCCN represents yet another uncertainty in estimating the influence of anthropogenic activity on climate.
Abstract
Observations from multiple satellites and large-eddy simulations (LESs) from the Regional Atmospheric Modeling System (RAMS) are used to determine the extent to which free-tropospheric clouds (FTCs) affect the properties of stratocumulus. Overlying FTCs decrease the cloud-top radiative cooling in stratocumulus by an amount that depends on the upper-cloud base altitude, cloud optical thickness, and abundance of moisture between the cloud layers. On average, FTCs increase the downward longwave radiative flux above stratocumulus clouds (at 3.5 km) by approximately 30 W m−2. As a consequence, this forcing translates to a relative decrease in stratocumulus cooling rates by about 20%. Overall, the reduced cloud-top radiative cooling decreases the turbulent mixing, vertical development, and precipitation rate in stratocumulus clouds at night. During the day these effects are greatly reduced because the overlying clouds shade the stratocumulus from strong solar radiation, thus reducing the net radiative effect by the upper cloud. Differences in liquid water path are also observed in stratocumulus; however, the response is tied to the diurnal cycle and the time scale of interaction between the FTCs and the stratocumulus. Radiative effects by FTCs tend to be largest in the midlatitudes where the clouds overlying stratocumulus tend to be more frequent, lower, and thicker on average. In conclusion, changes in net radiation and moisture brought about by FTCs can significantly affect the dynamics of marine stratocumulus and these processes should be considered when evaluating cloud feedbacks in the climate system.
Abstract
Observations from multiple satellites and large-eddy simulations (LESs) from the Regional Atmospheric Modeling System (RAMS) are used to determine the extent to which free-tropospheric clouds (FTCs) affect the properties of stratocumulus. Overlying FTCs decrease the cloud-top radiative cooling in stratocumulus by an amount that depends on the upper-cloud base altitude, cloud optical thickness, and abundance of moisture between the cloud layers. On average, FTCs increase the downward longwave radiative flux above stratocumulus clouds (at 3.5 km) by approximately 30 W m−2. As a consequence, this forcing translates to a relative decrease in stratocumulus cooling rates by about 20%. Overall, the reduced cloud-top radiative cooling decreases the turbulent mixing, vertical development, and precipitation rate in stratocumulus clouds at night. During the day these effects are greatly reduced because the overlying clouds shade the stratocumulus from strong solar radiation, thus reducing the net radiative effect by the upper cloud. Differences in liquid water path are also observed in stratocumulus; however, the response is tied to the diurnal cycle and the time scale of interaction between the FTCs and the stratocumulus. Radiative effects by FTCs tend to be largest in the midlatitudes where the clouds overlying stratocumulus tend to be more frequent, lower, and thicker on average. In conclusion, changes in net radiation and moisture brought about by FTCs can significantly affect the dynamics of marine stratocumulus and these processes should be considered when evaluating cloud feedbacks in the climate system.
Abstract
One problem in computing cloud microphysical processes in coarse-resolution numerical models is that many microphysical processes are nonlinear and small in scale. Consequently, there are inaccuracies if microphysics parameterizations are forced with grid box averages of model fields, such as liquid water content. Rather, the model needs to determine information about subgrid variability and input it into the microphysics parameterization.
One possible solution is to assume the shape of the family of probability density functions (PDFs) associated with a grid box and sample it using the Monte Carlo method. In this method, the microphysics subroutine is called repeatedly, once with each sample point. In this way, the Monte Carlo method acts as an interface between the host model’s dynamics and the microphysical parameterization. This avoids the need to rewrite the microphysics subroutines.
A difficulty with the Monte Carlo method is that it introduces into the simulation statistical noise or variance, associated with the finite sample size. If the family of PDFs is tractable, one can sample solely from cloud, thereby improving estimates of in-cloud processes. If one wishes to mitigate the noise further, one needs a method for reduction of variance. One such method is Latin hypercube sampling, which reduces noise by spreading out the sample points in a quasi-random fashion.
This paper formulates a sampling interface based on the Latin hypercube method. The associated family of PDFs is assumed to be a joint normal/lognormal (i.e., Gaussian/lognormal) mixture. This method of variance reduction has a couple of advantages. First, the method is general: the same interface can be used with a wide variety of microphysical parameterizations for various processes. Second, the method is flexible: one can arbitrarily specify the number of hydrometeor categories and the number of calls to the microphysics parameterization per grid box per time step.
This paper performs a preliminary test of Latin hypercube sampling. As a prototypical microphysical formula, this paper uses the Kessler autoconversion formula. The PDFs that are sampled are extracted diagnostically from large-eddy simulations (LES). Both stratocumulus and cumulus boundary layer cases are tested. In this diagnostic test, the Latin hypercube can produce somewhat less noisy time-averaged estimates of Kessler autoconversion than a traditional Monte Carlo estimate, with no additional calls to the microphysics parameterization. However, the instantaneous estimates are no less noisy. This paper leaves unanswered the question of whether the Latin hypercube method will work well in a prognostic, interactive cloud model, but this question will be addressed in a future manuscript.
Abstract
One problem in computing cloud microphysical processes in coarse-resolution numerical models is that many microphysical processes are nonlinear and small in scale. Consequently, there are inaccuracies if microphysics parameterizations are forced with grid box averages of model fields, such as liquid water content. Rather, the model needs to determine information about subgrid variability and input it into the microphysics parameterization.
One possible solution is to assume the shape of the family of probability density functions (PDFs) associated with a grid box and sample it using the Monte Carlo method. In this method, the microphysics subroutine is called repeatedly, once with each sample point. In this way, the Monte Carlo method acts as an interface between the host model’s dynamics and the microphysical parameterization. This avoids the need to rewrite the microphysics subroutines.
A difficulty with the Monte Carlo method is that it introduces into the simulation statistical noise or variance, associated with the finite sample size. If the family of PDFs is tractable, one can sample solely from cloud, thereby improving estimates of in-cloud processes. If one wishes to mitigate the noise further, one needs a method for reduction of variance. One such method is Latin hypercube sampling, which reduces noise by spreading out the sample points in a quasi-random fashion.
This paper formulates a sampling interface based on the Latin hypercube method. The associated family of PDFs is assumed to be a joint normal/lognormal (i.e., Gaussian/lognormal) mixture. This method of variance reduction has a couple of advantages. First, the method is general: the same interface can be used with a wide variety of microphysical parameterizations for various processes. Second, the method is flexible: one can arbitrarily specify the number of hydrometeor categories and the number of calls to the microphysics parameterization per grid box per time step.
This paper performs a preliminary test of Latin hypercube sampling. As a prototypical microphysical formula, this paper uses the Kessler autoconversion formula. The PDFs that are sampled are extracted diagnostically from large-eddy simulations (LES). Both stratocumulus and cumulus boundary layer cases are tested. In this diagnostic test, the Latin hypercube can produce somewhat less noisy time-averaged estimates of Kessler autoconversion than a traditional Monte Carlo estimate, with no additional calls to the microphysics parameterization. However, the instantaneous estimates are no less noisy. This paper leaves unanswered the question of whether the Latin hypercube method will work well in a prognostic, interactive cloud model, but this question will be addressed in a future manuscript.
Abstract
The authors’ previous idealized, two-dimensional cloud resolving model (CRM) simulations of Arctic stratus revealed a surprising sensitivity to the concentrations of ice crystals. In this paper, simulations of an actual case study observed during the Beaufort and Arctic Seas Experiment are performed and the results are compared to the observed data.
It is again found in the CRM simulations that the simulated stratus cloud is very sensitive to the concentration of ice crystals. Using midlatitude estimates of the availability of ice forming nuclei (IFN) in the model, the authors find that the concentrations of ice crystals are large enough to result in the almost complete dissipation of otherwise solid, optically thick stratus layers. A tenuous stratus can be maintained in the simulation when the continuous input of moisture through the imposed large-scale advection is strong enough to balance the ice production. However, in association with the large-scale moisture and warm advection, only by reducing the concentration of IFN to 0.3 of the midlatitude estimate values can a persistent, optically thick stratus layer be maintained. The results obtained from the reduced IFN simulation compare reasonably well with observations.
The longwave radiative fluxes at the surface are significantly different between the solid stratus and liquid-water-depleted higher ice crystal concentration experiments.
This work suggests that transition-season Arctic stratus can be very vulnerable to anthropogenic sources of IFN, which can alter cloud structure sufficiently to affect the rates of melting and freezing of the Arctic Ocean.
The authors find that the Hallett–Mossop riming splintering mechanism is not activated in the simulations because the cloud droplets are very small and cloud temperatures are outside the range supporting efficient rime splintering. Thus, the conclusions drawn from the results presented in this paper may be applicable to only a limited class of Arctic stratus.
Abstract
The authors’ previous idealized, two-dimensional cloud resolving model (CRM) simulations of Arctic stratus revealed a surprising sensitivity to the concentrations of ice crystals. In this paper, simulations of an actual case study observed during the Beaufort and Arctic Seas Experiment are performed and the results are compared to the observed data.
It is again found in the CRM simulations that the simulated stratus cloud is very sensitive to the concentration of ice crystals. Using midlatitude estimates of the availability of ice forming nuclei (IFN) in the model, the authors find that the concentrations of ice crystals are large enough to result in the almost complete dissipation of otherwise solid, optically thick stratus layers. A tenuous stratus can be maintained in the simulation when the continuous input of moisture through the imposed large-scale advection is strong enough to balance the ice production. However, in association with the large-scale moisture and warm advection, only by reducing the concentration of IFN to 0.3 of the midlatitude estimate values can a persistent, optically thick stratus layer be maintained. The results obtained from the reduced IFN simulation compare reasonably well with observations.
The longwave radiative fluxes at the surface are significantly different between the solid stratus and liquid-water-depleted higher ice crystal concentration experiments.
This work suggests that transition-season Arctic stratus can be very vulnerable to anthropogenic sources of IFN, which can alter cloud structure sufficiently to affect the rates of melting and freezing of the Arctic Ocean.
The authors find that the Hallett–Mossop riming splintering mechanism is not activated in the simulations because the cloud droplets are very small and cloud temperatures are outside the range supporting efficient rime splintering. Thus, the conclusions drawn from the results presented in this paper may be applicable to only a limited class of Arctic stratus.
Abstract
A cloud-resolving model coupled to an ocean model with high vertical resolution is used to investigate air–sea interactions in 10-day long simulations. Modeled fields showed good agreement with two different convective regimes during the Tropical Ocean Global Atmosphere Coupled Ocean–Atmosphere Research Experiment (TOGA COARE) Intensive Observing Period. The model simulates the formation of precipitation-produced, stable freshwater lenses at the top of the ocean mixed layer, with a variety of horizontal dimensions and lifetimes. The simulated fresh anomalies show realistic features, such as a positive correlation between salinity and temperature, the development of a surface jet in the direction of the wind, and, as a consequence, downwelling (upwelling) on its downwind (upwind) edge. The dataset generated by the coupled model is used to evaluate the contribution from several factors (ocean currents, gustiness, and correlations between wind speed and air temperature, wind speed and water vapor mixing ratio, and wind speed and SST) to the surface heat fluxes. Gustiness was shown to be a major contribution to the simulated surface heat fluxes, especially when convection is active. In a multiday average, the contributions from the other effects (currents and wind speed–air temperature, wind speed–water vapor mixing ratio, and wind speed–SST correlations) are small; however, they cannot be neglected under certain circumstances.
Abstract
A cloud-resolving model coupled to an ocean model with high vertical resolution is used to investigate air–sea interactions in 10-day long simulations. Modeled fields showed good agreement with two different convective regimes during the Tropical Ocean Global Atmosphere Coupled Ocean–Atmosphere Research Experiment (TOGA COARE) Intensive Observing Period. The model simulates the formation of precipitation-produced, stable freshwater lenses at the top of the ocean mixed layer, with a variety of horizontal dimensions and lifetimes. The simulated fresh anomalies show realistic features, such as a positive correlation between salinity and temperature, the development of a surface jet in the direction of the wind, and, as a consequence, downwelling (upwelling) on its downwind (upwind) edge. The dataset generated by the coupled model is used to evaluate the contribution from several factors (ocean currents, gustiness, and correlations between wind speed and air temperature, wind speed and water vapor mixing ratio, and wind speed and SST) to the surface heat fluxes. Gustiness was shown to be a major contribution to the simulated surface heat fluxes, especially when convection is active. In a multiday average, the contributions from the other effects (currents and wind speed–air temperature, wind speed–water vapor mixing ratio, and wind speed–SST correlations) are small; however, they cannot be neglected under certain circumstances.
Abstract
A two-dimensional cloud-resolving model (CRM) was used to simulate the evolution of convection over the western Pacific between 19 and 26 December 1992, during the Tropical Ocean Global Atmosphere Coupled Ocean–Atmosphere Response Experiment. A control simulation (CONTROL) was performed in which observed, time-evolving, spatially homogeneous SSTs were used as a lower boundary condition. It showed that the CRM was able to properly represent the evolution of the cloud systems.
Sensitivity experiments were carried out, in which the sea surface temperature was increased (SST+) or decreased (SST−) by 1°C and the same evolving large-scale forcing used in CONTROL. The similarities among all simulations suggested that the large-scale forcing is the dominant mechanism controlling the statistics of the cloud systems, including the total precipitation. However, the convective–stratiform partition of the cloud systems was altered, the convective part being favored in SST+ and the stratiform part favored in SST−. In terms of the radiative budget, the reduced low-level cloud coverage in SST+ acted to compensate the enhancement of high-cloud coverage produced by more vigorous convection (the opposite occurred in SST−). As a consequence, the surface downward radiation was approximately the same in CONTROL, SST+, and SST−.
Abstract
A two-dimensional cloud-resolving model (CRM) was used to simulate the evolution of convection over the western Pacific between 19 and 26 December 1992, during the Tropical Ocean Global Atmosphere Coupled Ocean–Atmosphere Response Experiment. A control simulation (CONTROL) was performed in which observed, time-evolving, spatially homogeneous SSTs were used as a lower boundary condition. It showed that the CRM was able to properly represent the evolution of the cloud systems.
Sensitivity experiments were carried out, in which the sea surface temperature was increased (SST+) or decreased (SST−) by 1°C and the same evolving large-scale forcing used in CONTROL. The similarities among all simulations suggested that the large-scale forcing is the dominant mechanism controlling the statistics of the cloud systems, including the total precipitation. However, the convective–stratiform partition of the cloud systems was altered, the convective part being favored in SST+ and the stratiform part favored in SST−. In terms of the radiative budget, the reduced low-level cloud coverage in SST+ acted to compensate the enhancement of high-cloud coverage produced by more vigorous convection (the opposite occurred in SST−). As a consequence, the surface downward radiation was approximately the same in CONTROL, SST+, and SST−.
