III.html

Janice Coen worked with Roy Rasmussen (joint appointment with RAP) and Gregory Thompson (RAP), Istvan Geresdi (visitor, Janus Pannonius University, Hungary), John Brown (NOAA), and Kevin Manning to improve the microphysics in MM5, Rapid Update Cycle, version 2 (RUC2), and other models. They reexamined nucleation modes parameterized in these models and searched for specific weaknesses resulting from the parameterized treatment of hydrometeors, particularly ice particles and drizzle. This involved using ice particle concentrations in clouds known to form ice through primary nucleation modes to prescribe the ice concentrations as a function of temperature in order to eliminate outdated formulas. Examining hydrometeor spectra in widely varying types of clouds, they also identified typical values of drop size distribution parameters that models use to calculate precipitation fall-speeds, which could lead to better forecasts of freezing drizzle.

Charles Knight nearly completed an extensive study of the crystallization of tetrahydrofuran (THF) clathrate hydrate from the melt. This is a model system for studying natural gas hydrates, important in the petroleum industry and possibly important for climate change, since most methane, an important greenhouse gas, is sequestered as methane hydrate beneath the sea floor. This work is in cooperation with the hydrate group under E. Dendy Sloan (Colorado School of Mines). He also continued studies of biological antifreezes, with collaborators from the University of Illinois and the University of Houston.

When ice forms in clouds it influences many important physical, chemical, electrical, and radiative properties in those clouds. Yet understanding of the primary formation and the multiplication and spread of ice in clouds has languished for a couple of decades. As weather and climate models gain finer and finer spatial resolution and become more sophisticated lack of understanding of the formation of ice in clouds and inadequate microphysical parameterizations are serious weaknesses. Recognizing the need for additional effort by the community in this area, MMM undertook an effort in FY 98 to develop an Ice Microphysics Initiative (ICE). James Dye is coordinating the effort in collaboration with William Cooper, Andrew Heymsfield, and Knight. Many university colleagues including Marcia Baker (University of Washington), Alan Blyth (New Mexico Institute of Technology), John Hallett (University of Nevada), Peter Hobbs (University of Washington), Dennis Lamb (Pennsylvania State University), David Rogers (Colorado State University), Jerry Straka (University of Oklahoma), Gabor Vali (University of Wyoming), and George Isaac (Atmospheric Environment Service of Canada) have been involved in early discussions. Planning meetings were held in Boulder and also at the Cloud Physics Conference in Everett, Washington. An overview document is being prepared, working groups have been formed, and a general community meeting is being planned for the spring of 1999 to prepare a written proposal for ICE. The effort would involve field, laboratory, modeling, and theoretical studies.

The Tropical Rainfall Measuring Mission (TRMM) satellite (U.S. and Japan) was launched in November of 1997 to provide radar, radiometric, and lightning observations of clouds in the tropics. A main goal of the TRMM project is to be able to use these measurements coupled with numerical models to better determine latent heating profiles in the tropics. Dye and Heymsfield, in collaboration with Jeffrey Stith (University of North Dakota) and Paul Lawson (SPEC, Inc., Boulder, Colorado) and many other investigators, participated in the TRMM ground validation projects conducted in Texas and in Florida. Their primary involvement was in working with microphysical measurements, especially a new microphysical instrument, the Cloud Particle Imager (CPI) developed by SPEC, Inc. The CPI images particles with pixel resolution of 2.3 x 2.3 micron, which is almost like looking at the particles under a microscope. This new instrument permits a much more detailed examination of ice particle habits, greatly improved measurements of small ice particles, and the ability to determine the degree of riming on the particles. They will continue to work with the TRMM team in relating the microphysical measurements to other TRMM satellite and ground validation measurements. The CPI will also be an important new instrument for the ICE initiative mentioned above.

MMM's ongoing Wildfire Research Program is directed at improving the understanding of fire behavior and involves collaboration with a growing number of colleagues in NCAR/ATD, the U.S. Forest Service, Monash University, University of Colorado, the Country Fire Authority (CFA) of Victoria, and Australia's Northern Territories Bushfire Council. The approach uses a combination of modeling, instrument development, and observations of forest fires and grass fires from field experiments conducted in the Northwest Territories of Canada, Australia, and the western United States.

The modeling studies applied a small-scale atmospheric prediction model that was coupled with an empirical fire spread model such that heat from the fire fed back directly affecting the atmosphere to produce fire winds. These studies included simulations of fires in idealized atmospheric and topographic conditions to understand basic characteristics of fire dynamics, case studies that examine fire spread in changing mesoscale conditions, and further model development.

Negative wind shear: A cause of blowups? Coen and Terry Clark designed and performed a series of simulations to reveal how negative vertical wind shear, perhaps from nocturnal drainage winds, microburst outflows, or gust fronts, might lead to fire blowups. This work showed that strong negative wind shear can lead to extreme wildfire behavior, producing strong fire whirls, rapid intensification of dynamics at the fire line, erratic spread rates, and strong updrafts capable of lofting burning embers away from the primary fire line. This extreme behavior appears to be sustainable through formation of a strong horizontal vortex along the fire line. Simulations also suggest that a fire might modify the winds in its environment so much that this modified local environment drives the fire behavior.

Mechanisms for the rapid spread of fires up hills. Coen and Clark simulated the passage of a wildfire over small hills of varying slope to examine why fires spread faster uphill. These experiments showed, firstly, that the model could reproduce the effect, with speed-up factors that were comparable to that predicted from an empirically based fire spread algorithm. These experiments also showed that the effect occurred in the model where only basic dynamic effects (such as acceleration of winds on hill due to potential flow and convective flow caused by the fire) are represented, suggesting that other widely-quoted mechanisms such as the slope lifting the fuel closer to the flames may be overemphasized. This work involves collaboration with Donald Latham and Francis Fujioka (USDA Forest Service).

Fire propagating over a small hill in katabatic flow conditions. To understand fire behavior in a common, more complex flow, Coen and Clark performed idealized simulations of a line fire propagating over a small Gaussian hill (height 200 m) with a relatively sharp slope. The ambient flow represented a katabatic flow or a gust front, i.e., a stable layer with strong easterlies overlaid by a neutral layer with moderate westerlies. Without a fire, this results in amplified winds on the lee side due to reflective effects from an overlaying critical layer. When a fire was ignited, it first moved up the slope, producing a rotating smoke column. One of the fire's flanks died out as the other intensified and, instead of propagating down the lee side, turned sharply along the ridge line producing what resembled the commonly-observed fire whirl forming in the lee of a ridge. Since the flow itself was symmetric, these results suggest that an instability amplified an asymmetric disturbance. In collaboration with Donald Middleton (SCD), video animations were produced to reveal some of the interesting features of the fire and flow.

Grass fire experiments. In collaboration with Paul Ginoux (ACD), Coen and Clark applied the model to simulate a small grass fire in the Congo, where fires are commonly set to clear and enrich agricultural land. Here, the fire spread rate is quite small, the fuel load relatively light, and the fire line very narrow. This simulation tested the model in a different regime of fire behavior and improved the model; confidence was gained in the model’s ability to represent a wide range of fire behavior.

The collaborative research project between Clark, Coen, Larry Radke (ATD), and Middleton resulted in the development of an image flow analysis software package. Robust statistics and least squares minimization procedures were used to extract fire winds. This software was applied to the infrared (IR) video data taken during the International Crown Fire Modelling Experiment in Canada's Northwest Territories from June-July 1997. The analysis gave estimates of fire winds showing amplitudes between 20 and 30 m/s. Using the derived vertical velocities and radiance temperatures, they were able to derive heat flux profiles in agreement with expected values thus corroborating the wind estimates. This work resulted in a paper believed to represent the first quantitative data set, albeit derived, showing the spatial and temporal structures of fire winds, heat flux profiles as well as derived vertical vorticity.

The image flow analysis software was successfully used in Wild Fire Experiment (WiFE) for a quick-look analysis of the observed wind fields. This post flight analysis allowed evaluation and assessment of the various infrared (IR) temperature ranges. Some further software development was performed to extract the ego motion of the observational platform, which is necessary before image flow analysis can be successfully applied. The combined image flow package with ego motion extraction represents a rather robust software system, which allows the quantitative analysis of IR data from unsteady platforms.

In ongoing work with James Hoffman (Space Instruments, Inc., Encinitas, California), Coen began developing techniques to use spectral data of fires from a thermal imaging radiometer to observe the various scales of wildfire dynamics. Because motions within the fire are obscured by smoke, clouds, debris, birds, and complex energy sources, one of the challenges of fire observations has been to identify a suitable combination of wavelengths that, when combined, provide the best view of a fire's structure and energetics; this work currently uses wavelengths within the 8-12 micron range. The figure shows one view of a FIRE in the 11-micron channel; the data values can eventually be scaled approximately linearly to temperature.

Sullivan, Chin-Hoh Moeng, Bjorn Stevens (ASP), Donald Lenschow, and Shane Mayor (University of Wisconsin) analyzed entrainment processes in clear convective planetary boundary layers (PBLs) using high-resolution large eddy simulations (LES). They described the role that thermal plumes play in the entrainment process over a range of bulk Richardson numbers, showed the results of a quadrant analysis of the buoyancy flux in the inversion region, proposed a new definition of the PBL height z_ibased on gradients in the potential temperature field, computed statistics of z_i, and examined the effect of finite inversion layer thickness in an entrainment rate parameterization.

Edward Patton (visitor, University of Minnesota), Christoff Kiemle (Institute of Atmospheric Physics, German Aerospace Research Establishment [DLR]), and Kenneth Davis (University of Minnesota) estimated entrainment zone thicknesses from LES simulations using a histogram technique developed by Kiemle and Davis for light detecting and ranging radar (LIDAR) measurements, and found many similarities between measured and simulated statistics of entrainment zone thickness.

Stevens, Moeng, and Sullivan studied the sensitivities of LES to subgrid scale (SGS) modeling-especially near the entrainment zone of the PBL. They found that SGS models that carry predictive equations for SGS energy are more susceptible to the model for SGS lengthscale than are models that diagnose SGS energy. They also showed that the sensitivity of LES to the SGS lengthscale model depends on the type of flow, and in particular, the strength of the inversion. The lengthscale sensitivity of SGS models that predict SGS energy is interpreted using analytic solutions to the SGS energy equation for conditions of no transport and fixed forcings.

Eileen Saiki (ASP), Moeng, and Sullivan continued to apply LES to the investigation of the nocturnal (stable) boundary layer (SBL). The SBL is characterized by intermittent turbulence combined with other phenomena such as gravity waves and low-level jets. A goal of this work was to study a rapidly cooling SBL. However, they found that in order to simulate this problem, modifications in the SGS model and gradual cooling of the boundary layer were needed. Several rapidly cooling windy SBLs were simulated and the results are being evaluated. This includes examining the evolution of the low-level jet that develops above the SBL, and performing detailed flow visualization of the SBL. Jielun Sun and James Howell (ASP) found, from analysis of data from the Microfronts experiment, that in light wind conditions significant heat flux divergence and deviations from Monin-Obukhov similarity can occur in the lowest few meters of the SBL.

In collaboration with Moeng and Sullivan, Jeffrey Weil (visitor, University of Colorado) used LES velocity fields to model passive "particle" dispersion in a Lagrangian framework. The objective is to calculate concentration mean and variance due to a point source in the convective boundary layer (CBL) and requires the total or absolute dispersion, including plume meander, and the relative dispersion about the local plume centroid. The total and relative dispersion is obtained from Lagrangian "one-" and "two-particle" models, respectively. This is the first calculation of relative dispersion with a two-particle model for the CBL. Results show that the total dispersion exhibits an initial linear growth with time consistent with Taylor's theory and tank experiments, whereas the long-time vertical spread tends toward a constant due to plume trapping in the mixed layer. For all source heights, the relative dispersion exhibits a nearly 3/2 power-law time dependence, consistent with Batchelor's theory, but the dispersion magnitude decreases with source height; the latter is attributed to the decrease in the turbulence dissipation rate. The net result is that the total and relative dispersion is closest in magnitude for surface sources and most widely separated for elevated releases. This suggests that the concentration variance should be much greater for elevated than for surface sources; model extensions and future calculations of the variance will be made to check this.

There is evidence from several sources for at least two scaling laws for small-scale statistics in turbulence. Work in MMM on the statistics of small-scale turbulence has focused on this issue. With Alexander Praskovsky (RAP), Robert Kerr (joint appointment with CGD and HAO) found that the old Moscow wind tunnel data shows not only different scaling laws for the transverse and longitudinal structure functions, but evidence for an additional dynamically important length scale within the inertial subrange. Two scaling laws suggest the existence of this second length scale. In collaboration with Michael Spector (visitor, University of Colorado), Kerr noted that there is nothing in the symmetry groups for structure functions that would rule out this possibility, refuting a recent claim in the literature. New analysis, published in two recent conference proceedings, of the singular solution of the Euler equations published in 1993 suggests a source for two small length scales. To have two small length scales in the singular solution also implies sharp changes in the angles between the vorticity vector and its time derivative. Kerr, in collaboration with John Gibbon (Imperial College, London) and Barak Galanti (Weizmann Institute, Israel), are working on a refined analysis of the singular solution of the Euler equations to identify this process. As part of this collaboration, Galanti has been given a version of Kerr's spectral code with modifications to allow him to do a forced 512 cubed simulation of turbulence on the new Cray J-9 at his institution.

Clark, William Hall, and Kerr continued their study of the 12 December 1992 Front Range windstorm in collaboration with F. Martin Ralph, Paul Neiman, and David Levinson (all of NOAA). Their analysis concentrated on identifying the source regions of clear-air turbulence (CAT) associated with the jet stream and associated upper level frontal passage. A major finding of this work is that the wave-like distortions of the jet stream aligned with the flow resulted in the major CAT source regions. Horizontal Vortex Tubes, resulting from these jet stream distortions, were identified as a likely cause of the aircraft incident over Evergreen, Colorado where a DC-10 lost 19 ft of wing plus one engine. Also, this work shows that the high-amplitude jet stream undulations can result in the underside of the jet stream acting to provide significant source regions of CAT. The work also highlighted the morphology of a "turbulent downburst" which occurred in the simulations when the jet stream undulations and internal gravity waves interacted to produce a confined region of wave breaking that penetrated to the surface. The nature of this surface gust was in agreement with observations, which showed intermittent surface gustiness along the Front Range. They continued the comparison of model results with available observations where many of the modeled structures such as HVTs and flow aligned cloud bands as seen by satellite corroborated their findings

Turbulence research in MMM extends from understanding the fundamental statistical structure of turbulence to applications to aviation safety. Analysis of the case study of a downslope windstorm over the Front Range near Boulder, Colorado on 9 December 1992 where part of a wing and one engine were torn off a DC-8 cargo jet has shown that the most intense vorticity is in the form of vortex tubes. Hall, Kerr, and Clark found that the most intense vortex tubes are located on the tropopause and are associated with the passage of a front from the north. The elevation of these tubes and the timing of the frontal passage are consistent with the time and location of the accident. The analysis leading to vortex tubes came from earlier fundamental investigations that led to the current understanding of the importance of vortex filaments in turbulence. Currently, Kerr, with Praskovsky and Spector, are considering the evidence for separate scaling functions for the transverse and longitudinal structure functions, which might point to the importance of more complicated structures.

Piotr Smolarkiewicz and Christoph Schaer (long-term visitor, ETH, Switzerland) conducted a collaborative study on alpine flows (relevant to the Mesoscale Alpine Project [MAP]). They investigated the predictability of stratospheric gravity wave breaking. The outcome appears important: (1) small-scale orographic features play a negligible role in defining location and timing of gravity wave breaking aloft; and (2) the spatial distribution of the overall breaking region is complex (e.g., in general, lower and upper breaking regions are not aligned) and observations of the wave-breaking at lower altitudes do not imply accurate predictions of the breaking aloft.

Joseph Prusa (Iowa State University), Smolarkiewicz, Rolando Garcia (ACD), and Andrzej Wyszogrodzki (visitor, University of Warsaw, Poland) continued their collaborative study of gravity wave activity in the upper mesosphere and lower thermosphere. New high resolution (544 x 291 x 80 grid points) simulations supported earlier computations (at 321 x 177 x 49 grid points) that showed the power spectra consisted of three spectral subranges: (1) an inertial range turbulence (IRT) at the smallest wavenumbers, (2) a buoyancy range turbulence (BRT) at mid wavenumbers, and (3) a reverse energy cascade (RC) at the longest wavenumbers. These three subranges are characterized by power spectra slopes of -5/3, -3, and -5/3 for the IRT, BRT, and RC, respectively. The critical wavenumber between IRT and BRT is correctly given by the Osmidov (buoyancy) scale, whereas the primary mode at which the wavefield is being forced gives the critical wavenumber separating BRT from RC. They conjectured that the RC is due to modifications of the mean state by the nonlinear 2-D wavefield prior to extensive wavebreaking and the onset of 3-D effects. Present work is focused on analysis of Eliassen-Palm flux divergence and 3rd order structure functions to test the validity of this conjecture. The high resolution run was obtained using a high performance version of the Eulerian/semi-Lagrangian numerical model (EULAG) for fluids. This code was successfully run on the 512 processor Cray T3E at the National Energy Research Scientific Computing Center (NERSC) in Berkeley, California.

Vanda Grubišic (ASP) was involved in a multi-model intercomparison of gravity wave breaking as a part of the pre-Mesoscale Alpine Project (MAP) research effort. The intercomparison involved eight nonhydrostatic mesoscale models including EULAG. The results of 2-D idealized high-resolution numerical simulations of the 11 January 1972 downslope windstorm case that occurred along the Colorado Front Range were encouraging with respect to the predictability of the upper level gravity wave breaking as seven out of eight models predicted the occurrence of wave breaking in similar locations. However, the structure of the wave breaking was found to be very sensitive to the vertical resolution, numerical algorithms and lateral boundary conditions as well as to the variations of shear and stability at altitudes above 10 km.

An effort is underway to investigate the utility of LES to enhance understanding of the interplay between chemical reactivity and turbulent transport. An initial study coupled a clear convective boundary layer with simple decay chemistry to determine characteristics of ozone-like and isoprene-like tracers and to investigate the top-down/bottom-up functions for chemically reactive species.

Patton and Kenneth Davis coupled the simple chemistry mechanism with a canopy-scale LES in which the plant canopy acts as a spatially distributed source of the species. Chemical loss enhances the scalar gradients substantially, but turbulent transport dominates even for a scalar with a 5-minute lifetime. The peak in a constant decay-rate scalar variance lies near the canopy top (due to both the increased velocity shear and scalar source in the vicinity) and diminishes in magnitude in accordance with the prescribed decay rates. However, the variance of a simplified isoprene species, whose decay rate depends on the local hydroxyl radical mixing ratio, is only slightly smaller than the conserved species due to a compensating variance source term resulting from chemical reactivity that appears in the isoprene variance equation. Numerical experiments are also underway using Sullivan's nested-grid LES coupled with both the plant canopy and the simple chemistry to investigate the influence of horizontally inhomogeneous chemical emissions and canopy storage on full PBL-scale statistics of reactive hydrocarbons.

Recently, Mary Barth (joint appointment with ACD), in collaboration with Patton, Kenneth Davis, and Moeng, performed a simulation coupling convective PBL dynamics with realistic hydrocarbon chemistry appropriate for rural conditions. Analysis of these results and simulation of a more urban chemical scenario is forthcoming.

Jordan Powers and Peter Hess (ACD) produced, and are experimenting with, a new version of a coupled mesoscale atmospheric/chemical transport model. The coupled system links the MM5 with a regional chemical transport model (CTM) developed by Hess. With the goal of understanding the regional transport of selected chemical species by deep convection, the coupled MM5-CTM is being used to investigate the intensively studied 10 July 1996 STERAO event. The current coupled model employs a newer version of the MM5 than a previous-generation prototype, and test simulations of the 10 July event with it have been favorable. The MM5 testing yielded a method to encourage convective initiation for this key STERAO case, and the preliminary MM5 results illuminated the event's mesoscale environmental conditions. Future work will turn to the CTM's behavior and its simulations of regional transport in the event.

Analysis of observations made during the STERAO/Deep Convection experiment was continued by James Dye in collaboration with Stith, Brian Ridley (ACD), Pierre Laroche (ONERA, France), Steve Rutledge (Colorado State University), colleagues at the NOAA Aeronomy Lab, and Thomas Matejka and Diana Bartel (both of NOAA). Major scientific objectives of the project were to investigate the transport of chemical constituents by deep convection and the production of NO_x (comprised of NO and NO₂) by lightning. It was unique in combining and coordinating extensive chemistry, air motion, and microphysical lightning measurements in and around thunderstorms.

An important component of the experiment was the ability to determine the locations of lightning in the storm using the ONERA lightning interferometer. Daniel Breed (joint appointment with RAP) and Martin Venticinque, working with Dye in collaboration with Laroche and Eric Defer (ONERA, France), evaluated the interferometer data and determined times and locations for some storms when the full three-dimensional interferometer location data are reliable. They also identified some regions near the interferometer lobes when larger location errors occur. These lightning data were then used in conjunction with the National Lightning Detection Network data to determine IC, CG, and total lightning flash rates and the fraction of total lightning which is IC for some STERAO storms. For many of the storms the ratio of IC to total lightning is in excess of 80 percent, for example for the 10 July 1996 storm during an intense multicellular stage and then a later quasi-supercellular stage the storm had IC to total flash rate ratios of greater than 99 percent for tens of minutes. The ratio as well as the total lightning activity was well correlated with intensification of the storms and presumably the updraft strength.

In comparing the lightning channel locations from the interferometer with the Doppler derived air flow fields in the storms prepared by Matejka and Bartels, Dye found that for the one 5-minute period examined so far, most of the lightning activity occurred in the regions with moderate updrafts. Surprisingly, most of the downdraft, especially the core, was almost devoid of lightning sources. There was some lightning in the updraft core but not as much as in the moderate updrafts just downshear of the core. This effort will be extended to look at other times and other storms.

William Skamarock simulated the 10 July storm 1996 using the COllaborative Model for Multiscale Atmospheric Simulation (COMMAS) model and, working with Dye and Matejka, used the observations to help refine the simulations so that they more closely match the character of the observed storm. The simulated storm showed that storm inflow was from the south-southwest at lowest inflow levels but from the west-northwest in the upper boundary layer. Interestingly, air between 0 and 500 m AGL was not ingested into the simulated storm; the main inflow layer extended from 500 m to 2 km AGL. Convective transport of CO from the boundary layer into the anvil show significant temporal variability, with net CO transport into the anvil varying by a factor of 3 during the lifetime of the simulated storm (see figure). Transport by the single mature quasi-supercell is found to be roughly equivalent to that produced by 3 to 5 cells in the multicellular stage. The next step in this effort is to specify lightning locations in the simulation based on observations from the interferometer. The investigators will then be able to examine NO_xproduction by lightning and compare with NO observations in the anvil.

Stith, in collaboration with Dye and Ridley, investigated narrow spikes of NO up to 19 ppbv observed inside or at the edges of some of the STERAO storms. These values were frequently much larger than the more commonly observed wide spread values of about 1 to 2 ppbv observed in the anvils of the storms. Values outside the anvils were typically 0.1 to 0.2 ppbv. By comparing the location of these spikes with lightning activity detected by the interferometer, the investigators found that the spikes were located in or downwind of electrically active regions of the storms. In a few cases individual flashes could be identified as probably being responsible for a particular spike. Using a simple plume model and the observed magnitude and width of the plumes they estimate NO production rates ranging from 7 x 10¹⁹ to 3 x 10²² molecules of NO per meter path length. These estimates are very rough but should be helpful in trying to constrain NO production rates from lightning in the storm simulations mentioned above and in other models.

Barth included soluble tracers into the COMMAS to examine how uptake by liquid hydrometeors affects the distribution of species. Amy Stuart (visitor, Stanford University), working with Barth and Skamarock, examined the effect of capturing soluble tracers in frozen hydrometeors when riming occurred compared to volatizing soluble tracers from liquid drops when riming occurred upon the distribution of soluble species. Results show that highly soluble species are distributed differently than an insoluble species. This behavior is even more pronounced when the soluble species are captured by frozen hydrometeors because the snow and graupel carry the soluble tracer to the boundary layer. Simulations with chemical reactivity in both the gas and aqueous phases within the cloud model will be performed to examine whether the aqueous chemistry influences the concentration of chemical species in the anvil region of a cloud. The ultimate goal is to be able to use the model coupled with the observations to quantify the vertical redistribution of constituents and the production of NO_x by lightning.

Richard Ramaroson (visitor; ACD, ONERA, France), along with Skamarock and Barth, examined two aspects of the 10 July 1996 COMMAS simulations. Cloud and hydrometeor fields from the model are being used to produce a 3-D map of the actinic fluxes needed for computing photolysis rates. The calculated photolysis rates are then included in parcel model gas and aqueous phase photochemistry calculations using trajectories from the COMMAS simulations. The calculated photolysis rates will be incorporated into the cloud and chemistry model for computing photochemical reactions online in the COMMAS model.

Barth, Philip Rasch (CGD), and Jeffrey Kiehl (CGD) continued to model global sulfur distributions (aerosol sulfate, sulfur dioxide, and dimethyl sulfide). The control simulations were documented where three aspects of model results are discussed: (1) comparison of model results to observations and the importance of cloud chemistry to the tropospheric aerosol sulfate distribution, (2) description of the sulfur cycle in the atmosphere and the potential for certain regions to affect the aerosol sulfate distribution, and (3) the radiative feedback of sulfate aerosols on the climate. Barth guided Andrew Church (visitor, SOARS) in examining the transport and distribution of sulfate aerosols emitted from southeast China and Mexico City. Next, the sensitivity to dimethyl sulfide emissions will be examined and parameterizations may be improved to gain better agreement between model results and observations.

Table of Contents	Director's Message
Significant Accomplishments	FY 98 Publications
Community and Educational Activities	Staff, Vistors & Collaborators