Free Troposphere

deep convective flux into BL +29.8 to +37.5

fair weather cumulus deep convective flux to free trop. -39.6 to-63.9

T T oxidation of CO I oxic

biogenic anthro. T

emissions emissions deposition

oxidation of HC +7.8

total horizontal flux out (E)

Units 10 kg/month Area 32.5-50N, 90-105W

Figure 9 Regional boundary layer CO budget for the central United States (32.5°N to 50°N; 90°N to 105°W). Note magnitudes of upward and downward deep convective transport components. Question marks signify that relative amounts of CO flux due to shallow convection and synoptic-scale systems are unknown. From Thompson et al. (1994).

aqueous chemistry, scavenging, and vertical mixing in the chemistry model of Chang et al. (1987). The vertical distribution of cloud microphysical properties and the amount of subcloud-layer air lifted to each cloud layer are determined using a simple entrainment hypothesis (Walcek and Taylor, 1986). Vertically integrated 03 formation rates over the northeast United States were enhanced by ~ 50% when the in-cloud vertical motions were included in the model.

Wang et al. (1996) simulated the September 26-27, 1992, TRACE-A mesoscale convective systems (MCS) and the June 10—11, 1985, PRE-STORM squall line with the NCAR/Penn State Mesoscale Model (MM5; Grell et al., 1994; Dudhia, 1993). Convection is parameterized as a subgrid-scale process in MM5; two convective parameterizations were tested in the Wang et al. (1996) work. These were the Grell (1993) and Kain and Fritsch (1993) schemes. Mass fluxes and detainment profiles from these schemes were used along with the three-dimensional wind fields in CO tracer transport calculations for the two convective events. The time-evolving tracer fields in the upper troposphere are different in the tropical MCS and the midlatitude squall line. The nearly stationary tropical system produced regions of large upper tropospheric CO that moved very little in the horizontal by the end of the 24-h simulation, whereas enhanced upper tropospheric CO propagates with the relatively fast moving midlatitude squall line. Using a grid size of 25 to 30km, the parameterized subgrid vertical transport represented 48% (Grell, 1993) and 41% (Kain and Fritsch, 1993) of the total upward transport in the tropical case and 64% (Kain-Fritsch) in the midlatitude case. Pickering et al. (1996) demonstrated that the MM5 convective transport (Fig. 10) reproduced the observed factor of three enhancement of upper tropospheric CO and that over several days downwind transport the enhanced upper tropospheric 03 precursor mixing ratios allowed 03 production to proceed at a rate ~ 4 times faster than would have occurred in undisturbed air. The U.S. Environmental Protection Agency (EPA) has developed a Community Multi-scale Air Quality (CMAQ) modeling system that uses MM5 with the Kain-Fritsch convective scheme as the dynamical driver (Ching et al., 1998).


Convective transport in global chemistry and transport models is treated as a subgrid-scale process that is parameterized typically using cloud mass flux information from a general circulation model (GCM) or global data assimilation system. Jacob and Prather (1990) simulated the distribution of radon-222 over North America using a three-dimensional CTM driven with meteorological fields from the NASA Goddard Institute for Space Studies (GISS) GCM II (Hansen et al., 1983), having a horizontal resolution of 4° x 5° and 9 layers in the vertical. Simulation of convective transport in the CTM follows the scheme used in the GCM to transport momentum, sensible heat, and moisture. The model gave a reasonable simulation of radon-222 observations over the United States, but with some significant discrepancies that were traced to problems in the GCM meteorology. Improved simulations of transport have been obtained using a newer convective parameterization of Del Genio and Yao (1988).

} j , £




■' JS&k


i" *<

|L ,„,


% -


-\f w

i fc55 „ -5

OH 0

.. -45

50 100 150 200 250 300 350

Average: 236 (obs.), 197 (model) Initial: 9J

50 100 150 200 250 300 350

CO mixing ratio(ppbv) at Z=9,5 km

CO mixing ratio(ppbv) at Z=11 km

Figure 10 MM5 simulation result for CO tracer following TRACE-A mesoscale convective events. Shown are CO mixing ratios at 1200 UT September 27, 1992, at altitudes 9.5 and 11 km. Region shown is fine-grid (30-km resolution) domain of MM5 simulation. Includes grid-scale and subgrid transport. From Pickering et al. (1996).

While GCMs can provide data only for a "typical" year, data assimilation systems can provide "real" day-by-day meteorological conditions, such that CTM output can be compared directly with observations of trace gases. The NASA Goddard Earth Observing System Data Assimilation System (GEOS-l DAS; Schubert et al., 1993) provides archived global data sets for the period 1980-1995, at 2° x 2.5° resolution with 20 layers in the vertical. Convection is parameterized with the relaxed Arakawa-Schubert scheme (Moorthi and Suarez, 1992). Pickering et al. (1995) showed that the cloud mass fluxes from GEOS-l DAS are reasonable for the June 10-11, 1985, PRE-STORM squall line based on comparisons with the GCE model (cloud-resolving model) simulations of the same storm (Fig. 11). In addition, the GEOS-l DAS cloud mass fluxes compared favorably with the regional estimates of convective transport for the central United States presented by Thompson et al. (1994). Allen et al. (1996a,b) have used the GEOS-l DAS data to drive global CTM calculations for radon-222 and for CO. However, Allen et al. (1997) have shown that the GEOS-l DAS overestimates the amount and frequency of convection in the tropics and underestimates the convective activity over midlatitude marine storm tracks.

Mahowald et al. (1995) investigated the behavior of seven different cumulus parameterization schemes in deriving convective transport from meteorological analysis data sets that did not routinely archive cloud mass fluxes. The derived convective transport was used in a column model and showed that the resulting vertical profile of trace gases was highly sensitive to the parameterization used.

June 10-11,1985

June 10-11,1985

Figure 11 Profiles of cloud mass flux for June 10-11, 1985, PRE-STORM squall line computed by GEOS-l DAS and by the GCE model. From Pickering et al. (1995).

Figure 11 Profiles of cloud mass flux for June 10-11, 1985, PRE-STORM squall line computed by GEOS-l DAS and by the GCE model. From Pickering et al. (1995).

Rasch et al. (1997) have described use of the output from the NCAR (National Center for Atmospheric Research) Community Climate Model (CCM3) in a chemical transport model. This CTM uses results from the CCM3 convective parameter-izations [(Zhang and McFarlane (1995) penetrative convection parameterization and the Hack (1994) scheme for shallow convection)].

0 0

Post a comment