Modeling Cloud Scale

The Goddard Cumulus Ensemble (GCE) model (Tao and Simpson, 1993) has been used by Pickering et al. (1991, 1992a,b, 1993, 1996), Scala et al. (1990), and

Stenchikov et al. (1996) in the analysis of convective transport of trace gases. The cloud model is nonhydrostatic and contains detailed representation of cloud micro-physical processes. Two- and three-dimensional versions of the model have been applied in transport analyses. The initial conditions for the model are usually from a sounding of temperature, water vapor, and winds representative of the region of storm development. Model-generated wind fields can be used to perform air parcel trajectory analyses and tracer advection calculations. Scala et al. (1990) conducted detailed air parcel trajectory analyses for an ABLE 2B storm to investigate flow patterns within the system. In this case the model showed that more than 50% of the air transported to the anvil region originated at or above 6 km, not from the boundary layer via undilute core updrafits. The trajectories also allowed diagnosis of a rotor-type circulation in the low to mid levels of the storm, which was responsible for thorough mixing of the lower troposphere (Fig. 5).

Pickering et al. (1991) used trajectory analyses derived from the GCE model wind fields for the ABLE 2A storm observed by Garstang et al. (1988) to identify air parcels that were undisturbed or modified by the storm (Fig. 6). Tracer transport calculations were performed for CO, 03, and NOx, and difference fields showing the changes in mixing ratio of each of these species due to convective transport were computed (Fig. 7). Enhanced values of ozone precursors (NOt and CO) in a biomass burning haze layer just above the boundary layer were redistributed upward and downward by the storm. Profiles taken from the two-dimensional tracer fields before and after convective transport were used in a one-dimensional photochemical

Figure 5 Composite schematic of the predominant transport pathways for the May 6, 1987, ABLE 2B simulated squall convection based on backward and forward trajectory analyses. The model cloud outline at 300 min in the simulation is shown. The horizontal dimension is 80 km. From Scala et al. (1990).

AUGUST 3,1985 ABLE 2A CLOUD MODEL SIMULATION TIME = 240 MIN

20 40 60 80

Figure 6 Summary of back trajectories produced by the GCE model for the August 3, 1985, ABLE 2A squall line. Numbers in the vertical column ahead and behind the cloud indicate the percentage of the air at that altitude that is outflow from the cloud. Most of the air pumped out of the boundary layer exits from the anvil (8 to 12 km) and the air in the "wake" has also been processed. Most of the air in the boundary layer ahead of the storm is unperturbed. Arrows indicate main flow paths. From Pickering et al. (1991).

20 40 60 80

Figure 6 Summary of back trajectories produced by the GCE model for the August 3, 1985, ABLE 2A squall line. Numbers in the vertical column ahead and behind the cloud indicate the percentage of the air at that altitude that is outflow from the cloud. Most of the air pumped out of the boundary layer exits from the anvil (8 to 12 km) and the air in the "wake" has also been processed. Most of the air in the boundary layer ahead of the storm is unperturbed. Arrows indicate main flow paths. From Pickering et al. (1991).

model to estimate ozone production rates. The upward transport of 03 precursors changed the photochemical tendency of the upper troposphere from that of 03 destruction to that of production. The same storm dynamics were used in a sensitivity study of convective transport and subsequent free tropospheric 03 production for conditions of more intense biomass burning pollution (Pickering et al., 1992b). Assuming a pristine middle and upper troposphere prior to convection, enhancements of 03 production postconvection potentially could be as great as a factor of ~ 50.

Similar methods were used by Pickering et al. (1992a) to examine transport of urban plumes by deep convection. Transport of the Oklahoma City plume by the June 10-11, 1985, PRE-STORM squall line and of the Manaus, Brazil, plume by the April 26, 1987, ABLE 2B squall line were simulated with the two-dimensional GCE model. In the Oklahoma event forward trajectories from the boundary layer at the leading edge of the storm showed that almost 75% of the low-level inflow was transported to altitudes exceeding 8 km. Over 35% of the air parcels reached altitudes over 12 km. For the Amazonian storm, 50% of the trajectories indicated transport to altitudes greater than 12 km. However, nearly 25% of the air parcels indicated air being detrained from the rear of the cloud between 4 and 8 km, and

CO Difference (ppb) (360 min-120 miri)

CO Difference (ppb) (360 min-120 miri)

40 60 80 100 120 140 160 180

Distance (km)

Figure 7 Difference (postconvection minus undisturbed) in model-computed CO tracer concentrations for the August 3, 1995, ABLE 2A squall line. Increases in CO are noted throughout the main updraft region and anvil, and decreases are seen in the downdraft region. From Pickering et al. (1991).

40 60 80 100 120 140 160 180

Distance (km)

Figure 7 Difference (postconvection minus undisturbed) in model-computed CO tracer concentrations for the August 3, 1995, ABLE 2A squall line. Increases in CO are noted throughout the main updraft region and anvil, and decreases are seen in the downdraft region. From Pickering et al. (1991).

15% became involved in a rotor-type circulation located behind the convective updrafts. In each of these cases tracer transport calculations were performed for CO, NOv, 03, and hydrocarbons. The three-dimensional version of the GCE model has also been run for the June 10-11, 1985, PRE-STORM case and for the September 26, 1992, event from TRACE-A. Figure 8 shows the redistributed CO from the rural Oklahoma boundary layer as simulated by the model-generated three-dimensional wind field. Free tropospheric 03 production enhancement of a factor of 2.5 for Oklahoma rural air and for the Oklahoma City case were calculated, while with a pristine preconvective upper troposphere an enhancement of a factor of 35 was estimated for the Manaus, Brazil, case.

Stenchikov et al. (1996) used the two-dimensional GCE model to simulate the North Dakota storm observed by Poulida et al. (1996). This storm showed the unusual feature of an anvil formed well within the stratosphere. The increase of CO and water vapor above the altitude of the preconvective tropopause was computed in the model. The total mass of CO across the model domain above this level increased by almost a factor of 2 during the convective event. Downward transport of ozone from the stratosphere was noted in the simulation in the rear anvil. Wang et al. (1995) simulated a tropical convective storm observed during CEPEX using the cloud dynamics and cloud transport models of Wang and Chang (1993).

PRE-STORM June 10-11th

CO (110 ppbv) isosurface at 4 hours

Figure H 1 so surface of CO mixing ratio (1 lOppbv) computed for the June 10, 1985. PRE-STORM squall line over Oklahoma using the three-dimensional GCE model. Measured rural CO mixing ratios used as initial conditions.

The simulated cloud tower extended into the lower stratosphere and a widespread anvil was produced. Intense mixing of boundary layer air into the cloud resulted in low ozone throughout the tower and the anvil. Stratospheric air with high-ozone mixing ratios was brought into the upper portion of the anvil. The model did not show any significant transport of boundary layer gases into the stratosphere.

0 0

Post a comment