Table of Contents 

3. MIDDLE ATMOSPHERE DYNAMICS AND CHEMISTRY

To understand the interactive three-dimensional radiative-chemical-dynamical structure of the middle atmosphere (10-100 km), and how it influences and is influenced by the regions above and below.

To understand the dispersion and chemistry of atmospheric trace gases.

To evaluate the sensitivity of the atmospheric system to human activities.

3.1 ATMOSPHERIC CHEMISTRY AND TRANSPORT

W.L. Chameides*         J.L. Moody***
S.-M. Fan                    W.J. Moxim
M. Gloor                      S. Oltmans****
P. S. Kasibhatla**       L.M. Perliski
A. Klonecki                   J. Sarmiento
H. Levy II                     J.J. Yienger*****
J.D. Mahlman

*Georgia Institute of Technology
**Duke University
***The University of Virginia
****Climate Monitoring and Diagnostics Laboratory/NOAA
*****University of Iowa

ACTIVITIES FY97

3.1.1 In situ Photochemical Module Development

A version of SKYHI with a detailed treatment of stratospheric photochemistry has now been run for over 10 model years. The photochemical scheme is described in detail in A96/P97. These simulations indicate that SKYHI is capable of generating generally realistic distributions of important stratospheric trace constituents. The version of SKYHI employed to perform the photochemical calculations does not have a parameterization of gravity wave drag, therefore the cold pole bias still exists in the calculations and has an effect on the trace species distributions through temperature-dependent photochemical processes. However, a simulation was performed with a version of SKYHI in which the zonal winds in the mesosphere were forced to their climatological values, thus warming the winter high-latitude temperatures to very close to the observed values. This resulted in a significant improvement in the winter high-latitude stratospheric ozone simulation. The results of this experiment is discussed in more detail below.

A scheme to calculate photolysis rates on-line in SKYHI has been incorporated into the GCM. This scheme results in more realistic sensitivities of photolysis rates to surface albedo, overhead ozone column and temperature. In addition, the scheme permits aerosol scattering processes to influence calculated photolysis frequencies. The previous photolysis rate computation used in SKYHI allowed sensitivity to solar zenith angle changes only.

In addition to a new photolysis rate calculation, a new photochemical scheme which includes a treatment of halogen chemistry (chlorine and bromine compounds) has been implemented in SKYHI. The previous scheme did not contain halogen chemistry for computational economy during model development, and it was also the strategy adopted for modeling the "pre-industrial" stratosphere when chlorine and bromine abundances were a small fraction of present day. The new chemical scheme includes a diagnostic calculation of type I and II polar stratospheric clouds. Background stratospheric aerosol abundances are currently prescribed from stratospheric ice aerosol gas experiment II (SAGEII) observations, and latitude-dependent percent weights of sulfuric acid are assumed. Currently, a test run that includes gas phase chlorine chemistry is being run. Bromine and polar ozone depletion processes will be enabled in future runs.

The existing polar stratospheric cloud (PSC) parameterizations assume that type I PSCs are composed of nitric acid trihydrate, but the composition and phase of polar stratospheric clouds is still uncertain. Recent observational evidence and laboratory measurements suggest that it is likely that at least some type I PSCs are actually liquid ternary solutions of water, nitric acid and sulfuric acid rather than nitric acid trihydrate. An algorithm to calculate the abundances, volumes and composition of these aerosols has been obtained from Carslaw and co-workers at the Max-Planck Institut fur Chemie in Mainz, Germany and is being added to the one-dimensional version of SKYHI's halogenated photochemical scheme for testing purposes and eventual incorporation into the model.

3.1.2 Off-line Photochemical Module Development

Extending the CO-CH₄-NO_x calculations of the O₃ chemical tendencies discussed in A96/P97, the tables previously used by the Global Chemistry Transport Model (GCTM) to obtain the chemical production and destruction terms for ozone were expanded by two new variables: water vapor and carbon monoxide. The resolution of the tables was increased by creating the tables every 10 degrees latitude (from 80S to 80N) and every two months. The ozone chemical tendencies are now calculated by a four-way interpolation routine that takes as input instantaneous value of NO_x, the monthly averaged CO (both previously computed with the same model), the instantaneous values of O₃ from the current simulation, and the monthly averaged H₂O. The tropospheric water vapor concentrations were obtained from a global set based on sonde measurements. These water data were checked against the tropospheric column relative humidities obtained from satellite measured water vapor radiances. In regions where the discrepancies between the two sets were larger then 25%, the sonde data was scaled to match the satellite data (the differences were mostly over regions where few or no sondes were flown) (ed).

These changes were incorporated into the box model described in A96/P97, which was then used to calculate the interconversion rates between NO_x, HNO₃, and PAN needed for the three tracer NO_y GCTM simulations. Five conversion rates (NO_x to HNO₃, HNO₃ to NO_x, NO_x to PAN, PAN to NO_x and fraction of PAC not reacting with NO₂) were saved in tables similar to those for ozone production and destruction. Preliminary NO_y simulations with the new rates were conducted.

3.1.3 Advection Schemes

A model experiment was designed to test the ability of several different advection schemes to produce realistic distributions of water vapor and atmospheric tracers such as N₂O. The schemes tested were SKYHI's default advection scheme, a centered difference scheme that is 4th order accurate in the vertical and 2nd order accurate in the horizontal, a pseudo-4th order vertical and horizontal centered difference scheme, a hiearchy of upstream difference schemes developed by Lin and Rood, and a semi-lagrangian scheme developed by Williamson and Rasch.

The first set of advection scheme experiments involve running a simulation of water vapor in SKYHI for each advection scheme for three model years. The stratospheric water vapor in the tropics calculated using the centered difference schemes generally shows a great deal of vertical structure compared to the Lin and Rood scheme which appears to be significantly drier. This vertical structure is believed to be due either to dispersion errors associated with the schemes, or to different representations of horizontal transport processes. (A model simulation in which water vapor is transported with the Williamson and Rasch scheme has not yet been completed.) In addition, a predictive cloud scheme is currently being tested in SKYHI (2.4.1). Preliminary results show that the tropical tropopause temperature increases substantially, which should increase the water vapor in the lower tropical stratosphere. In light of this, the water vapor advection experiments may be re-run with the new version of SKYHI.

The second set of advection scheme experiments involves a three-year SKYHI simulation of N₂O for each advection scheme. The results of these experiments suggest that after three model years, the N₂O calculations for the pseudo-4th order scheme and a higher-order Lin and Rood scheme are highly correlated, with both schemes producing horizontal N₂O gradients near the polar vortex edge equally well. The Williamson and Rasch and mixed 4th-2nd order centered difference schemes produce much shallower N₂O gradients across the vortex, implying that more N₂O is diffused into the vortex than for the pseudo-4th order and Lin and Rood schemes. This effect is roughly equivalent to running a lower spatial resolution calculation, and has possible implications for polar vortex ozone depletion chemistry. In addition, the tropical vertical profile of N₂O for the calculation using the Williamson and Rasch scheme shows higher amounts of N₂O at high altitudes than the other schemes, suggesting that this advection scheme is more rapidly transporting trace species out of the troposphere.

3.1.4 Tropospheric Reactive Nitrogen

A study which constrains the global emission of NO_x by lightning within a relatively narrow range of 3-5 TgN/year was first discussed in A95/P96 and has recently been published [1410]. While at the low end of current estimates, which range as high as 20 TgN/year, this level still leaves lightning as the largest natural source of NO_x outside of the atmospheric boundary layer and actually controls present NO_x levels in the tropics. Another study quantifying the impact of PAN chemistry on the global distribution of NO_x and identifying its control of NO_x levels over the remote Oceans was also first discussed in A95/P96 (1372).

Work continued on improvements to the six independent NO_x sources used in the GCTM (Fig. 3.1). While preindustrial NO_x emissions occurred primarily in the tropics, present NO_x emissions occur predominantly in the northern midlatitude metro-agro-plexes of North America, Europe, and Asia. Tropical emissions may play a significant, though relatively minor role. Future growth in NO_x emissions is expected to be dominated by growth in fossil fuel combustion in Asia. A new collaboration established with the Center for Global and Regional Environmental Research (CGRER) at the University of Iowa yielded an improved emissions

estimate for Asia. The fossil fuel component was augmented with a new emission inventory for Asia from the CGRER based "RAINS ASIA" research initiative. With new time-dependent fossil fuel and lightning NO_x sources, and with continuing refinements to the group's soil-biogenic source, the GFDL GCTM now operates with the most advanced "state-of-the-art" emissions sources available to the global modeling community.

3.1.5 Stratospheric Ozone

A ten-year SKYHI simulation with HO_x-O_x-NO_x chemistry has shown that the photochemical scheme is stable and gives reasonably good agreement with observed ozone. Comparisons of SKYHI O₃ in the upper troposphere/lower stratosphere (300-25 mb) with long-term average ozonesonde measurements show reasonable agreement between the model and measurements. At high latitudes in the Northern Hemisphere winter, the calculated ozone agrees with the average ozonesonde measurements to within the measurement error. In the tropics, the calculations are systematically slighter higher than the ozonesondes, but still within measurement error. However, at midlatitudes in the summer Southern Hemisphere, the model ozone is higher than observed by as much as 35% in the lower stratosphere. The overprediction of ozone by the model is likely due to the lack of a polar ozone depletion scheme in the simulation since the measured lower ozone is influenced by equatorward transport of low ozone air from polar regions. Future photochemical simulations performed with a version of SKYHI that treats ozone "hole" chemistry should improve the agreement with observations significantly.

Comparisons of SKYHI stratospheric ozone with MLS (microwave limb sounder) measurements reveal reasonable agreement between the satellite data and the model in the tropics and midlatitudes. The model calculations tend to be 10-20% higher than observed for these regions. This difference is likely due to the absence of halogen chemical cycles in the model. Simulations are underway with the new version of the model, which treats halogen chemistry and should give better agreement with observed ozone. At high latitudes, the model's ozone levels are significantly higher than the observations, especially during winter and spring when the model calculates excessively cold temperatures. During October at 75S, for example, the model calculates temperatures that are on the order of 40 degrees too cold between 2 and 5 mb. The calculated ozone is about a factor of 4 higher than observed due largely to the effect of the cold temperatures on temperature-dependent photochemical processes.

A model run was performed with a version of SKYHI that contains a gravity wave drag parameterization. This parameterization is described in detail in section (3.2.6), and consists of a forcing of the zonal winds in the mesosphere to observed climatological values. Incorporation of this parameterization into SKYHI results in temperatures that are very close to observed values throughout the stratosphere at all seasons. The effect on ozone is a reduction at high latitudes, reducing significantly the difference with MLS observations. The balance of the difference is likely due to missing photochemical cycles, such as the chlorine cycle.

3.1.6 Tropospheric Ozone

Tropospheric ozone, which both controls the chemical reactivity of the lower atmosphere and is a significant greenhouse gas, is thought to have increased significantly in a number of regions, possibly throughout much of the globe, over the last 100 years. The key issues to be addressed are the present and future impacts of human activity on tropospheric ozone levels and the mechanism or mechanisms responsible. A variety of observational, analytical and numerical techniques are being applied to this general problem.

Collaborations with colleagues from the Air Ocean Chemistry Experiment [AEROCE] and the North Atlantic Regional Experiment (NARE) continued. A preliminary regional simulation of summertime ozone over the North Atlantic (1421) found that human activity has more than doubled the ozone levels in the bottom half of the troposphere. A second study developed a modern tropospheric ozone climatology based on ozonesonde series from a number of islands in the region (1424), and a third study (1423) identified the transport processes responsible for some major events of elevated ozone over the North Atlantic.

A global study of the impact of human activity on tropospheric ozone was first discussed in A96/P97 (1445). An 11-level version of the GCTM has been employed within a conceptual framework of stratospheric injection, CO-CH₄ background tropospheric chemistry, parameterized pollution production in the continental boundary layer, and surface deposition to simulate realistic global distributions of present and pre-industrial tropospheric ozone (O₃). An evaluation with seasonal measurements from 12 surface sites, 21 ozonesonde sites and one aircraft campaign finds reasonable agreement (25% or better] with 73% of the observations while identifying systematic errors in the wintertime high-latitude Northern Hemisphere (NH), as well as the Southern Hemisphere (SH) tropics during biomass burning and the remote SH. The present total mass of tropospheric ozone [298 Tg O₃] is 39% larger than the preindustrial level of 215 Tg O₃. The two dominant components of the ozone budget are stratospheric injection (essentially constant at 696 Tg O₃/yr) and loss through dry deposition (which increases from 459Tg O₃/yr to a present level of 825 Tg O₃/yr). Tropospheric chemistry's contribution reverses from a preindustrial net-destruction of 236 Tg O₃/yr to the present net-production of 128TgO₃/yr, itself the balance between two large terms, 558 Tg O₃/yr of destruction in the background troposphere and 686 Tg O₃/yr of production in the polluted continental boundary layer. A detailed breakdown of the global budget is given in Table 3.1.

The feasibility of incorporating SKYHI stratospheric zonal averaged monthly ozone fields in the GCTM was explored in an effort to alleviate the GCTM's stratospheric biases, which accumulate excess ozone in the midlatitude lower-stratosphere of the NH, and too little ozone in the SH, thus resulting in an overestimate of transport into the NH upper troposphere during winter and spring and an underestimate of transport into the SH. A detailed comparison of available ozonesonde data and simulated SKYHI ozone profiles at representative grid boxes was performed, which showed that SKYHI values in the lower stratosphere agreed well with the observations. The SKYHI zonal mean mixing ratios for each month were converted to the GCTM's vertical grid using an appropriate mass weighted interpolation. These values were then inserted into the GCTM at 110 mb in the subtropics and tropics (30N-30S) and at 190mb elsewhere. Analysis showed that the best agreement with observed upper tropospheric ozone was obtained by utilizing a ten day relaxation of the GCTM-transported ozone to SKYHI values, which provided a softer link between the model's three-dimensional transport meteorology and the SKYHI zonal ozone.

Ozone simulations were then conducted with both the new chemical tables discussed (3.1.2) and the lower-stratospheric relaxation to SKYHI ozone discussed above. The results, when compared with a global set of 352 observations compiled from ozonesonde data, surface measurements and aircraft missions, show that the transport biases discussed above are significantly reduced (Fig. 3.2) and that the global simulation is now essentially quantitatively correct. Most large remaining differences are the result of either unresolved and/or poorly simulated local meteorology or errors in the simulated chemistry of the tropical biomass burning regions.

3.1.7 Atmospheric Sulfur Chemistry and Transport

Monthly-mean simulated column burdens of SO₄, discussed in A96/P97, have been used to calculate the clear-sky radiative forcing due to sulfate aerosols and the resulting forcings have been incorporated into the GFDL climate model. See (2.4.3) for details.

3.1.8 Tracer Transport

Tropospheric transport was evaluated in two GFDL tracer models in order to facilitate inverse modeling of atmospheric CO₂. Important results and progress cover three major topics. First, the SKYHI model, including a new parameterization of "aggressive vertical mixing" for unstable conditions, appears to provide a more realistic radon distribution in the lower troposphere than the standard parameterization. Second, the chemistry and transport of CFC11 simulated in the SKYHI model (using estimated global emissions from 1986-1989) compares well with observations, considering both the synoptic scale and seasonal variances. The modeled meridional CFC11 gradient, as represented by annual mean mixing ratios from ten observation sites from the South Pole to Alert, is in excellent agreement with observations, indicating that the meridional and interhemispheric transports have been well simulated in the SKYHI model. Third, transport of SF₆ was simulated, using global emissions from 1989-1993, in the SKYHI model and in the GCTM model. Both the SKYHI and GCTM simulations show a close match to the observed data. The results were submitted to the international tracer models intercomparison program TransCom-2, and analysis of the transport is underway.

3.1.9 Transport Dynamics

An overview/synthesis of the dynamics governing transport in the upper troposphere (1481) has led to some important conclusions and insights. First, the mechanistic insight and modeling capability is reasonably good in the midlatitude middle troposphere and tropopause region, though important uncertainties remain in the understanding and simulation of transport across the top of the planetary boundary layer and in the tropics. The concept of "barriers" to transport shows that the processes shaping the stratospheric circulation act to concentrate stratospheric tracers above the tropopause. The final "push" "across the tropopause requires crossing substantial barriers due to high static stability and large horizontal potential vorticity gradients. The crossing of these barriers ultimately depends upon the "breaking" of planetary-scale, cyclone-scale, and gravity waves due to non-linear breakdown. Interestingly, the structure of the tropopause itself is hypothesized to be a direct result of these same wave-breaking and transport shaping processes. Ultimately, the presence of tropopause and jet stream transport barriers in the upper troposphere are argued to be caused by the attempt of the atmosphere to homogenize itself due to the wide range of wave-induced transport processes (Fig. 3.3).

3.1.10 CO₂ Inverse Modeling

The inverse modeling effort of the Carbon Modeling Consortium (CMC) (see also section 5.4) is aimed at estimating the spatial distribution of uptake of CO₂ by the oceans and terrestrial biosphere, given the surface concentrations provided by CMDL/NOAA sampling network, and the atmospheric transport predicted by an atmospheric GCM (SKYHI or GCTM). For the inversion, the globe is divided into seven terrestrial and ten oceanic regions, according to large scale biome and ocean circulation features. The carbon sources considered are fossil fuel emissions, annually balanced land biotic flux (from the CASA terrestrial ecosystem model), and net oceanic and terrestrial uptake.

The first inverse technique uses the singular value decomposition (SVD) method for matrix inversion. The method was evaluated by recovering fluxes from simulated data within the same atmospheric model and across the two different atmospheric models. Sensitivity to sub-regional spatial variability of fluxes was also examined. Observations of atmospheric CFC11 for 1988 from the GAGE and CMDL networks were used in an inversion for land regions. These results were compared to industry production and consumption estimates for that year. Inverse modeling of atmospheric CO₂ for the decade 1981-1992 suggests 1-2 PgC/yr of terrestrial uptake of CO₂ in temperate North America and boreal Eurasia. Temperate Eurasia and boreal North America appear to be weak carbon sources. These findings are consistent with recent reports of forest regrowth, a lengthening of the growing season, and a large scale stimulation of forest productivity due to enhanced atmospheric CO₂ and widespread deposition of fixed nitrogen. The inverse model also confirms the existence of a large oceanic sink overall. Significant progress was also made towards resolving seasonal fluxes, using both SVD and a Kalman filter as a second inverse method.

The surface CO₂ monitoring network was analyzed to select optimal locations for new stations. In general, the inversion is best served by networks with about equal numbers of sampling stations per region. Optimal locations are found where the signal from individual sources is maximum, including the downwind boundaries of net source regions, and the centers of atmospheric high pressure cells. Point sources, such as fossil fuel combustion in concentrated urban areas, are problematic due to the large spatial variability. The optimization algorithm chooses stations away from these locations. There is considerable uncertainty in the vertical distributions of CO₂ in atmospheric models. Vertical profiles would be most useful constraints for savanna and rainforest regions in Africa and South America.

PLANS FY98

The GFDL GCTM will be updated to include CO emissions for preindustrial, present, and projected future conditions. The model will then be used to investigate global impacts of chemical emissions from Asia into the next century.

An analysis of the GFDL GCTM simulated tropical distribution of tropospheric ozone will be initiated, coincident with existing observational programs in three remote regions: the eastern South Pacific off the west coast of South America, the Indian Ocean, and the South Pacific Ocean near Samoa. The origin of the ozone will be determined by separating the transported tracer into three components: stratospheric ozone, ozone produced by CO/CH₄ chemistry, and ozone produced in the polluted boundary layer.

An NO_y simulation will be conducted with the GFDL GCTM using the new CO-NO_x-O₃-H₂O chemical tables. The simulated NO_x, HNO₃, and PAN will be compared with available measurements and with the previous simulation of NO_x, HNO₃, and PAN driven by off-line chemistry.

A five-tracer simulation with interacting NO_x, HNO₃, PAN, O₃, and CO will be conducted with the GFDL GCTM using the new CO-NO_x-O₃-H₂O dependent chemical tables. The resulting fields will be compared both with available observations and with previous independent simulations of NOy, O₃, and CO.

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

A three-year simulation of SF₆, N₂O, and CO₂ will be conducted using the SKYHI model as an extension of a previous four year simulation. The model results will be compared to aircraft measurements in the lower stratosphere and in the upper troposphere.

Development of polar ozone depletion chemistry will continue and an ozone "hole" "simulation will be performed using SKYHI.

NO_x sources from the GFDL GCTM will be added to SKYHI's troposphere in order to more realistically simulate the chemistry of the upper troposphere. The ozone budget of SKYHI's upper troposphere will be examined.

A systematic evaluation of the direct radiative forcing of tropospheric aerosols and the dynamical response of the atmosphere to the combined effect of this and other forcings will be initiated. Sulfur chemistry and emissions will first be incorporated in the SKYHI GCM, and preliminary global-scale simulations will be performed with no feedback between the chemistry and radiation.

3.2 ATMOSPHERIC DYNAMICS AND CIRCULATION

V. Balaji             C. Kerr***
D.G. Golder        J.D. Mahlman
K. Hamilton        L. Perliski
Y. Hayashi           R.J. Wilson
R. Hemler

***International Business Machines
****Los Alamos National Laboratories/DOE

ACTIVITIES FY97

3.2.1 SKYHI Model Development

The development of the new standard SKYHI model has continued. A preliminary version of the model containing all the major new parameterizations has been created and is undergoing testing. This model contains the upgraded long- and short-wave radiation codes, the new surface albedo scheme, new prognostic cloud scheme options, and a new formulation of vertical mixing under unstable conditions. Evaluation of the combined effect of these changes is underway. Several less significant scientific upgrades remain to be added, in addition to numerous computational and analysis upgrades.

The new SKYHI source may now be run on either the Cray T90 or Cray T3E computing systems simply by choosing the appropriate set of preprocessor options for the desired platform. The model may be executed on as many processors as the user desires, and will reproduce answers independent of the number of processors. This capability will ultimately allow SKYHI experiments to be run on either computing system as available system resources allow.

Inclusion of a parcel trajectory package within SKYHI has been completed. This package is currently undergoing checkout and evaluation. Also, efforts will continue toward generalizing and simplifying the procedures necessary to run SKYHI with arbitrary vertical and horizontal resolution and to analyze its output.

3.2.2 SKYHI Control Integrations and Basic Model Climatology

Control integrations were conducted on the CRAY T90 computer with various versions of the SKYHI model. These include two ten-year integrations with the 3 x 3.6 latitude-longitude version, one with the standard 40-level vertical grid (extending to ~80 km) and one with a 46level grid extending to ~96 km. A five-year integration has been also been conducted with the 46-level 2 x 2.4 version. Shorter integrations have been performed with an 80-level version of the model (to ~80 km) at both 3 x 3.6 and 1 x 1.2 horizontal resolution.

A number of conclusions have emerged from analysis of the global-scale circulation in these various integrations. Moving the upper boundary from ~80 km to ~96 km has a significant effect on the mesospheric circulation, with the winter (summer) polar temperatures in the mesosphere being warmer (colder) with the raised lid. This brings the simulation further from radiative equilibrium and into better agreement with observations. It appears that the upper boundary suppresses the mean meridional circulation and thus forces the temperatures near the model top to be unrealistically close to radiative equilibrium.

A preliminary examination of the results of the integration with the 80-level 1 x 1.2 version has proven very interesting. Results from June through August are now available, and the SH winter mesospheric circulation in this model appears to be significantly improved over that in the 40-level version. In particular, the SH polar stratosphere and mesosphere are warmer and the polar night jet weaker in the 80-level model. More dramatic differences are seen in the tropical circulation. The easterly peak of the semiannual oscillation (SAO) in July near 1 mb is much more realistic in the 80-level version, with the equatorial easterlies being stronger and more sharply peaked in height than in the 40-level version. It also appears from the limited integration conducted thus far that the 80-level model is adjusting towards a somewhat different equatorial wind climatology in the lower stratosphere than that exhibited by the 40level model.

3.2.3 Studies of Diurnal Variability

A diurnally-varying integration with a 54-level 3 x 3.6 resolution version of the SKYHI model extending to ~130 km has proceeded for two years. This integration is now being analyzed to examine tidal behavior in the stratosphere, mesosphere, and lower thermosphere. Fig. 3.4 shows a cross-section of the amplitude of the sun-synchronous component of the semidiurnal tidal oscillation in the zonal wind. The diurnally-varying integration will be compared with the results from a parallel diurnally-averaged control to determine the effects of the tides on the mean circulation in the middle atmosphere.

The role of latent heat release in forcing the upward-propagating diurnal tide was investigated using a linear tidal model in a project conducted in collaboration with J. Forbes (University of Colorado) and M. Hagan (NCAR). The results indicate a very significant component of the non-sun-synchronous tide seen in the mesosphere and lower thermosphere may be attributed to tropospheric latent heat excitation (fg).

3.2.4 Fine Horizontal Resolution Integration with SKYHI

The integration of a version of the 40-level SKYHI model with 0.333 x 0.4 latitude-longitude resolution has progressed from early May into September. The results for the circulation in the SH winter middle atmosphere have been particularly impressive. Fig. 3.5 shows the zonal-mean June-average temperature at 70S from the 0.333 x 0.4 integration compared with the 10-year mean 3 x 3.6 model climatology and two separate years from the 1 x 1.2 model control run. The tendency for the simulated temperatures in the polar regions to increase with improved resolution is apparent. At 3 x 3.6 resolution, the simulation has high latitude SH temperatures near 1 mb that are ~60C colder than observed. At 1 x 1.2 resolution this has improved to ~25C, while the 0.333 x 0.4 version has almost no cold bias in the high latitude SH temperatures in June. Similar results are obtained in July, but in August the familiar "cold winter pole" bias returns somewhat in the high resolution model, with temperatures colder than observations by as much as 20C near 1 mb at the pole. The improved temperature simulation at high resolution is associated with a more realistic polar night jet structure, which displays the observed equatorward-upward slope of the vortex

edge. This is by far the most realistic GCM simulation of the SH winter stratosphere and mesosphere that has been obtained without the imposition of a parameterized subgrid-scale drag on the mean winds. The NH summer mesospheric simulation is much less sensitive to the horizontal resolution employed and, even in the highest resolution version, a significant warm bias persists near the summer mesopause.

Examination of the tropical tropospheric simulation in the 0.333 x 0.4 resolution model reveals the presence of very strong tropical cyclones in the Western Pacific. This includes one "super-typhoon" with minimum surface pressure of 906 mb and near-surface winds of more than 70 m-s^-1. This storm also has a warm core in the upper troposphere of ~18C at its most intense phase. This appears to be the strongest tropical storm ever seen in a global climate model simulation, and is only slightly weaker than the most intense typhoons that have been observed in the real atmosphere. About 15 West Pacific tropical cyclones are found during the June-August period of the simulation, and the general pattern of formation, intensification and movement of these storms appears to be quite realistic. The model is much less successful in simulating Atlantic tropical storms, however (fb).

3.2.5 Explicit Simulation of Convectively-Forced Gravity Waves

The GFDL limited-area nonhydrostatic (LAN) model is being applied to the simulation of stratospheric gravity waves generated by tropical moist convection. Initial efforts have focussed on high-resolution two-dimensional simulations of isolated regions of convection. Fig. 3.6 shows results from a simulation starting with horizontally-symmetric initial conditions appropriate for a moderately unstable tropical profile. A heat flux is applied in a narrow region near the center of the domain and the model is integrated for several hours. The green region shows the outline of significant condensed water after 5.5 hours of integration, while the red and blue contours show the vertical velocity field after 9 hours of integration. The initial mean wind in the upper troposphere and stratosphere is rightward at 5 m-s^-1. Fig. 3.6 shows that the convection excites rightward and leftward moving gravity waves, some of which have reached the lower thermosphere by 9 hours. These simulations will be repeated for various initial conditions, mean winds, and surface heat fluxes. The results will be analyzed to investigate a number of issues that are critical to the formulation of physically-based gravity wave parameterization schemes for global models. In particular, the simulations will be used to study the dependence of the momentum flux spectrum above convection on the intensity, suddenness, geometry, and mesoscale organization of the convective elements, and on the mean wind shear.

3.2.6 Parameterized Gravity Wave Drag in the SKYHI Model

An efficient prescribed drag has been developed for the 40-level 3 x 3.6 SKYHI model which ensures that zonal-mean temperatures throughout the middle atmosphere remain within about 5C of the observed climatology (1441). This drag is now being applied in the version of SKYHI including the stratospheric photochemistry. The same drag is also being employed in a zonally-symmetric version of the model. The mean diabatic circulation in the two models should be almost identical, so comparison of the three-dimensional and zonally-symmetric versions will allow a direct determination of the effects of three-dimensional eddy fluxes on the stratospheric chemistry and transport. Integrations of the SKYHI model including a version of the Lindzen gravity wave drag parameterization scheme have also continued.

3.2.7 GCM Simulation with an Imposed Tropical Quasi-biennial Oscillation

A 48-year simulation using the 3 x 3.6 SKYHI model with an imposed zonal momentum source designed to force a realistic QBO in the tropical stratosphere described in A96/P97 was analyzed (ex). The extratropical circulation in the NH winter stratosphere was found to be affected by the tropical QBO in a manner similar to that which has been observed. In particular, the polar vortex is generally weaker in winters in which there are easterlies in the tropical middle stratosphere. Roughly 2/3 of the largest midwinter polar warmings occur when the equatorial 30 mb winds are easterly, again in rough agreement with observations. Despite this effect, however, the total interannual variance in the zonal-mean extratropical circulation in the model apparently is not increased by the inclusion of the tropical QBO. The observed dependence of the winter-mean stratospheric extratropical stationary wave patterns on the QBO is also quite well reproduced in the model.

The QBO was also found to have a profound influence on stratospheric stationary waves at low latitudes. Near and above 10 mb, the NH stationary waves were found to penetrate across the equator during the westerly QBO phase, but to be restricted to latitudes poleward of ~10N during the easterly phase. This means that the equatorial QBO in the prevailing wind near and above 10 mb has a significant non-zonally-symmetric component. If this is also true in the real atmosphere, there are important implications for the adequacy of the current observational rawindsonde network near the equator.

A detailed comparison is being made of the deviations from zonal symmetry near the equator in the model simulation and in results from the High Resolution Doppler Interferometer on the UARS satellite. This work is in collaboration with D. Ortland (U. of Michigan) and preliminary results are very encouraging. In order to extend this comparison, the SKYHI model experiment will be repeated using a mean flow evolution based on observations during the UARS period rather than the somewhat idealized QBO imposed in the 48-year run.

3.2.8 GCM Simulation of the Ozone Quasi-biennial Oscillation

An integration of the SKYHI model with stratospheric chemistry and incorporating the QBO forcing described above has now proceeded for more than five QBO cycles (27months each). Analysis of calculated total ozone shows that the model is producing a QBO signal in total ozone with a peak-to-peak amplitude of about 25 DU, in good agreement with satellite observations. The QBO signal in total ozone appears to propagate poleward such that the midlatitude total ozone QBO is out of phase with the tropical QBO and has a peak-to-peak amplitude of about 20 DU, somewhat less than the tropical amplitude.

Time-height analyses of the calculated ozone reveal two regions of response to the imposed QBO located roughly between 25 and 100 mb and between 2 and 10 mb. In the lower region, where ozone is under dynamical control, the ozone QBO is due to the vertical circulation that arises in association with the vertical shear zones between the alternating regions of westerlies and easterlies. The ozone QBO in the upper region appears to be due to a QBO in NO₂, which effectively changes the catalytic ozone loss due to nitrogen chemistry in this region. The NO₂ QBO results from the secondary circulation associated with the vertical shear zones.

3.2.9 Observational Studies Using Radiosonde Data

An analysis of high resolution radiosonde data from several stations in the Western Pacific during the TOGA-COARE experiment has been completed. Rather remarkably, very short (~2-3 km) vertical wavelength waves were found to be coherent across the ~35 longitude spread of the stations considered. The temperature perturbations travel eastward with a very slow (~3 m-s^-1) phase speed. This wave is very strongly damped above the troposphere, and the eddy flux convergence associated with this wave could play a significant role in the dynamics and constituent transport near the tropical tropopause (1469). Efforts have also been made to obtain some high resolution European radiosonde data, which will be analyzed for gravity wave signals as part of a world-wide project (coordinated by SPARC) to establish a gravity wave climatology for the lower stratosphere.

A preliminary examination was undertaken of the global archive of wind and temperature observations taken during ozonesonde ascents. These are of particular interest due to the rather high ceilings often reached by the balloons employed. More than a hundred soundings extending up to 5 mb are available at each of several stations in the NH midlatitudes. An examination of the temperature observations from these stations revealed long-term-mean mid-stratospheric summer temperatures that are lower by ~5C than those in climatologies based on the Nimbus-7 radiometer data from the 1970s (eu). This is consistent with some of the more recent satellite data now available (1418).

3.2.10 Dynamics of the Martian Atmosphere

Transport of aerosols and water vapor is a central aspect of the description of the Martian climate. Radiative heating due to aerosols is a major source of thermal forcing for the atmospheric circulation, which in turn, strongly controls the raising and distribution of the dust. This interrelation is most prominently evident during major dust storms where the interaction between radiative heating and advection by the intensifying circulation can result in the deep and wide-spread distribution of dust around the planet. A realistic simulation of this phenomenon and comparison with observations from the Viking mission have been completed (1422). This simulation confirmed theories that the Hadley circulation rapidly expanded and intensified in response to the deepening distribution of aerosols. It was found that the thermal tides play an important supporting role in producing a strong warming at the winter pole.

Spacecraft observations have indicated a strong hemispheric asymmetry in atmospheric water vapor column abundance due to seasonal and spatial variations in the exchange with surface sources, and sinks and transport within the atmosphere. These observations are complemented by Hubble Space Telescope images of widespread water ice during certain seasons. Transport of water vapor and condensate has been introduced in the Mars general circulation model to explore the role of baroclinic and stationary waves in the seasonal water cycle and the possible interactions between water ice clouds and the aerosol distribution. It is anticipated that the upcoming Mars Global Surveyor mission will yield a wealth of new data for the investigation of aerosol and water vapor.

PLANS FY98

Work will continue toward the production of the new standard version of SKYHI. Evaluation and improvement of the new scientific parameterizations will proceed, in addition to computational upgrades.

Development of the procedures needed to both generate initial data and analyze model output for models with arbitrary resolution will continue.

Integrations of SKYHI with both high vertical and horizontal resolution will be run on the GFDL CRAY T90 and T3E.

Analysis of the tropical cyclones in the very high horizontal resolution version of SKYHI will continue. The analysis will focus on both the ability of the model to simulate realistic tropical storm structure and climatology, and the role of tropical cyclones in generating stratospheric gravity waves.

Integrations of SKYHI with parameterizations of stratospheric and mesospheric gravity wave drag will be continued, with the goal of obtaining a consistent formulation that is applicable over a range of horizontal and vertical resolutions.

Investigation of the diurnal cycle in the SKYHI model will continue. Detailed analysis will focus on the mechanisms of nonlinear saturation of the diurnal tide in the mesosphere and lower thermosphere and the interaction of the tide with the mean flow. Also, an attempt will be made to characterize the interaction of the tide with explicitly-resolved vertically-propagating gravity waves.

Development of a simple, dry, planar-geometry model for studies of vertical wave propagation will continue. This will be applied to the study of the saturation of a spectrum of vertically-propagating gravity waves.

The limited-area model simulations of stratospheric gravity waves forced by tropospheric moist convection will continue. The relation of gravity wave fluxes to the intensity, suddenness and mesoscale organization of the convection will be studied.

The QBO experiment with stratospheric photochemistry will be analyzed with a focus on the role of the QBO in modulating tropical-extratropical ozone transport.

 Table of Contents