Grantee Research Project Results
Final Report: Development and Evaluation of a Shallow Convection Parameterization for Mesoscale Models
EPA Grant Number: R825254Title: Development and Evaluation of a Shallow Convection Parameterization for Mesoscale Models
Investigators: Seaman, Nelson , Kain, John S.
Institution: Pennsylvania State University
EPA Project Officer: Hahn, Intaek
Project Period: October 1, 1996 through September 30, 1999
Project Amount: $449,000
RFA: Air Quality (1996) RFA Text | Recipients Lists
Research Category: Air Quality and Air Toxics , Air
Objective:
A shallow-convection parameterization suitable for both marine and continental regimes is developed for use in mesoscale meteorological models. This work is designed to support the numerical generation of improved meteorological inputs for air-chemistry models. The original objectives of this project were to:- Develop and refine a detailed parameterization of shallow convection designed for use in mesoscale meteorological models with grid lengths of 10 to 60 km.
- Link the new shallow-cloud parameterization to existing advanced parameterizations of deep convection and turbulent planetary boundary layer (PBL) processes, with an emphasis on providing a smooth transition between the various modes of atmospheric convection.
- Perform detailed evaluations based on the meteorological fields generated in 1-D and 3-D model simulations, using the new shallow-convection parameterization scheme in various convective environments, and evaluate the influence of the simulated shallow-cloud processes in an air-chemistry model.
Summary/Accomplishments (Outputs/Outcomes):
The new Penn State shallow-convection parameterization was designed to be consistent with the explicit-moisture, deep-convection, radiation and turbulence physics of the 3-D Penn State/NCAR mesoscale model, known as the MM5 (Grell, et al., 1994). In particular, it transitions to the Kain-Fritsch (1990) deep convection scheme when simulated cloud depths exceeds a critical depth, DKF, or to the Dudhia (1989) explicit moisture scheme when shallow clouds spread to saturate a grid element in a more stable environment. The parameterized shallow convection is triggered primarily by boundary-layer turbulence via the model-predicted turbulent kinetic energy (TKE) (Shafran, et al., 2000) and the surface-layer temperature and moisture. Cloud depth is calculated using parcel buoyancy theory, but the cloud top grows gradually as a function of the updraft velocity maximum, rather than instantaneously as in the Kain-Fritsch scheme. The closure for the cloud-base mass flux is based on a hybrid formulation that combines a convective available potential energy (CAPE)-removal closure for deeper clouds (as in the Kain-Fristch deep convection) and a TKE-based closure for very shallow clouds. This flexible hybrid assumption has been tested in case studies over land and oceans and has been found to perform reasonably well in all cases.Cloudy air in the sub-grid convective updrafts is detrained from the 1-D cloud model into a class of approximately neutrally buoyant clouds, or NBCs. Prognostic equations are used to predict the sub-grid NBC fraction and liquid/ice content. The NBCs can dissipate through several physical processes, including evaporation at cloud edge due to horizontal turbulent mixing, vertical diffusion, precipitation, ice settling, and cloud-top entrainment instability (CTEI). Although the competed sub-model is designed primarily for use in mesoscale numerical weather prediction models, it is adaptable to climate models and air-chemistry models as well.
Following the development phase of the shallow-cloud sub-model, testing and evaluation studies were performed. Using a 1-D version of the Penn State/NCAR MM5, the shallow-convection parameterization first was evaluated for six different marine and continental environments. The simulations produced qualitatively realistic thermodynamic structures and cloud fields in all cases. In the first application, the 1-D model was initialized for a marine case, using a composite sounding from the ASTEX (Atlantic Stratocumulus Transition Experiment). This experiment correctly produced a quasi-steady solution with ~6-13 percent cumulus-cloud areas topped by a solid stratus deck beneath the inversion of the Bermuda High. The sub-grid scale shallow convection was shown to be important for preventing over-moistening of the convective turbulent boundary layer (CTBL) by forcing compensating subsidence in the cloud environment.
In a second slowly evolving marine case, the shallow-cloud parameterization was applied to an ASTEX case using a moving Lagrangian framework. In this second experiment, the 1-D MM5 successfully simulated the observed transition from stratus to stratocumulus to trade cumulus as an air column moved southward ~1400 km from a relatively cool-water regime near 41 N to a region with warmer subtropical seas near 30 N. The simulated cloud fields agreed reasonably well with a variety of observed cloud characteristics, such as the distribution of cloud fraction, cloud water, precipitation and the liquid-water path. Model results also compared favorably with those produced by six other models described in the literature.
Four more cases were studied using the 1-D model framework in continental environments, where the diurnal forcing was much greater than over the ocean. The third experiment investigated a case over OK from April 1997, in which an observed field of post-frontal stratocumulus clouds was simulated rather well by the model. The net surface radiation, and the sensible and latent heat fluxes at the surface were reproduced quite reasonably. This forcing led to the successful evolution of the primary boundary-layer and cloud characteristics, such as height of cloud top and cloud base, the PBL depth, and cloud liquid-water path.
Next, the fourth experiment studied a transitional cumulus and stratocumulus case in the vicinity of Pittsburgh, PA, for which skies were initially clear. As daytime heating developed, many small shallow cumuli were initiated before mid-morning, and then rapidly grew until reaching the inversion base near 700 mb. Capped by a mid-tropospheric inversion, the strong vertical moisture transport by the cumulus clouds and detrainment from their updrafts soon led to formation of a stratocumulus layer. The stratocumulus in the model eventually covered 90 percent of the sky by mid-afternoon, matching observed cloud-area growth quite well. Meanwhile, simulated cloud-base heights rose in response to the growth of the turbulent boundary layer, gradually causing cloud depths to decrease. In contrast to the marine cases, this and other continental cases displayed strong forcing and rapid changes in cloud characteristics, demonstrating the versatility of the scheme.
In the fifth experiment, another case over OK from July 1997, only widely scattered fair-weather cumulus humilis were observed. For this case the 1-D shallow-convection model developed mostly accurate representations for the cloud area, base height and depth. As described, the 1-D model produced very different shallow-cloud-field responses in each of these five experiments. The different character of the cumulus clouds observed was primarily due to differences in the strength of the inversion layer above the planetary boundary layer, and variations in the surface temperature and moisture, which caused to large differences in the surface fluxes. The 1-D model represented these changes rather well in all cases, which contributed significantly to the success of the model's simulations of the relative-humidity structure and cloud-fields.
For the final continental case, also over OK in July 1997, a sixth experiment investigated the ability of the shallow-convection parameterization to detect a conditionally unstable environment in which widespread deep convection was observed, and to pass control to the Kain-Fritsch deep convection scheme. In this case, the 1-D MM5, with the shallow-convection scheme, successfully simulated the deep convection, including the approximate updraft initiation times, height of cloud top and cloud base, and mixed-layer depth. Water mass detrained from the cloud updrafts formed convective downdrafts and anvil clouds, which periodically cooled the surface, reduced the surface fluxes and temporarily suppressed the mixed layer. For this case, the model developed three separate deep-convection events, similar to those observed at the central ARM-CART facility at Lemont, OK. Thus, the new cloud scheme demonstrated considerable versatility for a wide range of atmospheric environments.
Following the evaluation experiments, nine additional experiments were conducted to investigate the model's sensitivity to several key factors: the closure assumptions, NBC dissipation mechanisms and model vertical resolution. It was found that the shallow-cloud scheme's hybrid mass-flux closure behaved smoothly, even if the number of updrafts calculated in a grid cell was arbitrarily doubled or halved. In these cases, the cloud sub-model developed compensating trends in the vertical velocities and water flux that tended to mute the impact on the total cloud-base mass flux. Predictably, replacement of the hybrid closure by a closure dependent only on the boundary-layer TKE (i.e., removal of the CAPE-sensitive part of the hybrid closure) worked well when the clouds were quite shallow (less than ~1 km depth). However, for situations with deeper clouds where CAPE becomes larger, the behavior of the sub-model was degraded significantly. That is, the ability of the shallow-cloud sub-model to process boundary-layer air through the cloud updrafts was impaired for deeper clouds when the CAPE-sensitive element of the hybrid closure was removed.
Moreover, it was found that the shallow-cloud scheme is highly dependent on the drizzle mechanism and cloud-edge evaporation that remove liquid water from the NBCs. This suggests that it is important to link the shallow-cloud sub-model to a moist microphysics scheme that handles light precipitation well for clouds with low to moderate water content. However, the cloud sub-model was not very sensitive to the model vertical resolution in the planetary boundary layer and in at least the lower part of the cloud layer, until that resolution became rather coarse (at least 160 m).
When the shallow cloud parameterization was installed in the 3-D MM5 mesoscale model, it performed quite well at 36-km horizontal resolution and with 32 layers in the vertical direction. The model reproduced the horizontal distribution of shallow-cloud area over KS and OK rather well in a springtime case with widespread stratocumulus clouds.
To summarize, the main conclusions of this study can be listed as follows:
- The hybrid mass-flux closure and the cloud-trigger mechanism based on boundary layer properties perform quite well and are numerically stable.
- The shallow-convection parameterization successfully reproduces a variety of observed cloud characteristics in the 1-D MM5 for both marine and continental environments.
- The shallow-convection sub-model is relatively insensitive to errors in the estimation of the number of updrafts because the hybrid mass-flux closure compensates to limit the impact on total cloud mass flux.
- It is recommended that, for best results, the shallow-convection sub-model be applied with vertical resolutions no coarser than ~160 m in the PBL and lower half of the shallow cloud layer.
- The shallow-convection sub-model performed well when installed and applied in the 3-D Penn State/NCAR mesoscale model at 36-km horizontal resolution.
- This shallow-convection sub-model is now ready for a wide range of testing and applications in the 3-D model, including applications in support of air-quality studies.
- Evaluate the shallow-convection parameterization for a wider variety of cloud environments, especially for the tropics and arctic regions.
- The sub-model should be applied in the 3-D MM5 for a range of grid resolutions.
- The cloud-sensitive variables generated by the sub-model, such as liquid water content and mass flux, should be passed to an air-chemistry model to study the implications of these clouds for chemical transport and aqueous reactions.
- Broader testing should be conducted to better reveal the sub-model's sensitivity to the updraft and NBC formulations, and to the interactions among these components and other model physics.
- Comparison should be made with results from Large Eddy Simulation (LES) studies in moist conditions to better evaluate many of the cloud-model parameters used in the present formulation.
- Studies should be conducted to replace the present formulation, that represents clouds in a grid cell using a single updraft size, by an ensemble of shallow-cloud sizes, which should yield more accurate results.
Thus, the first two objectives have been fulfilled completely. The third objective has been met in part. That is, the original intent was to apply the fields simulated by a 3-D mesoscale meteorological model (the Penn State/NCAR MM5 model), with and without the shallow-convection scheme, in an air-chemistry model (MASQSIP). Although the shallow-cloud sub-model was transitioned into the 3-D MM5, and was tested and evaluated successfully, there was insufficient time to apply these fields in the MASQSIP chemistry model. Nevertheless, there is no technological obstacle preventing this final step at a future time, since the shallow-cloud sub-model is now operational in the Penn State version of the MM5.
References:
Dudhia J. Numerical study of convection observed during the winter monsoon experiment using a mesoscale two-dimensional model. Journal of Atmospheric Science, 1998;46:3077-3107.
Grell GA, Dudhia J, Stauffer DR. A description of the fifth generation Penn State/NCAR Mesoscale Model (MM5). NCAR Technical Note, NCAR TN-398-STR, 1994, 138 pp.
Kain JS, Fritsch JM. A one-dimensional entraining/detraining plume model and its application in convective parameterization. Journal of Atmospheric Science 1990;47:2784-2802.
Shafran PC, Seaman NL, Gayno GA. Evaluation of numerical predictions of boundary-layer structure during the Lake Michigan Ozone Study (LMOS). Journal of Applied Meteorology 2000;39:412-426.
Journal Articles on this Report : 2 Displayed | Download in RIS Format
Other project views: | All 17 publications | 2 publications in selected types | All 2 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Deng AJ, Seaman NL, Kain JS. A shallow-convection parameterization for mesoscale models. Part II: Verification and sensitivity study. Journal of Atmospheric Sciences 2003;60(1):57-78. |
R825254 (Final) |
Exit |
|
Deng AJ, Seaman NL, Kain JS. A shallow-convection parameterization for mesoscale models. Part I: Submodel description and preliminary applications. Journal of Atmospheric Sciences 2003;60(1):34-56. |
R825254 (Final) |
Exit |
Supplemental Keywords:
air, ambient air, atmosphere, acid deposition, global climate, tropospheric, precipitation, chemical transport, clouds, aqueous chemistry, actinic flux, physics, modeling, climate modeling, mesoscale modeling, acid rain, particulates, modeling, Northeast, Pennsylvania, PA, EPA Region 3., RFA, Scientific Discipline, Air, Geographic Area, particulate matter, State, Atmospheric Sciences, tropospheric ozone, EPA Region, particulates, mesoscale models, shallow convection parameterization, Pennsylvania, Region 3, air sampling, boundary layer turbulence, cloud-based mass flux, air pollution models, atmospheric transport, Acute health effects, acute toxicity, cloud radar, modeling studies, subgrid scale clouds, particle transport, acid particlesProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.