U.S. Dept. of Commerce / NOAA / OAR / ERL / GFDL *Disclaimer  

 

2. ATMOSPHERIC PROCESSES

GOALS

2.1 RADIATIVE TRANSFER

ACTIVITIES FY98

2.1.1 Solar Benchmark Computations

Computations of the suite of overcast sky cases initiated in A97/P98 were completed. These cases account for absorption by all the major gaseous constituents and clouds, with the calculations being performed using the "line-by-line + doubling-adding" (LBL+DA) technique. The high cloud case was performed using ice crystal optical parameters based on a formulation developed by Q. Fu of Dalhousie University. Additional benchmark cases were generated for both ice and water clouds, including geometrically thick systems, in order to focus on the absorption effect of large in-cloud vapor contents.

A collaborative research project investigating the direct radiative forcing due to sulfate aerosols, and which utilized GFDL's LBL+DA algorithm, has been completed (1569).

2.1.2 Characteristics of Solar Fluxes

An analysis of the near-infrared flux disposition in plane-parallel overcast sky cases has been completed (ga). Using the updated catalog of benchmark computations, this study has now been extended to include the complete solar spectrum. In particular, analysis has focused on whether the dependence of solar absorption on drop optical properties and location in the atmosphere, seen to be critical for the near-infrared spectrum, is also important for the total spectrum fluxes. This inquiry has particular relevance for notions about the invariance of solar absorption in clear and overcast atmospheres and for proposed methods that seek to obtain surface solar flux from satellite-measured top-of-the-atmosphere flux.

From Fig. 2.1a, it is apparent that the total solar flux absorbed in overcast atmospheres, besides differing considerably from the clear-sky case, is not the same when clouds at different altitudes (e.g., 500-600 mb versus 800-900 mb case), different optical depths (d), different size/shape of particles (CS-stratocumulus, CL-cumulonimbus, FU-ice crystals), and different geometrical thicknesses (e.g., CS 180-900, 300-900 and 500-900 mb cases versus CS 800-900mb) are considered. An especially interesting consequence is the relationship between cloud forcing at the surface and at the top of the atmosphere (TOA). Observational studies suggest that this ratio is nearly invariant, with an inferred value of ~1.5, while results from some parameterization model studies suggest a value of ~1.0. Fig. 2.1b shows the LBL+DA computed ratio for various model cloud cases, as a function of the drop optical depth, cloud type and location in the atmosphere. Note that the ratio is between 1.0 and 1.5 for all the cases considered. For d > 5, there is little dependence of this ratio on d for a specific cloud type and location. Notable decreases with increasing d are seen to occur for d < 5. Of greater significance is the variation that occurs among the cloud cases for a fixed d, with absorptivity differences > 0.1 possible for the same cloud type located in the middle versus the lower troposphere. The computational results suggest that either the measurements and/or their interpretations leading to the inference of an invariant value are incorrect, or that solar radiative transfer in realistic atmospheres is somehow inconsistent with known radiative transfer principles. The results also suggest that the total solar surface forcing may not be simply estimated from a knowledge of the TOA forcing, without an accompanying knowledge of the properties of the cloud present.

2.1.3 Parameterization

The effect of the increased stratospheric heating associated with the new solar parameterization (A96/P97) was investigated utilizing the latest version of the N30 SKYHI GCM. Fig. 2.2 shows the change in the annual-mean stratospheric temperature obtained from a difference of 10-year runs with the old and the new shortwave schemes. Temperature increases of more than 2 K occur throughout the stratosphere, with increases of more than 4K occurring in the upper stratosphere, due mostly to an improved accounting of the solar CO2 heating. This improvement brings the modeled stratospheric temperatures into a better quantitative agreement with satellite observations, thus overcoming a deficiency noted for earlier model results using the older shortwave parameterization, i.e., the long-acknowledged cold bias of the model's middle and upper stratosphere (A97/A98). The changes in high latitudes do not appear to be statistically significant due to the high degree of interannual variability there.

The new shortwave algorithm has now been tested against reference computations for the case of clouds containing ice crystals (in contrast to water drops, as reported in prior

years). The parameterization reproduces the "exact" results reasonably well and hence should be suitable for application in models that consider ice crystal occurrences explicitly.

2.1.4 Clear-Sky Shortwave Radiative Flux

An investigation of the possible factors involved in accounting for the observed zonal and interannual variations of the clear-sky solar flux has been completed (1547).

Using the new shortwave parameterization, sea surface albedo values, and the known distributions of the atmospheric gases, the top-of-the-atmosphere outgoing solar flux was computed over oceanic regions for clear skies (i.e., devoid of aerosols and clouds). The results were compared with the corresponding ERBE observations. It was found that the computed results underestimate the observed reflected flux by up to 12 W/m2 in the annual-mean. The most likely reason for this bias is the reflective effects of aerosols in the real atmosphere. Calculations including aerosols suggest that both the naturally occurring sea-salt and anthropogenic sulfate, biomass and dust particles are contributing to the reflected flux in clear skies. Thus, it is concluded that the ERBE clear-sky flux measurements contain signatures of the aerosols' presence over the world's oceans.

PLANS FY99

The "benchmark" cases will continue to be analyzed for complex multi-layer clouds. The longwave radiation algorithm will be modified to account explicitly for non-gray aerosol and cloud absorption. Further computations of the effect of aerosols on the top-of-the-atmosphere and surface fluxes will be performed, and the quantitative roles of natural and anthropogenic aerosols will be analyzed.

2.2 CONVECTION-CLOUDS-RADIATION-CLIMATE INTERACTIONS

2.2.1 Cumulus Parameterization

ACTIVITIES FY98

A major, but elusive goal in parameterizing cumulus convection for general circulation models is to capture the interaction between deep cumulus towers and the mesoscale and synoptic-scale cloud systems in which they are embedded and in whose development they play important roles. The latter, spatially extensive cloud systems are major regulators of the earth's radiant energy system. A new conceptual framework for dealing with this problem has been developed (1133). The frequently used concept of mass-flux cumulus parameterizations was extended to include a statistical treatment of the vertical velocities in cumulus convection. These vertical velocities can be used to drive microphysics at physically appropriate scales, and thereby provide representations of the interactions between cumulus-scale clouds and larger-scale, radiatively important cloud systems (1349).

Preliminary testing of a parameterization built on this new framework in the SKYHI GCM is nearly complete. The parameterization produces plausible distributions for both deep cumulus towers and upper-tropospheric mesoscale circulations associated with deep convection. The mass fluxes associated with these upper-tropospheric stratiform systems are significant fractions of those in the deep cumulus towers, in accordance with observational studies. The vertical penetration of heat and moisture produced by this cumulus parameterization is significantly deeper than that associated with the saturated adiabatic adjustment currently used in SKYHI.

PLANS FY99

Final testing of the parameterization will be completed. The impact of cumulus convection on the climate and climate sensitivity of SKYHI will be evaluated. Cloud-convection-radiation interactions will be examined, including the relationship between deep convective systems and other upper-tropospheric cloud systems (1502).

2.2.2 Limited-Area Nonhydrostatic Models

ACTIVITIES FY98

An ongoing study of deep convection and its associated mesoscale circulations using the Lipps-Hemler (885) cloud-system model (fy) focused on the structure of tropical convective systems, the role of these systems as sources of heat and sinks of moisture for larger-scale flows, the relationship of the behavior of these systems to the properties of larger-scale flows, and the transport and transformation of atmospheric sulfate by these systems.

The cloud-system model was used to study deep convection and associated mesoscale circulations for a composite easterly wave in the tropical eastern Atlantic Ocean (fy) and a westerly wind burst in the tropical western Pacific Ocean. The model successfully captures the observed life cycles of deep convection, which begin with deep cumulus towers and end with mesoscale stratiform systems in the upper troposphere. These studies revealed that very strong interactions between radiation and convection can occur under some circumstances. The extent to which these radiative-convective interactions proceed depends strongly on the manner in which ice microphysics is treated in the cloud-system model and, in particular, the rate at which ice crystals are removed by sedimentation.

The cloud-system model was also used to simulate sulfate-cloud interactions in the western tropical Pacific (1537, 1543). The model included a new, detailed sulfate aerosol microphysics scheme, designed to estimate the effects of sulfate on cloud microphysics and radiative properties, and the effects of deep convection on the transport and redistribution of aerosol. Model results showed that the presence of convective clouds has a significant impact on the spatial and temporal distribution of sulfur species. An instantaneous simulated total sulfate concentration in the cloudy domain in the western Pacific is illustrated in Fig. 2.3 for a case with high sulfur from the Pacific Exploratory Mission West B (PEM West B) observations during February-March 1994. Increases in sulfate as observed during PEM West B cause a significant decrease of the effective radius of cloud droplets, an increase in the cloud droplet concentration, and an increase of the shortwave reflection above clouds.

PLANS FY99

Treatments for microphysics and radiation in the cloud-system model will be modified. The current radiative treatment of ice crystals as small, spherical particles will be replaced with an approach in which the ice particles are treated as larger crystals. Techniques such as geometric ray tracing will be used to model radiative transfer through ice (1502). The formulation for ice sedimentation will probably be changed from its current form, in which cloud ice does not fall unless snow is also present (fy). Most likely, an approach will be used in which cloud ice falls at terminal speeds which depend on ice concentration. Shortwave and longwave fluxes at the top of the atmosphere and the surface observed in recent field experiments in the tropical western Pacific, will be used to constrain the radiative and microphysical parameterizations in the cloud-system model.

The relationships described to date between sulfate concentration and the radiative properties of cloud systems (1537, 1543) are based on empirical data. However, using a new approach to the treatment of particle sizes in the cloud-system model, in which several moments of the size distributions for microphysical species are treated as prognostic variables, a more general approach may be possible and will be investigated. This strategy will be pursued in FY99.

2.2.3 Radiative-Convective Equilibria with Explicit Moist Convection

ACTIVITIES FY98

Simulations of radiative-convective equilibrium in a horizontally homogeneous atmosphere are being conducted with a cloud-resolving nonhydrostatic model. These integrations address fundamental issues in the theory of moist convective turbulence and cloud-radiative feedbacks. The studies are being carried out with both two- and three-dimensional models at various resolutions with a vertically stretched grid and cyclic lateral boundary conditions.

Recent focus has returned to detailed studies of the energy and entropy budgets of the two-dimensional model. This analysis has led to the significant finding that an important way in which kinetic energy is dissipated in the model is through the force exerted by precipitation. From a microscopic perspective, the surprising implication is that a large part of the kinetic energy dissipation occurs in shears in the immediate vicinity of falling hydrometeors and is not a consequence of a turbulent cascade to small scales. A simple theory has been developed that relates the energy dissipation associated with falling hydrometeors to the precipitation rate and the average distance that a hydrometeor falls. The resulting estimate for the global mean of this term, which has been neglected in all discussion of energy dissipation in the atmosphere, is surprisingly large: 1.5 W/m2. This is comparable to classical estimates of the frictional loss of kinetic energy from large scales that occurs in the planetary boundary layer.

The entropy budget also has surprising features. In particular, kinetic energy dissipation is not the dominant irreversible source of entropy; rather, the evaporation of water into unsaturated air is the dominant term. This complexity in the entropy budget and in the sources of kinetic energy dissipation cast doubt on the viability of several recent theories that use the entropy budget to constrain convective velocity scales and the average CAPE (convective available potential energy) of the tropical atmosphere.

Insights into the energy and entropy budgets, as well as other features of moist-convective turbulence, has been gained by varying the value of the latent heat of condensation and studying the consequences for radiative-convective equilibria. This series of integrations includes the limit of zero latent heating, in which water vapor is simply a passive tracer in a dry radiative-convective model. Convection in the dry limit is found to penetrate and overshoot radiative equilibrium at the tropopause much more dramatically than in the case with realistic latent heating, forming a sharp inversion at the tropopause.

Preliminary radiative-convective equilibrium calculations have been performed with the three-dimensional model with 2 km resolution and a 128 x 128 grid, fully interactive cloud-radiation feedback with two different values of surface temperature (25 and 30C) and two different treatments of the ice microphysics. Differences in microphysics cause dramatic changes in the upper level cloud cover and in the dynamics of the convection in the upper troposphere. This sensitivity to the microphysical assumptions is currently the most serious limitation to the utilization of this model for a variety of purposes.

PLANS FY99

The study of radiative-convective equilibria as a function of the strength of the latent heating will be pursued, as well as the energy and entropy budget analyses. Experimentation with the three-dimensional model will continue, as well. In addition, the study of inhomogeneous statistically steady states will begin with a model of a Walker cell in a relatively small domain in which the surface temperature possesses an east-west gradient.

2.2.4 Physical Parameterization Tests in Single Column Models

ACTIVITIES FY98

The parameterization of cloud microphysics and convection is a source of large uncertainty in simulations by general circulation models. A creative way to test physical parameterizations is to simulate the vertical structure of temperature, water vapor and clouds at a single point. In these so-called "single-column" models (SCMs), the effects of large-scale dynamics must be specified. If they are specified from observations, then the predictions of the SCM, which are the result of the interaction between the specified forcing and the physical parameterizations, can be compared to observations taken at the same point.

Tests of the model parameterizations in the flexible/modular modeling system (3.1.1) have been carried out with forcing specified by data taken from the Atmospheric Radiation Measurement (ARM) program's Oklahoma site in July 1995. The results of these tests have been submitted to an intercomparison project of SCMs from other major research labs (NCAR, ECHAM, etc.). Results indicate that the GFDL SCM's temperature tends to drift warm in simulations relative to ARM data, but that this error is common to many of the other SCMs.

SCMs can also be used to demonstrate the sensitivity of model simulations to physical parameterizations. An example of this is shown in Fig. 2.4, which shows how the predicted amount of cloud liquid plus ice differs when the convection scheme is changed from the Relaxed Arakawa-Schubert (RAS) parameterization to Moist Convective Adjustment (MCA). The simulated cloud field occurs at a higher altitude (lower pressure) when RAS is used. This reflects the fact that RAS is a significantly more penetrative scheme and detrains water vapor at a higher altitude than MCA.

PLANS FY99

Participation in the SCM program of ARM will continue. It is anticipated that tests of the SCM will occur for other seasons using data from ARM. In addition, as new parameterizations become available, they will be evaluated in the SCM with ARM data.

2.2.5 "Predicted" Cloud Distributions in SKYHI

ACTIVITIES FY98

The SKYHI GCM integration using the new cloud prediction scheme with "part black" "high clouds (A97/P98) has been extended to include a period of 10 model years. Comparison of this GCM simulation with a 10-year simulation using the standard prescribed-cloud distribution indicates that the predicted-cloud approach leads to increased diabatic heating in the upper troposphere, and yields significant improvements in the simulation of upper tropospheric temperatures and lower stratospheric water vapor mixing ratios.

Figure 2.5 displays the 10-year average differences in the zonally averaged temperature and fractional change in the mixing ratio between the two simulations. Changes in the zonally averaged fractional cloud amount are similar to January results (A97/P98). The increase in temperatures in the tropical upper troposphere (100-150 hPa) exceeds 4 K, and attains ~8 K near the equator. This confirms the robustness of the results previously reported (A97/P98). The H2O mixing ratio increases in the stratosphere by at least a factor of 2, with increases of a factor of 4 in much of the lower stratosphere. This moistening represents a significant improvement in the GCM water vapor simulation of this region. The results demonstrate that the global stratospheric water vapor amount depends on cloud-convective-radiative interactions occurring in the troposphere.

PLANS FY99

SKYHI GCM calculations with increased horizontal and vertical resolution are planned. Tests of the effects of modified cloud prediction schemes and boundary layer formulations on the cloud, moisture and temperature distributions will be pursued.

2.3 ATMOSPHERIC CHEMISTRY AND TRANSPORT

ACTIVITIES FY98

2.3.1 Tropospheric Photochemistry

HOx radicals are the key to chemical reactivity in the troposphere and acetone is considered to be an important source of them in the upper troposphere, where mixing ratios of water vapor are small. Reactions of acetone and its derivatives were added to the box model that is used to calculate the tables of ozone production and destruction (A97/P98). The GFDL Global Chemical Transport Model (GCTM) ozone simulation run with tables that included the additional source of HOx from acetone showed that the impact of introducing acetone is greatest at the top tropospheric level of the model (190 mb) and decreases quickly for lower levels. The simulated impact on OH at this level for the month of July varies between 10% in the moist region over southeast Asia to more than 50% in some regions of the

Southern Hemisphere. At lower altitudes, the acetone-induced change in OH is much smaller, generally less than 10% by 315 mb.

The associated changes in the chemical ozone production follow the absolute changes in the concentration of OH, so these changes are also greatest in the tropics and subtropics at 190mb, where it increases by 0.25 to 1.00 ppbv/day, amounting to a 50 to 100% increase. At 315 mb, the largest increase in ozone production due to acetone is only 10-20%. The influence on ozone itself is much less, with the increased production leading to only 5-10% more ozone in the tropics at 190 mb, and progressively less at lower altitudes. In summary, while acetone plays a major role in HOx chemistry of the upper troposphere, the resulting impact on ozone levels is minor.

2.3.2 Fast Photochemical Solver Development

In current GCTMs with in situ chemistry, 75-90% of the computational time is spent determining the chemical tendency terms for the continuity equations of the transported species. A recent effort has begun to examine an approach whereby, after a one-time off-line expenditure of significant computational resources, on-line chemical tendency terms, which are equivalent to those produced by a full on-line chemistry, can be calculated very rapidly.

The High Dimensional Model Representation (HDMR) method was used to construct the Fully Equivalent Operational Model (FEOM) of CO-CH4-NOy-H2O chemistry for calculating the O3 chemical tendencies. The HDMR method maps out the relationship between sets of high dimensional input variables and output variables through explicit algebraic equations which can be rapidly evaluated. These explicit algebraic equations constitute the core of the FEOM, in contrast to the ordinary differential equations which normally must be solved to generate the chemical tendencies. The FEOM can then be used in place of the current look-up tables in the GCTM for obtaining the time-dependent chemical production and destruction terms of ozone, as well as OH and the O3 net chemical tendency, for all model levels, all months, every 10 band of latitude, and all tropospheric values of H2O, CO, NOx and O3. This represents a first step in the development of a FEOM for the complex non-methane hydrocarbon chemistry of the full troposphere, including the polluted continental boundary layer. While calculated off-line, the FEOM is expected to reproduce the chemical tendencies resulting from a fully time-dependent on-line chemical model, while requiring much less on-line computation.

2.3.3 Tropospheric Carbon Monoxide

Carbon monoxide (CO) plays both a primary and secondary role in determining air quality. In urban areas with large CO emissions, concentrations can be large enough to cause health problems. However, even far from its emission source, CO affects air quality as part of catalytic ozone production and destruction cycles, and as an important sink for OH radicals. The lifetime of CO varies from about 15 days to more than one year, which is long enough to transport CO long distances and affect air quality throughout the hemisphere. Thus, the chemistry and transport of CO, as well as its emissions, must be considered when examining air quality globally.

The GFDL three-dimensional GCTM was used to examine the evolution and distribution of carbon monoxide, and to investigate the specific role of each contributing source: fossil fuel emissions, biomass burning, oxidation of biogenic hydrocarbons, and oxidation of methane. An understanding of the global CO budget will aid analyses of tropospheric ozone, evaluations of emission estimates for contributing sources, and should lead to improvements in the calculation of global hydroxyl concentrations. Model results were compared with observations from the 33 CMDL/NOAA global cooperative flask sampling network stations which measure CO. Over most of the globe there is good agreement, with only 14% of seasonal averages differing by more than 25% between model and observations. For all of these outliers (18 points out of 132), the model overestimated CO concentrations, thus revealing a high bias to the simulation. An investigation into the specific causes for model discrepancies is underway. In the Southern Hemisphere, an overestimation of CO from biogenic hydrocarbons is a likely source of error.

2.3.4 Tropospheric Reactive Nitrogen

The 11-level GFDL GCTM has been used to simulate the tropospheric fields of NOx, PAN, HNO3, NOy, and NO3-deposition, which are then evaluated with available observations from surface stations and aircraft missions and analyzed for the contributions of individual natural and anthropogenic sources. This simulation includes all known sources of tropospheric NOx, transports three families (nitrogen oxides, PAN, HNO3), and employs pre-calculated chemical conversions among them. With the exception of outliers due to either anomalous local observations or local errors in the GCTM's NOx source and simulated precipitation, the GCTM's HNO3 wet deposition is highly correlated with observations and clearly captures the observed spatial patterns of wet deposition. However, there appears to be a significant positive bias (~20%) in the simulated U.S. deposition, though not for the rest of the world. The GCTM's NOx fields are in reasonable agreement with the large majority of the observations, show no systematic global biases, display the observed vertical profiles, have high levels (~1 ppbv or greater) in the polluted boundary layer (BL), and the very low values in the remote BL. At Mauna Loa Observatory, while summer and fall simulated NOx is clearly in deficit, the PAN is in surplus and the sum of the two agrees well with observations. In general, the level of agreement between simulation and observation is as good as the agreement between separate, but simultaneous observations of NO, NOx or NOy.

The BL sources of NOx, which are primarily anthropogenic, exceed free tropospheric (FT) emissions, which are primarily natural, by more than 8:1 and completely dominate lower tropospheric NOx levels (except for very remote regions where BL and FT sources have a comparable impact). At higher altitudes, the smaller FT sources play a larger role and generally dominate in the upper troposphere. A more detailed analysis of individual natural and anthropogenic NOx source contributions is shown in Fig. 2.6. Anthropogenic emissions from surface fossil fuel combustion and biomass burning dominate in the lowest layers with a significant contribution from biogenic emissions in remote regions that do not have biomass burning. In the lower troposphere, surface fossil fuel combustion dominates in the Northern Hemisphere (NH) extratropics, while lightning is the primary source in the tropics and biomass burning plays a major role in much of the Southern Hemisphere (SH). In the middle troposphere, lightning dominates in both the tropics and much of the midlatitudes, while transported surface fossil fuel emissions supply the NH high latitudes, and transported emissions from biomass burning do the same in much of the SH. In the upper troposphere, lightning dominance extends farther poleward, while stratospheric injection is the major source in the NH high latitudes. Remnants of biomass burning, along with stratospheric injection dominate in the SH high latitudes. Although seldom dominant, aircraft emissions do play a significant role in the upper troposphere and lower stratosphere of the NH extratropics.

In an effort to assess the impact of pre-calculated, monthly-mean, zonally-averaged (off-line) NOy interconversion rates, a GCTM simulation of NOx, HNO3, and PAN was conducted with an interconversion rate table calculated with the chemical box model for the same set of variables as the rates of ozone production and destruction (3.1.2, A97/P98). The interconversion rates were interpolated at every GCTM time-step using the GCTM's time varying fields of NOx and O3, and monthly averaged fields of simulated CO and observed H2O. An additional source from isoprene oxidation was added to PAN formation, and photolysis was added to PAN loss. The primary impact of the new treatment was higher average OH, due mostly to higher observed values of middle and upper tropospheric water vapor. As a result, the NOx chemical lifetime was significantly shortened and the global scaling of the lightning source was increased from 4TgN/yr to 10TgN/yr. There was also a significant zonal variation in the NOx lifetime in the lower troposphere, though there was much less variation in the upper troposphere, as a result of the switch to on-line chemistry. The NOx and PAN data from this simulation were then compared with the available data from aircraft campaigns and surface sites. The NOx values were generally lower than observed with approximately 60% of the compared points within 50% of the observed values. The level of agreement with observations was comparable to the earlier PAN simulation (1372), but not quite as good as it was for the earlier NOx simulation with the pre-calculated chemical interconversions. In general, the switch to on-line chemistry did not have a major impact, though the improved water vapor fields did increase upper tropospheric OH and lead to a significant increase in the lightning source.

The sensitivities of the NOx simulations to uncertainties in the observed specific humidities, simulated OH, and simulated ozone were tested in separate GCTM simulations. A 50% reduction in the specific humidities reduced the OH levels by 30-50% in the middle and lower troposphere and by 10-40% at 190 mb. The resulted in NOx increases of 10-30% in the upper troposphere and 20-40% in the middle and lower troposphere. A 25% reduction in ozone increased NOx by 15-20% in the upper and middle troposphere and by 5-10% in the lower troposphere. Increasing ozone by 25% lowered NOx levels by about 10% in the upper and middle troposphere, and by 5-10% in the lower troposphere. The NOx simulations were significantly less sensitive to uncertainties in ozone than to uncertainties in specific humidity and OH.

2.3.5 Tropospheric Ozone

A previous GFDL GCTM simulation of ozone (A97P98), which agrees well with ozone observations, has been further improved by the inclusion of acetone in the HOx chemistry (3.1.1), and by an improvement in the seasonality of ozone production in the polluted lower troposphere. The GCTM was then employed to study two of the key issues in tropospheric chemistry: a) the relative roles of stratospheric injection and tropospheric chemical production in the ozone budget; and b) the nature of tropospheric chemical production. A three-tracer experiment was designed to assess the fractional contributions to tropospheric ozone from stratospheric ozone (Blue), ozone produced in the background troposphere (Green), and ozone produced in the polluted lower troposphere (Red). The source of Blue ozone was direct injection across the model's tropopause. Green ozone was produced by CO-CH4-NOy-H2O chemistry in the background (NOx < 200 pptv) troposphere, and Red ozone was produced by a NOx dependent parameterization in the polluted boundary layer (NOx > 200 pptv). All three ozone loss processes, photochemical destruction, nighttime NOx based loss, and dry deposition, were partitioned among the three tracers in each grid-box according to their individual mixing ratios. Globally, Green ozone from the background troposphere is the largest single component (54%), while Blue ozone transported from the stratosphere is next in importance (38%), with Red ozone from the polluted continental contributing only 8%. However, Red ozone does dominate in its source region, the polluted continental boundary layer, and also accounts for ~30% of the ozone in the summertime boundary layer over the North Atlantic. Generally, Blue ozone dominates in the extratropical Southern Hemisphere, with Green ozone dominating in the tropical troposphere. Both Blue and Green have comparable roles in the extratropical Northern Hemisphere. Red ozone and its associated complex hydrocarbon photochemistry do not play major roles in most of the troposphere.

The GFDL GCTM simulation of tropospheric ozone previously used to study the impact of human activity on tropospheric ozone (1445; 3.1.6, A97/P98) has also been used to investigate the role of photochemistry in the winter-spring ozone maximum in the NH midlatitude free troposphere (770 mb-240 mb; 30N-60N). Free tropospheric ozone mass slowly builds up in the winter and early spring, with net chemistry and transport playing comparable roles. Winter and early spring conditions are favorable to net ozone production for two reasons. First, winter conditions (cold, low sun angle, and dry) reduce HOx and lower the level of NOx needed for chemical production to exceed destruction. Second, throughout the winter, NOx increases above normally net-destructive levels in the remote atmosphere. As a result, net production in the midlatitude NH free troposphere maximizes in early spring while NOx is still relatively high, while the ozone chemical balance point is still relatively low, and while increasing insolation is speeding up photochemistry. Conceptually, the net ozone production is associated with an annual atmospheric "spring cleaning" in which accumulated winter-time NOx is removed via sunlight driven OH oxidation. Furthermore, human activity was found to have a major impact on both the simulated levels of tropospheric ozone and the role of chemistry in the NH midlatitude where anthropogenic NOx emissions dominate. In that region, modern ozone levels increased by ~20% in the winter and ~45% in the spring. At the same time winter-spring chemistry switched from net destructive to net productive, the winter-spring balance between transport and chemistry switched from transport dominance in pre-industrial times to the present parity, and the pre-industrial February maximum progressed to March-April. Estimated 2020 levels of NOx emissions were found to lead to even greater net production and to push the O3 spring maximum later into April-May.

2.3.6 GCM Simulation of Carbonaceous Aerosol Distribution

ACTIVITIES FY98

A model of the carbonaceous aerosol cycle has been implemented in the GFDL SKYHI model. The fossil fuel emissions of carbonaceous aerosol (both black and organic) have been calculated on the GFDL SKYHI GCM N30 (3 x 3.6) grid. These emissions have been derived from a 1 x 1 database of black and organic carbon emissions1 using a simple linear interpolation scheme.

Tests have been performed to determine the best transport scheme for use with aerosols. A comparison has been made between the 4th-order and Lin-Rood transport schemes and the Lin-Rood scheme was found to be more appropriate in cases where the emission fields have large gradients. Since this is the case for the emission of carbonaceous aerosol, the Lin-Rood transport scheme has been utilized for the model integrations.

The carbonaceous aerosol model that has been implemented is similar to that of Cooke and Wilson2. Briefly, the scheme uses two tracers for each carbonaceous aerosol component. These represent, respectively, the hydrophobic and hydrophilic parts of the carbonaceous aerosol. The sinks for these aerosols are dry and wet deposition, respectively. In addition, there is also a transformation from hydrophobic to hydrophilic aerosol with a time constant of 1 day. This can be thought of as a representation of the aging of the aerosol, which causes it to become more hydrophilic with time.

The SKYHI model was run for 15 months and the results for the final 12 months were compared to surface concentration and wet deposition measurements. Initial comparisons indicate a reasonable agreement between the modeled and measured concentrations of the BC aerosol. Fig. 2.7 shows the monthly mean global distribution of black carbon (BC) (ngm3) in July as simulated by the model. The distribution of BC near the surface is similar to that of sulfate3 with large concentrations near the major source regions of northeast United

States, eastern Europe and China. It is also apparent that the entire Northern Hemisphere is affected by this aerosol species, with plumes of aerosol advecting off the east coasts of North America and Asia, and northward from Europe. The Southern Hemisphere is not impacted as severely, although there are areas with significant amounts of black carbon in the atmosphere.

The seasonality of the modeled BC concentrations has been compared with inferred BC concentrations from long-term measurements of light absorption at the CMDL/NOAA sites of Sable Island, Bondville, Barrow and Mauna Loa. The concentrations of this aerosol type are reproduced fairly well for most of the year at Sable Island. At Bondville, the modeled BC concentrations do not account for all of the light absorption in the fall. This may be due to noncarbonaceous absorbing aerosols present at this time of year. The modeled and measured seasonal cycles at Barrow and Mauna Loa agree fairly well, although the model's peak values at these sites are too low. This may be due to stratified transport in the Arctic in the case of Barrow, which is not well represented in models, and to long-range transport of dust from Asia in the case of Mauna Loa. Longer term model runs will be needed to determine whether the differences are due to interannual variability in the atmospheric transport.

PLANS FY99

The contributions to tropospheric ozone from stratospheric ozone, ozone produced in the relatively clean troposphere and ozone produced in the polluted boundary layer will be quantified, and the dependence of each source of ozone on season and location will be analyzed.

The GFDL GCTM simulated tropical distribution of tropospheric ozone will be analyzed, coincident with existing observational programs in three remote regions: the tropical South Atlantic, the Indian Ocean, and the South Pacific Ocean near Samoa.

The GCTM study of carbon monoxide will be completed, with an investigation into the pre-industrial distribution. The GCTM simulations will be used to examine cross-border air pollution in Asia, with a focus on possible changes into the next century.

GCTM simulations of NOx, CO and O3 resulting from Indonesian fires will be compared with available observations from the region for both normal and El Niño years, and the impact of the extensive fires in 1997 on local, regional and global air quality will be quantified.

Sensitivity of the global distribution of NOx and O3 to rapidly changing emissions in Asia will be investigated with the GFDL GCTM. Specific goals include quantifying the export of NOx and ozone from Asia's polluted BL to the free troposphere, and investigating the impact of this export on the balance of ozone production and destruction in the Central Pacific.

A five-tracer simulation with interactive NOx, HNO3, PAN, O3, and CO will be conducted with the GFDL GCTM using a FEOM representation of the fully coupled in situ CONOx-O3-H2O chemistry. The resulting fields will be compared with available observations and previous independent simulations of NOy, O3 and CO.

Development of the Fully Equivalent Operational Model (FEOM) for the complex non-
methane tropospheric chemistry will be initiated. Specifically, the FEOM development will focus on the carbon-bond mechanism (CBM-IV).

Sulfate emissions will be integrated into the SKYHI model in the coming year. Global-scale calculations of the distribution of this aerosol component will be compared to measured global distributions. In addition, a comparison of the relative abundance of sulfate and carbonaceous aerosols will be conducted and the direct radiative forcing of both these aerosols will be calculated. Sensitivity studies with respect to various microphysical and hydrologic assumptions will also be performed.

2.4 ATMOSPHERIC DYNAMICS AND CIRCULATION

2.4.1 SKYHI Model Development

ACTIVITIES FY98

The development of the new standard SKYHI model has continued. A preliminary version of this model containing all of the code upgrades from the last two years (including new long- and short-wave radiation, surface albedo, prognostic cloud and unstable vertical mixing packages) was produced and made available to users. This version is now being used by all SKYHI investigators who are developing and testing additional new model features, so that intercomparison of results between developers may be more easily performed. Some deficiencies still present in the new parameterizations have been identified and efforts are underway to remedy them. Computational upgrades to the model continue, as does the development of generalized initialization and analysis programs.

The parcel trajectory and stratospheric chemistry packages which had been available with older versions of SKYHI are now included in this new release, providing investigators with access to the latest physics parameterizations and model upgrades.

The new SKYHI source has been run on both the Cray T90 and Cray T3E systems. Evaluation of the performance and scaling characteristics of SKYHI on the T3E has identified some model design features which may need modification in order to obtain better performance on distributed memory architectures. Efforts are underway to test the effects of making these changes.

PLANS FY99

Work will continue on the evaluation and improvement of the new scientific parameterizations. The major focus of code development will shift to the task of incorporating the essential functionality of SKYHI into the new GFDL grid-point model, currently under development.

2.4.2 SKYHI Control Integrations and Basic Model Climatology

ACTIVITIES FY98

Control integrations were continued with various versions of the SKYHI model. These now include a 10-year integration of the 3 x 3.6 40-level model, two years with the 1 x 1.2 40-level version, two years with the 1 x 1.2 80-level version, two months with the 1 x 1.2 160-level version, and seven months with the 0.33 x 0.4 40-level version. The 160-level and the 0.33 experiments represent integrations at unprecedented spatial resolution (at least for global climate simulation models). Preliminary analysis of these new integrations has concentrated on the simulation of the large-scale circulation in the middle atmosphere. Overall, the extratropical simulation appears to improve with finer grid spacing, but the effects of increasing horizontal resolution are more significant than vertical resolution. Comparing the 40-level versions, the SH winter polar temperature near 1 mb (~50 km) has a cold bias relative to observations of ~70C in the 3 version of the model, ~35C in the 1 version and less than 10C in the 0.33 version. By contrast, the change from 40 levels to 80 or 160 levels seems to improve this cold bias in the 1 model by only a few C. In the tropics, however, the changes in vertical resolution were found to lead to much more significant effects on the simulation. In particular, the 80-level and 160-level models develop very strong shear zones in the tropics which descend with time, rather like the observed QBO (Quasi-Biennial Oscillation) shear zones (2.4.3).

Results from the long integration of the 3 x 3.6 version of the model have been incorporated in the middle atmospheric GCM intercomparison organized by the SPARC (Stratospheric Processes and their Role in Climate) initiative of the WCRP (World Climate Research Program) .

PLANS FY99

Analysis will continue of the high-resolution integrations, including an attempt to compare some of the small-scale and high-frequency variations in the simulated middle atmosphere with available observations.

2.4.3 Spontaneous QBO-like Tropical Wind Oscillations in SKYHI Simulations

ACTIVITIES FY99

As noted earlier, a strong sensitivity of the tropical circulation to the vertical resolution in SKYHI has been discovered. The top panel of Fig. 2.8 shows the height-time evolution of the equatorial zonal-mean zonal wind during two years of the control integration with the 40-level 1 version of SKYHI. The winds near the stratopause undergo a semiannual oscillation with somewhat realistic features, but the winds in the tropical lower stratosphere are nearly constant (typical behavior in most GCMs). The bottom panel of this figure shows the equatorial wind in the 80-level version of the model, in which the equatorial winds and the vertical shears are much stronger. Above about 1 mb, the semiannual variation still dominates, but in the lower and middle stratosphere a longer period oscillation is evident. The mean wind evolution in the lower and middle stratosphere displays both the downward propagation of wind regimes and the concentration of vertical shear into narrow zones that are characteristic of the QBO in the real atmosphere. However, the period of the wind oscillation appears to be about one year, which is much shorter than that of the real QBO. From the limited two-year simulation available, it is not clear whether the period of the lower stratospheric wind variations is exactly one year or not. The model was rerun with fixed equinoctial conditions, and the mean wind evolution in the lower and middle stratosphere was found to be quite similar to that in the seasonally-varying run, showing that this oscillation is indeed a spontaneous "QBO-like" "variation rather than being forced by the annual cycle. Detailed analysis of the momentum budget in these experiments has shown that the main driver for the mean flow accelerations in the QBO-like oscillation comes from vertically-propagating waves.

PLANS FY99

Further analysis of the tropical wind variations in the fine vertical resolution models will continue. Additional experiments will be carried out to examine the dependence of the strength and period of the equatorial wind oscillation on the model resolution, subgrid-scale dissipation and cumulus parameterization scheme.

2.4.4 Low-Frequency Variability of Simulated Stratospheric Circulation

ACTIVITIES FY98

Long control integrations of the 40-level 3 x 3.6 SKYHI model have revealed an impressive degree of interannual variability in the NH winter middle atmospheric circulation. The standard deviation of the monthly-mean zonal-mean temperature compares quite well with observations (1275). A striking aspect of the variability is the significant degree of quasi-decadal variation produced by the model. This occurs even in the absence of interannual variations in the prescribed sea surface temperatures or in external forcings (solar flux, volcanic aerosols, etc.). These results are interesting in light of recent claims of significant quasi-decadal variations in the real NH stratosphere, and caution against automatically attributing such long-period variability to effects of external forcing or decade-scale variations in sea surface temperatures.

Some recent studies with simplified models suggest that the zonal-mean zonal momentum in the subtropics may provide some effective interannual memory for the high-latitude stratospheric circulation. To investigate the role of this effect in producing interannual variations in the SKYHI model, the long integrations are being repeated with an arbitrary constraint on the zonal winds in the tropical and subtropical stratosphere.

PLANS FY99

Further analysis of the control and constrained integrations will continue.

2.4.5 Horizontal Spectra from High-Resolution SKYHI Integrations

ACTIVITIES FY98

The horizontal kinetic energy spectra from the high horizontal resolution SKYHI runs have been analyzed in detail. The spectra at tropospheric levels show two distinct regimes with relatively steep spectral slopes for scales larger than ~500 km and significantly shallower at smaller scales. This corresponds well with observations which demonstrate the existence of a shallow "mesoscale" spectral regime in the troposphere. Fig. 2.9 shows a comparison between one-dimensional power spectra in the midlatitude upper troposphere in the 0.33 SKYHI simulation and results of observations taken from many thousands of commercial aircraft on long flights. The agreement is quite good over most of the range and it appears that the 0.33 SKYHI model is the first global climate model to successfully simulate a significant portion of the mesoscale regime.

Analysis of the detailed spectral kinetic energy budget is proceeding. At each wavenumber the contributions to the energy from horizontal advection, vertical advection, pressure gradient and dissipation terms are being computed. An interesting finding is that, at

the 0.33 resolution, the dissipation of kinetic energy by the vertical subgrid-scale diffusion actually dominates that from the horizontal diffusion for most of the spectral range.

PLANS FY99

The analysis of the kinetic energy budget will continue, along with analysis of the available potential energy and tracer variance budgets. A key issue to be addressed is whether the dominant energy cascades in the mesoscale regime are upscale or downscale.

2.4.6 Parameterized Gravity Wave Drag in the SKYHI Model

ACTIVITIES FY98

A version of the Alexander-Dunkerton gravity wave drag parameterization scheme has been tested in the 3 40-level version of SKYHI. Earlier simplified relaxation experiments (1441) have been repeated using the current version of the model to provide a comparison for some preliminary off-line tuning of the parameters.

PLANS FY99

Work towards the efficient implementation and appropriate tuning of the gravity wave scheme will continue. The sensitivity of the residual circulation and the large-scale tracer distribution to the incorporation of the gravity wave drag will be examined.

2.4.7 Effect of Advection Schemes in Simulating Stratospheric Transport

ACTIVITIES FY98

The age of stratospheric air has been computed in the 3 x 3.6 version of the SKYHI model by three methods: 1) integrating particle trajectories initialized at the tropical tropopause; 2) employing a Green's function approach; and 3) simulating a passive tracer with a linearly increasing source at the surface. Trajectories have been computed using both the model "sigma" levels and potential temperature as vertical coordinates. In the Green's function and the linear source experiments, four gridded advection schemes have been used: the SKYHI 2nd and 4th order centered-difference schemes, the Lin-Rood upstream scheme, and the NCAR semi-Lagrangian (SLT) scheme. This is the first time that such a wide variety of transport schemes (ranging from Eulerian through semi-Lagrangian to fully Lagrangian) have been simultaneously employed in a GCM. Besides computing the age spectrum, the Green's function experiments, in which a passive tracer is advected from a concentrated source in multiyear integrations, constitute a more stringent test of the various advection schemes than the usual tests involving idealized one- or two-dimensional flows and short integrations of three-dimensional simulations of tracers with chemical sources or sinks.

The experiments have demonstrated that a wide range of mean age distributions can be obtained within the same GCM depending on the choice of the advection scheme. Sigma coordinate trajectories exhibit spurious cross-isentropic motions and, as a result, lead to ages younger than those produced by potential temperature coordinate trajectories. Except in the lowermost tropical stratosphere, the ages computed using gridded advection schemes are too young compared with measurements, most likely as a result of diffusion errors inherent in all Eulerian and semi-Lagrangian advection schemes. This conclusion is supported by the fact that the longest ages overall are obtained using potential temperature coordinate trajectories, which suffer from minimal numerical diffusion. The SLT scheme produces the youngest ages as a result of spurious vertical motions akin to those produced by the sigma-coordinate trajectories. Since the calculation of these trajectories forms the essence of the SLT scheme, this finding, and the contrast between these two trajectory schemes, suggest that the performance of this scheme might be improved if its trajectories were calculated in theta coordinates. Among the gridded schemes, the centered-difference schemes produced the longest ages (and closest to those observed), suggesting that their lack of numerical diffusion outweighs their dispersion errors in this application. The large sensitivity of the computed ages to the model numerics calls for caution in interpreting features of model-generated mean age distributions solely in terms of the model simulation skill.

2.4.8 GCM Chemical Simulation with an Imposed Tropical Quasi-biennial
Oscillation

ACTIVITIES FY98

A 48-year simulation using the 3 x 3.6 SKYHI model with an imposed QBO zonal momentum source (A97/P98) has been analyzed (ex). More than 16 years of this experiment have now been repeated in a version of the model with a detailed prognostic photochemical code. This model produces a simulation of the QBO in ozone near the equator that compares very well with observations (A97/P98). Recent analysis has focussed on the details of the transport and chemical response in the subtropics and midlatitudes. One crucial feature apparent in the observations of stratospheric ozone is the tendency for the annual cycle of ozone and the QBO to be strongly coupled. The power spectrum of total ozone variations near the equator has a strong peak at the QBO frequency (1301), but there is also a strong peak in the subtropics near ~20 months which corresponds to one of the beat frequencies between the QBO and the annual cycle. Fig. 2.10 shows the spectrum of zonally averaged total column ozone at 19.5 latitude computed from 14 years of Nimbus-7 TOMS (Total Ozone Mapping Spectrometer) data. The results are shown separately for the components of the ozone symmetric and antisymmetric between the NH and SH. The symmetric component is dominated by a peak centered near 27 months, but the antisymmetric component has a more prominent peak near 20months. Also shown is the same analysis applied to the ozone in the SKYHI QBO experiment. The agreement with observations is quite impressive and this is an indication that the model is able to capture the QBO modulation of the annual cycle of trace constituent transport in the subtropics.

PLANS FY99

Analysis of the transport and chemical effects of the QBO in the integration will continue. The SKYHI model experiment will be repeated using a prescribed zonal-mean flow evolution based on detailed radiosonde and satellite observations.

2.4.9 GCM Simulation of Long-Term Variations in Stratospheric
Tracer Concentration

ACTIVITIES FY98

Analysis of the results for the ozone concentration in the SKYHI integration with an imposed QBO has revealed the occurrence of long-period variations, particularly in the NH mid and high latitudes. Fig. 2.11 shows the least-squares linear trend of the zonal-mean total column ozone in the SKYHI simulation. Values are shown for 14 years (omitting the first two years of the experiment as spinup period) and also for just the last 10 years. The results reveal a positive total ozone trend which is particularly pronounced in the NH midlatitudes. The magnitude of the trend in the NH midlatitudes turns out to be more than 50% of the magnitude of the (decreasing) trend actually seen in the 14-year (1979-92) TOMS dataset. This suggests that natural internally-generated transport variability may be able to generate decade-length trends that are of comparable magnitude to those that have been observed, and thus that natural variability might possibly account for a significant portion of the trends seen in NH midlatitude ozone data. Somewhat similar long-period variability is seen in the lower stratospheric N2O field in the 48-year simulation (ex).

PLANS FY99

The full chemistry version SKYHI simulation will be continued in order to see how the long-period ozone variations develop. The transport mechanisms contributing to the long-period trace constituent variations will be analyzed in detail.

2.4.10 Observational Studies Using Radiosonde Data

ACTIVITIES FY98

Six years of high-resolution radiosonde data at Payerne in Switzerland have been obtained. These data are now being analyzed for gravity wave signals as part of a worldwide project (coordinated by SPARC) to establish a gravity wave climatology for the lower stratosphere. In addition, a study has commenced of the statistics of cold extremes in the winter lower stratosphere temperature data. This will be useful in determining how much mesoscale temperature variations might contribute to activation of low temperature heterogeneous chemistry that is thought to be responsible for significant mid- and high-latitude ozone loss.

A review of pre-1950 balloon observations of stratospheric winds at near-equatorial sites was undertaken (1566). Scattered observations exist as early as 1908, and these data strongly suggest that the QBO has been a fairly stable feature of the equatorial circulation for at least the last century.

PLANS FY99

The analysis of high-resolution radiosonde data will continue and the results will be incorporated into the worldwide SPARC gravity wave analysis project.

2.4.11 Review Papers on Middle Atmospheric Dynamics

ACTIVITIES FY98

Two review articles on aspects of middle atmospheric dynamics have been prepared. One is a historical survey of the development of the field with a focus on the period before about 1975 (gl). The other is a review of the state-of-the-art in observations and theory of the general circulation of the tropical stratosphere and mesosphere (gt).

2.4.12 Dynamics of the Martian Atmosphere

ACTIVITIES FY98

Viking Infrared Thermal Mapper (IRTM) data provide the foundation for much of the knowledge of the current martian climate. The IRTM data were collected over a period in excess of two Mars years and contain a wealth of information on the spatial and temporal variation of surface and atmospheric temperatures on diurnal to seasonal time scales. In particular, the 15 m channel radiance measurements have provided brightness temperatures (T15) for a deep layer of atmosphere centered at the 0.5 level (~25 km). The latitude and seasonal variation of T15 is shown in Fig. 2.12. The T15 data have indicated a prominent seasonal modulation of global mean temperature that is the result of a large seasonal variation in atmospheric dust loading. The effect of aerosol heating is also indicated by the large diurnal variation in martian T15 temperatures, particularly during global dust storm periods. Episodic global dust storms, such as the two observed in 1977, represent a significant interannual component in the climate description.

Work has continued in comparing the IRTM T15 observations with Mars GCM simulations of diurnal and seasonal variability. The recent effort to reconcile the spatial pattern of diurnal variability evident in the T15 data with that predicted by the GCM has resulted in the identification of a systematic error in the retrieved 15 m radiances which were evidently additionally sensitive to radiation from the surface. The surface emission resulted in a 15 K temperature bias in the estimate for midday tropical temperatures for relatively clear sky conditions when the contrast between atmosphere and surface temperature was largest. The downward revision of IRTM temperatures during the Northern Hemisphere spring and summer seasons also implies a lower water saturation level and thus, an increased possibility for water ice cloud/dust aerosol interaction. This interaction may be an important aspect of the seasonal variation in atmospheric dust opacity, as has been suggested by one-dimensional

model simulations employing coupled microphysics, radiative transfer, and turbulent mixing of aerosol (hl).

PLANS FY99

Ice cloud microphysics and improved radiation parameterizations will be incorporated in the Mars GCM in order to examine the influence of microphysical/thermal/dynamic feedback mechanisms on martian climate.

2.5 CLIMATIC EFFECTS DUE TO ATMOSPHERIC SPECIES

ACTIVITIES FY98

2.5.1 Lower Stratospheric Ozone and Temperature Trends

Model and observational results indicate that the observed loss of ozone in the global lower stratosphere since ~1980 has led to a cooling of that altitude region (1193, 1394). In recent years, the WCRP SPARC Program has initiated a project to obtain a thorough quantitative assessment of the temperature trends in the entire global stratosphere. This project seeks to: a) assess the temperature change in the stratosphere from a variety of available measurements; and b) evaluate the degree to which these can be attributed to changes in atmospheric species and/or natural variations. Fig. 2.13 shows the 50 and 100 hPa trends over the 1979-1994 period drawn from radiosonde and analyzed datasets, and those derived from satellite data. The Berlin, Angell, Russia and UK/Raob plots represent different radiosonde-based datasets, while the rest of the non-satellite data come from analyses that either do not involve a GCM (NOAA/ Climate Prediction Center or CPC) or that do involve a GCM (NASA's Goddard Space Flight Center; NOAA/NCEP's re-analyses, or "Reanl"). The satellite data consists of the Microwave Sounding Unit (MSU Channel 4) and the thermal infrared Stratospheric Sounding Unit (SSU 15X channel, labelled here as Nash). The MSU senses signals from a region ~50-150 hPa with a peak at ~90 hPa, while the SSU peaks at 50 hPa and has a broad span encompassing ~20-250 hPa.

In the northern midlatitudes, all the datasets exhibit coherence with regard to the magnitude of the cooling. They are also coherent with respect to the statistical significance of the cooling (not shown). In the low latitudes, there is some divergence between the various datasets with regard to the magnitude of the cooling. In the Southern Hemisphere, which does not have as dense a radiosonde monitoring network, the datasets do not exhibit the same degree of agreement. In addition, the annual-mean results for the Southern Hemisphere are not statistically significant at any latitude for most of the datasets, although the Antarctic springtime and the midlatitude Southern Hemisphere during the early part of the year do exhibit a statistically significant cooling (1394). At the high southern latitudes, warming is found at the 50 hPa level, with cooling at 100 hPa. (Note that the lower stratosphere, where considerable ozone loss occurs, is at a higher pressure level at this latitude.) The satellite data contains a signal not only from the lower stratosphere, but also nonnegligible components from the region above the lower stratosphere and, more particularly, the upper troposphere. In general, the satellite trend at the NH midlatitudes agrees well with the other datasets in terms of the magnitude and with respect to statistical significance. However, at low latitudes, the satellite trends are smaller than for the radiosonde, likely due to signals from upper troposphere which mask the pure lower stratospheric trend. In the high southern latitudes, the satellite results yield a cooling that is similar to the 100 hPa trend shown in panel (b). At the high northern latitudes, the entire 50-100 hPa layer undergoes a large cooling in contrast to the non-satellite

data estimates for the high southern latitudes. Model studies of ozone depletion effects (1193, 1394) indicate a warming trend above the region of cooling in the lower stratosphere at high southern latitudes due to dynamical changes. This differs from the high northern latitudes, where the warming tendency is quite weak and cooling occurs over a deeper vertical extent. The fact that the 50 and 100 hPa trends are of opposite sign in the high southern altitudes, and that the low latitude satellite signature may well contain a substantial input from the upper troposphere, illustrates the necessity of extra care in quantitative intercomparisons of radiosonde and satellite trends in the lower stratosphere. It should be noted that the current results regarding lower stratospheric temperature trends constitute a principal highlight of the recently concluded WMO/UNEP 1998 Ozone Assessment.

Simulations of the effects of stratospheric ozone loss (A97/P98) were analyzed further. When ozone losses are distributed over a broader vertical extent than just the lower stratospheric region, the cooling of the lower stratosphere is altered only modestly in magnitude. However, the strength of the warming seen above the ozone depletion regions can be considerably reduced, especially in the high southern latitudes (a result of the change in the stratospheric residual circulation).

The net radiative flux changes at the tropopause due to depletion of lower stratospheric ozone, computed using the SKYHI GCM and Fixed Dynamical Heating (FDH) method, have been compared. Differences in the longwave flux for the two approaches are considerable and arise due to the fact that: a) the GCM's lower stratospheric temperature change differs from the FDH; and b) the troposphere is not held fixed in the GCM, as in the FDH approximation. Thus, caution must be exercised in comparing FDH-computed radiative flux changes with those observed. This is because flux observations already incorporate a measure of the altered dynamics (and thereby the climate system's response to the ozone losses) and thus cannot be simply related to a FDH computation.

2.5.2 Radiative Forcing Due to Tropospheric Aerosols and Ozone

The global radiative forcing features due to sulfate and black carbon aerosols have been examined (1529). A review has been conducted (1523) of the state-of-the-art models used to compute the forcing due to scattering and absorption by atmospheric aerosols, as well as the present methods used to estimate the absorption ability of these aerosols, e.g., single-scattering albedo. It is concluded that it is not possible to generalize the range of the single-scatter albedo at present. Further, most of the measurement techniques suffer from a lack of calibration checks and a systematic overestimation of light absorption coefficients. On the modeling side, the necessity of including humidity-dependent effects on light extinction cannot be overemphasized.

The radiative effects of changes in tropospheric ozone since preindustrial times has been examined (1555) and the results are illustrated in Fig. 2.14. The forcing due to ozone has been found to be as important as that due to sulfate and carbonaceous aerosols (A97/P98). The requirement of a radiative-dynamical equilibrium in the stratosphere for the determination of the tropospheric ozone radiative forcing, as demanded by the strict FDH definition, is not found to be a significant factor, unlike the case of the lower stratospheric ozone changes.

The radiative forcing due to an external mixture of sulfate and soot aerosols (A97/P98), combined with that due to tropospheric ozone, is shown in Fig. 2.15 which reveals substantial

gradients in the geographical distribution of the climate forcing. It is interesting to note that while the forcing from each of the three species has been deemed by the IPCC to be important in its own right, the global-mean is a small residual (-0.12 W/m2). Nevertheless, the large positive and negative values occurring in the geographical distribution are relevant when considering regional climate effects.

PLANS FY99

The effects due to stratospheric ozone changes will be studied further, with emphasis on middle and upper stratospheric temperature trends. As part of the SPARC investigation of temperature trends, observational records will also be analyzed. Investigations concerning the global radiative forcing due to trace gases and aerosols will continue.

2.5.3 Radiative Effects of Aerosol-Cloud Interactions

ACTIVITIES FY98

Aerosol and water drop size distributions in clean maritime, continentally influenced maritime, and polluted clouds were modeled using data from the Monterey Area Ship Track Experiment (MAST) and a size- and composition-resolved externally-mixed aerosol model4. These results were used to compute the corresponding Mie scattering parameters, which then act as an input to a delta-Eddington exponential-sum-fit solar radiation algorithm (A96/P97). The resulting radiation quantities are analyzed to investigate, on a microphysical level, the corresponding changes in albedo, absorption, and transmission as a consequence of cloud evolution.

Analysis of a clean maritime cloud influenced by a ship track plume reveals an almost order of magnitude increase in cloud optical depth at 550 nm wavelength, with drops in the >7 micron size range (derived from the plume particles that are activated) making the largest contribution. The increase in optical depth translates into a near doubling of the otherwise clean maritime cloud's visible albedo.

The influence of a clean maritime cloud on atmospheric absorption above and inside the cloud as a function of wavelength has also been investigated. The results show a 17 W/m2 increase in above-cloud absorption at wavelengths less than 2.0 microns due to reflection by the cloud and absorption by ozone and water vapor above the cloud. There is a 26 W/m2 increase in the in-cloud absorption at wavelengths greater than 0.7 microns due to scattering by the cloud drops and absorption by water vapor inside the cloud, and to absorption by the cloud drops themselves.

PLANS FY99

Analyses and comparisons of the results for the clean maritime cloud with those for continentally influenced clouds will be conducted. An emphasis will be placed on distinguishing the contributions to the cloud radiative properties by particles of different sizes and compositions. These results, plus additional idealized simulations, will be used to assess parameterizations of supersaturation, aerosol activation, and the "Twomey effect." An attempt will be made to extend the study toward an evaluation of the aerosol-cloud interaction and the "indirect" aerosol effect in one-dimensional radiative convective and three-dimensional general circulation models.

2.5.4 Radiative Forcing due to Stratospheric Aerosols

ACTIVITIES FY98

Using the new shortwave parameterization (A96/P97) code (which includes stratospheric aerosols), calculations of changes in the downward and upward fluxes and heating rates were performed for volcanic aerosol size distributions typical of those observed after the El Chichon and Pinatubo eruptions. Line-by-line (LBL) calculations were also performed for these size distributions for the entire shortwave spectral range (1-50,000 cm-1). The LBL calculations enable a quantitative evaluation of the contributions from different spectral regions and for stratospheric aerosols of different sizes. Downward and upward fluxes and the heating rates obtained from the parameterization were compared with the LBL results for the near IR, visible, and UV spectral regions, separately. The net and the spectral upward and downward fluxes obtained from the parameterization are about 5-10% larger than the LBL results. The heating rates obtained from LBL calculations were found to be consistently higher than the parameterized values, with most of the bias coming from the near IR and visible spectral regions.

Volcanic aerosols in the stratosphere can produce significant long-term solar and infrared radiative perturbations, producing a radiative and dynamical response in the climate system. A spectral-, space-, and time-dependent set of zonally monthly averaged aerosol parameters for two years after the Pinatubo eruption (June 1991-May 1993) was obtained5 and formatted for the SKYHI latitude grids and for the 26 spectral bands of the new shortwave parameterization. Two sets of SKYHI GCM integrations were performed with predicted clouds and monthly varying SSTs, one with aerosols and the other without ("control"). Differences between these runs illustrate the warming of the stratosphere due to the aerosols and the evolution of this warming over the two-year period.

A simple one-dimensional model was developed to study the time evolution of a volcanic aerosol layer and its associated particle size distribution based on the microphysical processes of growth, coagulation, diffusion and sedimentation. The simulated size distributions for the Pinatubo volcanic event compare reasonably well with available observations. Using the shortwave parameterization code, the changes in the fluxes and heating rates were obtained for the aerosol size distributions simulated at various times. These changes are larger during the initial stages following the eruption, and decrease with time. Since the aerosol size distribution contains more large particles and higher number concentrations initially, the evolution of the radiative changes is consistent with the simulated microphysics.

Using a newly developed radiative-convective model (RCM) with a mixed layer ocean, and incorporating typical volcanic stratospheric aerosol parameters, equilibrium runs have been performed with and without aerosols, and the changes in the global, annual-mean surface and stratospheric temperatures have been determined. Results show a peak stratospheric warming of about 3 K (for a aerosol = 0.1) for a Pinatubo type aerosol size distribution located between 15 and 30 km, which is quite consistent with other model estimates.

PLANS FY99

Efforts are underway to incorporate the non-gray longwave radiative effects due to particulates in eight longwave bands from 560 to 1400 cm-1. This will be implemented in both the RCM and SKYHI GCM radiation codes and used to study the infrared radiative perturbations due to the presence of volcanic aerosols. Using the time-dependent Pinatubo aerosol dataset, an ensemble of runs will be performed to estimate the changes in the net radiative fluxes, temperatures in stratosphere and troposphere, and their temporal variations and other changes in tropospheric climate. The volcanic aerosol effects will be compared with responses due to other types of forcings, such as the lower stratosphere cooling in mid- and high latitudes due to ozone depletion. Further study of the microphysical and radiative impacts of volcanic aerosols is planned by coupling the aerosol microphysical model to the RCM. Also, the aerosol microphysical model will be incorporated into the SKYHI GCM, and the evolution and transport of Pinatubo aerosols at various stratospheric altitudes will be studied.

  1. Cooke W.F., C. Liousse, H. Cachier, and J. Feichter, Construction of a 1 x 1 fossil fuel emission dataset for carbonaceous aerosol and implementation in the ECHAM-4 model, submitted to J. Geophys. Res., 1998.
  2. Cooke W.F., and J.J.N. Wilson, A global black carbon aerosol model, J. Geophys. Res., 101, 19395-19409, 1996.
  3. Kasibhatla P., W.L. Chameides, and J. St. John, A three-dimensional global model investigation of seasonal variations in the atmospheric burden of anthropogenic sulfate aerosols, J.Geophys. Res., 102, 3737-3759, 1997.
  4. Russell, L.M., and J.H. Seinfeld, Size- and composition-resolved externally-mixed aerosol model, Aerosol Sci. Technol., 28, 403-416, 1998.
  5. Stenchikov, G.L., I. Kirchner, A. Robock, Hans-F. Graf, J.C. Antuna, R. Grainger, A. Lambert, and L.W. Thomason, Radiative forcing from the 1991 Mount Pinatubo volcanic eruption, Report no. 231, MPI, 1-40, 1997.


Back to Table of Contents

*Portions of this document contain material that has not yet been formally published and may not be quoted or referenced without explicit permission of the author(s).