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

          The need for short-term warning and forecast products covers a broad spectrum of environmental events which have lifetimes ranging from several minutes to several weeks. Some examples of these events are tornadoes, hurricanes, tsunamis, and coastal storms, as well as "spells" of unusual weather (warm, cold, wet, or dry). Benefits of these products can be measured in terms of lives saved, injuries averted, and expenses spared. NOAA's vision for improvement in this area involves operational modernization and restructuring, strengthening of observing and prediction systems, and improved applications and dissemination of products and services.

          Efforts at GFDL are centered around the development of numerical models which may be used in the prediction of "short-term" atmospheric and oceanic phenomena. Simulations from these models are studied and compared with observed data to aid in the understanding of the processes which govern the behavior of the various phenomena.

          With regard to tropical weather systems, efforts are aimed at understanding and forecasting the genesis, growth, and decay of tropical storms and hurricanes. In extratropical regions, interest includes the development of severe weather systems, the interaction of medium-scale atmospheric flow with that on larger scales, and the influence of the underlying topographic features. Experimental prediction of regional-scale weather parameters weeks in advance is being pursued; included in this context is the study of "ensemble forecasting." With regard to the marine environment, forecasts of coastal conditions on a day-to-day basis can be made by coupling of ocean and atmosphere models. Ocean models are also used to simulate and forecast coastal and estuarine environments, the response of coastal zones to transient atmospheric storms, and Gulf Stream meanders and rings.

ACCOMPLISHMENTS FY99

          Skillful forecasts of intensity have been a challenging goal in hurricane forecasting. For the 1998 season, a new initialization scheme was installed in the National Weather Service (NWS) operational GFDL system, marking the first time that a four-dimensional (4-D) approach to hurricane initialization was taken. When forecasts from 1998 were evaluated, the GFDL model exhibited useful intensity forecasting skill for tropical storms and weak to moderate hurricanes (6.1.2).

          The impact of tropical cyclone-ocean interaction on the intensity of real storms was evaluated using the new GFDL coupled hurricane-ocean model in near real-time for Atlantic cases in 1998. The results showed that the forecast central surface pressure error was reduced by 26% compared to the operational GFDL model (6.3.2).

          An upgrade of the GFDL hurricane forecast system has been initiated, with studies underway on increased resolution (6.4), improved physical parameterizations (6.4), 4-D initialization (6.2.3) and assimilation (6.2.2), and extended range prediction (6.2.1). It is anticipated that these improvements will lead to more realistic and accurate forecasts of storm track, structure, and intensity. The conversion of the system to the computer industry's standard of massively parallel, distributed memory architecture has begun.

          The life cycle of baroclinic eddies in a controlled storm track environment has been examined by means of long model integrations with the Zeta model. A coherent picture has emerged of the growth, maturation, and decay of the eddies within a storm-track. The analysis revealed that the low-level vorticity centers that migrate poleward tend to follow isotachs that closely correspond to the zonal phase speed of the eddies. Since the phase speed of the eddies is a quantity relatively easy to predict, its use could improve the forecast skills of these events by current GCMs (7.2.1).

          Two distinct patterns of eddy evolution have been discovered which are associated with two distinct types of storm track. These patterns can be referred to as "baroclinic wave packets" and "couplets". Baroclinic wave packet require restoration of the downstream surface baroclinicity by packet propagation through a zonally elongated storm track. Alternatively, a basin-scale storm track can support "couplets", self-sustaining structures in which the upstream baroclinic eddy grows through heat fluxes in the area of stronger surface baroclinicity and then fluxes energy to the upper-level downstream eddy. The signature of the couplet at upper levels is an omega-like pattern with two troughs separated by a "building" "ridge. The couplet pattern is frequently observed in storm tracks of the Northern Hemisphere winter. Baroclinic wave packets are more characteristic of the zonally elongated storm tracks that are a frequent feature of the Southern Hemisphere (7.2.2).

PLANS FY00

           The GFDL limited-area non-hydrostatic model will be used to study the interaction between continental convection and radiation. It will also be used to evaluate the representation of physical processes in the cumulus parameterization used for GCM studies of convection. The non-hydrostatic model will also be used to study aerosol residence times and transport in the context of the recently completed Indian Ocean experiment.

           The GFDL hurricane model studies of the increase in grid resolution of multiply-nested meshes will continue. More realistic representation of the eye and eyewall of intense hurricanes, as well as of the large-scale environmental flow, will enhance accuracy in research applications. The evaluation of prediction skill due to increased resolution will proceed with multiple case studies.

           The initialization of the GFDL hurricane model will be improved, with continued development of a 4-D approach in which both routine observations and additional data can be ingested.

           A major thrust for the coming year will be the conversion of the GFDL hurricane forecast system to NCEP's new distributed memory system in preparation for the 2000 hurricane season. This conversion should make it possible to run at high resolution, and with improved physical parameterization packages and ocean interaction.

           The sensitivity of cyclone evolution to position within the storm track will continue to be evaluated. In particular, the mechanisms responsible for the poleward progression of low-level cyclonic eddies will be further analyzed. A clarification of the role that the cyclone-frontal structure plays in timing the meridional temperature gradient reversal and other critical events will be pursued. Model simulations and observations will be used to document in greater detail the dynamics of couplet structures. Further analysis of the differences in frontal development between the various 3-D models will help to elucidate the impact of such differences on the life cycle of the eddies. Work will begin on including moist processes in both the frontal-cyclone structure and the storm tracks.

          Seasonal to interannual climate fluctuations have far-reaching consequences for agriculture, fishing, water resources, transportation, energy consumption, and commerce, among others. Short-term climate anomalies which persist from a season to several years affect rainfall distributions, surface temperatures, and atmospheric and oceanic circulation patterns. Reliable climate forecasts may be used to reduce the disruption, economic losses, and human suffering that occur in connection with these anomalies. NOAA's vision for improvement in this area is based on better predictive capability, enhanced observations, greater understanding of climate fluctuations, and assessment of impacts.

          The study of seasonal to interannual climate fluctuations at GFDL is based on both theoretical and observational studies. Available observations are analyzed to determine the physical processes governing the behavior of the oceans and atmosphere. Mathematical models are constructed to study, simulate, and predict the coupled ocean-atmosphere, land-surface, sea-ice system.

          Simulations based on the numerical models maintained at GFDL, in conjunction with observations, are used to study climate variations on seasonal and longer time scales. Processes under study include large-scale wave disturbances and their role in the general circulation, the effects of boundary conditions such as sea surface temperature and soil moisture, influence of clouds, radiation, and atmospheric convection, and the "teleconnection" of atmospheric anomalies across the global atmosphere. Furthermore, experimental model forecasts are used to evaluate atmospheric predictability and to assess skill in forecasting atmospheric and oceanic climate anomalies, both in general and in connection with the El Niño-Southern Oscillation phenomenon. Also, a more accurate representation of the state of the global ocean is being studied through data assimilation for better initialization of seasonal-interannual forecasts.

ACCOMPLISHMENTS FY99

          A dataset of monthly precipitation and discharge for almost 200 large river basins has been constructed to support diagnostic studies of land water balance and evaluation of hydrologic and climate models. This dataset reveals that the sensitivity of annual runoff to annual precipitation in any basin is determined mainly by the ratio of mean runoff to mean precipitation, consistent with a simple theory which assumes that interannual water storage is negligible (1.6.1).

          A simulation of the effect of Pinatubo aerosols on stratospheric temperatures yields very good agreement with the global-mean observed temperature perturbations over the entire two year period following the eruption. However, the low latitude temperature changes are in agreement with the observations for only the first nine months; thereafter, there is a departure of the simulation from observations despite a substantial aerosol optical depth. Poleward of 30 degrees, temperature changes in both the simulation and observations lack statistical significance (2.5.4).

          The impact of low clouds and thresholds for convective activity on seasonal predictions with coupled models has been examined. Relatively subtle changes in these physical parameterizations can have enormous impact on both the mean of the large-scale model circulation and its variability. Results are being used to improve parameterizations for coupled model prediction (3.2.1.5).

          As part of a collaborative seasonal prediction project, GFDL's experimental prediction model has been compared to models from a number of other centers. For seasonal predictions, the GFDL model was found to have the highest extratropical anomaly correlation and lowest root mean square error, showing that it is one of the best seasonal prediction models currently in use (3.2.1.7).

          Ensembles of coupled model forecasts for the tropical Pacific Ocean temperature were found to split in some years, with some forecasts predicting La Niña and others predicting El Niño conditions after only a few months of integration. The only difference in the forecasts was the details of the atmospheric initial condition. While this is only a model result, it suggests that the seasonal prediction of tropical Pacific Ocean temperatures may be highly non-deterministic for some ocean initial conditions (3.2.2.1).

          An initial version of an adjoint of the Modular Ocean Model (MOM) has been constructed as the first phase of a project to construct a four-dimensional variational ocean data assimilation system for seasonal prediction (3.3.1).

          A study employing a global atmospheric inverse model has found a large North American sink for atmospheric carbon dioxide during the years 1988-1992 of 1.7 gTC/yr. This is 2/3 of the terrestrial sink for the whole Northern Hemisphere (NH). A time-dependent inversion shows that the uptake over NH continents is manifested as increased growing season productivity (4.4.1).

          The influences of El Niño-Southern Oscillation on the precipitation and circulation over the principal monsoon regions of Asia and Australia have been studied using a suite of multi-decadal GCM experiments with prescribed SST anomalies. In broad agreement with the observations, the simulations reproduce the negative summer rainfall anomalies in India and northern Australia, as well as positive winter rainfall anomalies in southeast Asia during warm El Niño episodes. The large-scale anomalous circulation pattern over the monsoon regions is similar to that of a Rossby-wave pattern associated with a condensational heat source or sink in the western equatorial Pacific. Air-sea interactions in the Indian Ocean Basin lead to a negative feedback on perturbations in the monsoon flow in that region (5.3.1).

          The influences of El Niño events on air-sea interactions in the extratropics have been investigated by diagnosing the output from GCM experiments with prescribed SST forcing in the tropical Pacific and two-way air-sea coupling elsewhere in the world oceans. It has been demonstrated that some of the recurrent modes of SST variability in the extratropical North Atlantic and North Pacific on interannual time scales are driven by the overlying atmospheric circulation which is, in turn, responding to SST forcing in the tropical Pacific. The model results are in good agreement with the corresponding diagnoses performed on datasets based on ship observations (5.3.2).

          A one-dimensional ocean mixed-layer model with variable depth has been successfully coupled to an atmospheric GCM. A more comprehensive set of processes has been incorporated in this new version of the mixed-layer model, as compared to the previous version with a constant depth. An ensemble of four 46-year integrations has been completed with this new coupled model. This model will be a useful tool for investigating ocean-atmosphere variability in different parts of the globe on interannual and interdecadal time scales (5.3.4).

PLANS FY00

           The FEOM (Fully Equivalent Operational Model) technique for a substantial speed-up of chemical kinetics calculations will be expanded to account for non-methane tropospheric chemistry. Further analyses of the GFDL GCTM simulations and investigations of the chemistry-transport interactions will continue, including a quantification of chemical species originating from Asia and arriving in North America. The investigation of the seasonal growth and decay of the tropical South Atlantic ozone levels and comparisons with observations will continue. Also, the contributions to tropospheric ozone levels due to intrusion of air from the stratosphere, as well as the quantitative dependence of tropospheric ozone on multiple sources will be estimated.

           Experiments to better understand the impact of clouds on seasonal predictions will be performed. Several new cloud prediction schemes will be evaluated in both atmosphere-only and coupled seasonal prediction models.

           The role of the Madden-Julian Oscillation and tropical westerly wind bursts will be examined using a hybrid coupled model and fully coupled seasonal prediction models. A better understanding of what role transient surface wind events play in initiating El Niño events should result.

           Initial versions of a four-dimensional variational assimilation system for MOM will be constructed. When completed, this system should significantly improve the quality of ocean initial conditions for seasonal prediction.

           New methods for localizing terrestrial carbon sinks are under development that will address questions about the interannual variability in the carbon system. A trajectory analysis will be used to look at sources and sinks affecting individual parcels of air. A new "forest dynamics" model will be employed to look at net ecosystem carbon sequestration.

           The seasonal evolution of anomalies in the extratropical air-sea coupling system during prominent El Niño events will be examined in greater detail using both observational and model-simulated datasets. Emphasis will be placed on the role of the perturbed atmospheric circulation in forcing extratropical SST changes during winter, and the feedback of the ensuing oceanic anomalies on the atmospheric circulation in the spring season.

           The suite of integrations of a coupled model consisting of an oceanic mixed-layer model and an atmospheric GCM will be expanded in an effort to enlarge the ensemble size of this experiment. The increased number of samples will facilitate analyses of inter-sample and inter-event variability of the global atmosphere-ocean system.

          Events such as the Sahel drought, the dust bowls in the Midwest, the Little Ice Age, stratospheric ozone depletion, and global warming may define eras in history. Events such as these have lifetimes of decades to centuries and their causes can be either natural or anthropogenic. An ability to predict such changes and to assess the causes is essential in long-range policy making. Adapting to these changes and reducing the effects of human activities will require enhanced predictive capability. NOAA's vision for improvement in this area is based on a commitment to research in climate and air quality, as well as to insure long-term climate and chemical records.

          The related research efforts at GFDL require judicious combinations of theoretical models and specialized observations. The modeling efforts draw on principles from the atmospheric, oceanic, chemical, and biological sciences. One area of focus is long-term climate variability and secular change associated with the atmosphere and oceans. This area encompasses a number of topics, including the effects of changes in the concentration of atmospheric gases such as carbon dioxide, the simulation of past climates, and the variability of the oceanic thermohaline circulation. Another area of focus is the formation, transport, and chemistry of atmospheric trace constituents. This area addresses problems such as: the transport of quasi-conservative trace gases; the biogeochemistry of climatically significant long-lived trace gases; the transport, sources, and sinks of aerosols; the chemistry of ozone and its regulative trace species; the effects of clouds and aerosols on chemically important trace gases; and the impact of anthropogenic chlorofluorocarbons on stratospheric ozone amounts. Yet another area of focus relates to the modeling of the marine environment. It includes the dispersion of geochemical tracers in the world oceans, the oceanic carbon cycle and trace metal geochemistry, and ecosystem structures.

ACCOMPLISHMENTS FY99

          A climate model has been successfully integrated for over 1,000 years with twice the resolution in both atmosphere and ocean as previous integrations of this length at GFDL. The model has more vigorous atmospheric eddies, a better defined intertropical convergence zone, and a less diffusive ocean than the previous, lower resolution model. The model exhibits enhanced variability on the multi-decadal time scales of particular importance for the issue of climate change detection. The simulated variability involves large-scale exchanges of heat and fresh water between the Arctic and the North Atlantic (1.2.1).

          The connection between the Arctic Oscillation (AO) and Northern Hemisphere surface temperatures has been analyzed in the control run of a climate model and in observations. The control integration shows a geographical distribution of temperature anomalies associated with the AO that is very similar to that observed, but the hemispheric mean of this pattern is smaller than that obtained with the last few decades of the observed record. Analysis of the observations suggests that the connection between the AO and Northern Hemisphere mean temperature in recent decades is not representative of earlier periods (1.2.4).

          An ensemble of five global warming scenario integrations, run over the period 1865-2089, is nearing completion, using a model with twice the resolution in both ocean and atmosphere as compared to previous scenario integrations at GFDL. The ensemble mean warming is broadly similar, but delayed by roughly 10 years as compared with the earlier model. The geographical distribution of simulated surface temperature trends is consistent with observations over the last several decades in most regions, taking into account the internal variability as estimated from the model (1.3.1).

          One member of the set of five global warming scenario integrations with the latest version of the climate model captures the time evolution of the 20th century Northern Hemisphere temperature record -- warming in the 1920's and 30's followed by a period of nearly steady temperatures until the 1970s, when a more uniformly distributed warming begins. The early 20th century warming in this realization is primarily a result of the model's internal variability. In contrast, the more uniform warming starting in the 1970's is broadly reproduced by every member of the ensemble (1.3.1).

          The cause of the reduction in strength of the Atlantic thermohaline circulation in climate model simulations of global warming has been examined with a set of five multi-century low resolution simulations. Changes in net surface fresh-water fluxes, including runoff from land, are found to be responsible for about two-thirds of the weakening of the model's thermohaline circulation in its global warming scenario. Surface heat flux changes associated with the warmer temperatures account for the remaining third. Wind stress variations have negligible impact on changes in the model's overturning strength in these scenario integrations (1.3.2).

          In global warming simulations, the surface westerlies and the midlatitude storm track in the Southern Hemisphere shift polewards. This shift has been found to be very similar to that produced by one of the phases of the dominant pattern of month-to-month zonal wind variability in the Southern Hemisphere, found both in the model and in observations. Atmospheric integrations with fixed sea surface temperatures and sea ice show that the poleward shift can be captured by simply increasing sea surface temperatures uniformly (1.3.3).

          Accurate steady state simulations with a control, a doubled CO₂, and a quadrupled CO₂ scenario have now been achieved with a low resolution coupled model. The lengths of these integrations range from 4,000 to 12,000 years. The thermohaline circulation in the Atlantic ocean, which is responsible for most of the oceanic poleward heat transport in the present climate, is remarkably similar in pattern and amplitude among the three integrations. This contrasts with the transient weakening of this model's circulation in the first few centuries following a buildup of CO₂ (1.3.5).

          Using a low resolution atmosphere coupled to a mixed-layer ocean and a sea ice model, the climatic response to changes in the Earth's orbit over the past 120,000 years has been simulated. Changes in the Earth's orbit are a prime candidate for forcing glacial-interglacial fluctuations of climate. The orbital variations are accelerated by a factor of 30 to reduce the length of the computation. Initial analysis shows monsoon climates to respond more strongly to change in the tilt of the Earth's axis than to precession of the equinoxes (1.4).

          Several multi-decade integrations of a moderate-resolution version of the SKYHI model have been performed. These reveal quite significant spontaneously-generated decadal-scale variations in the Northern Hemisphere extratropical stratospheric circulation, behavior that has implications for climate change detection and attribution. Results from further perturbed model experiments suggests that much of the effective interannual "memory" of extratropical stratospheric circulation resides in the tropical mean winds (2.4.4).

          The vertical profile of the temperature trend in the northern midlatitude (~45°N) stratosphere (~16 to 50 km) over the 1979-1994 period indicates a statistically significant cooling at all altitudes, and is consistent among several different datasets. The cooling is about 0.8 K per decade over the 15-35 km region, and increases with height to 2.5 K per decade at ~50 km. While ozone loss contributes strongly to the cooling in the lower stratosphere, changes in ozone and well-mixed greenhouse gases (notably, CO₂) contribute to the middle and upper stratospheric cooling. However, the observed trend at the higher stratospheric altitudes exceed those estimated by present model simulations (2.5.1).

          A simple theoretical framework has been developed for integrating the effects of Southern Hemisphere winds and eddies into the global thermohaline circulation. This framework qualitatively predicts the responses seen in more complicated three-dimensional ocean GCMs (4.1.2).

          A new model for glacial-interglacial changes in atmospheric CO₂ has been developed which views reduced ventilation of the ocean's deepest water as the key process in glacial CO₂. This new model does not rely on increased biological production or changes in nutrient chemistry to reduce atmospheric CO₂ (4.4.6).

          A new approach for estimating global rates of N2 fixation and denitrification has been applied to the Pacific Ocean using WOCE data. The N2 fixation rate has been estimated to be 40 Tg N/yr (4.4.7).

          A project aimed at assessing the feasibility of estimating long-term climate trends on the basis of historical radiosonde measurements has been completed. This project made use of advanced statistical techniques and an extensive compilation of meta-data describing historical changes in radiosonde instruments at various station sites. It was found that trend estimations were highly sensitive, both locally and globally, to the particular method used to identify instrumental changes in the station records (5.1.1).

          It was found that the global-mean temperature anomalies, as estimated by Microwave Sounding Unit measurements from satellite platforms, agree remarkably well with the corresponding lower tropospheric temperatures computed from the objectively-analyzed global radiosonde network. The high level of consistency between the two records had not been previously demonstrated and lends credence to both the satellite-based and ground-based observations (5.2.2).

          A series of experiments were performed using the GFDL Hurricane Prediction System coupled to a regional version of the Princeton Ocean Model (POM) to evaluate the role of an interacting ocean in the intensification of hurricanes during a global warming scenario. Preliminary results confirm that intensification still occurs under elevated CO₂ conditions even when hurricane/ocean coupling effects are included. However, a larger ensemble of experiments appears to be necessary to quantify how the hurricane/ocean coupling affects the degree of CO₂ warming-induced intensification and how this effect may vary from one tropical basin to another (6.3.1).

PLANS FY00

           Control runs of the new coupled climate model will be continued and the internal variability of the model analyzed in detail. Atmosphere-only integrations will be studied to help characterize the dynamics of the dominant mode of inter-decadal variability in this model. Extreme outliers in the internal variability of a 12,000 year control integration of the lower resolution coupled model will also be studied.

           The size of the global warming scenario ensembles will be increased, including integrations using several of the IPCC-2000 scenarios. The ensemble member that has been found to closely resemble observations in its 20th century temperature record will be the subject of particular scrutiny.

           New atmospheric integrations will be designed to study the dynamics of the poleward shift of the Southern Hemisphere circulation in global warming simulations and its relationship to the dominant pattern of month-to-month variability of the winds in the Southern Hemisphere. This Southern Hemisphere response will be contrasted with the more complex dynamics of the Northern Hemisphere circulation's response to global warming.

           The recently completed simulation of the climatic response to the orbital variations of the past 120,000 years will be studied extensively, with particular emphasis on the relative importance of obliquity and precessional changes for a variety of climatic regimes and indices.

           Simulated temperature trends in the global stratosphere (extending to the middle and upper stratosphere) will be compared with observational data. Both natural and anthropogenic factors, and forced and unforced variations will be considered, and their respective contributions to the trends will be evaluated.

           Analysis will continue of some SKYHI integrations with enhanced vertical resolution which display long-period oscillations in the tropical stratospheric mean winds with properties very similar to the observed quasi-biennial oscillation. This analysis will focus on understanding the details of the eddy-mean flow interactions responsible for forcing the mean flow oscillation.

           A major study is underway to determine the role of mesoscale eddies in balancing the northward Ekman transport in the latitude band of the ACC. A series of models with progressively finer resolution will illustrate how the ACC changes in response to variable winds when eddies are explicitly resolved.

           A more temporally homogeneous radiosonde temperature data base suitable for detecting climate trends will be assembled. Artificial discontinuities due to instrumental changes in a 86-station network will be identified. An adjustment procedure aimed at removing the artificial variability will be refined and applied where possible. The adjusted data will then be used to assess long-term trends and compared to a number of observational and GCM datasets for the assessment of global change.

          A number of the research topics at GFDL cut across the various time scales characteristic of each of the foregoing sections. Progress on these topics impacts many other research areas which depend critically on the successful representation of numerous lower-level processes which are common to problems at all scales. Topics which fall into this category include hydrological processes, radiative transfer (including the effects of aerosols and clouds), cloud prediction/specification, "teleconnection" of atmospheric anomalies across the global atmosphere, satellite data interpretation, transport processes, gravity wave effects and parameterization, model resolution effects, and many other model enhancement efforts. As these processes become better understood and more accurately represented, benefits will accrue to a multitude of other research efforts.

ACCOMPLISHMENTS FY99

          As preparation for the construction of the next-generation coupled climate model, the ocean model has been upgraded to include variable meridional resolution (with higher resolution in the tropics to improve ENSO simulations), a free upper surface that enables a more realistic treatment of surface fresh water fluxes, Gent-McWilliams mixing of tracers, an explicit mixed layer parameterization, and partial cells at the ocean bottom to better resolve bottom topography. The resulting water mass structure of the world ocean is greatly improved over that produced with the ocean model currently in use for global warming studies (1.5.1).

          By analyzing a cloud-resolving non-hydrostatic model of an atmosphere in radiative-convective equilibrium, novel techniques have been developed to analyze the energy and entropy budgets of deep moist convection. It has been found that frictional dissipation is not primarily a consequence of a turbulent cascade of energy to small scales but instead mostly occurs in the shear zones surrounding individual hydrometeors. Equivalently, the energy cycle of the convection is dominated by the potential energy needed to lift water, not by the kinetic energy flowing through the convective motions (1.7.2).

          The new shortwave radiation parameterization for GCMs, developed on the basis of benchmark computations, yields errors of less than ~15% in the cloud absorbed flux and less than 5% for atmospheric absorption for a variety of cloudy sky cases studied. The improvements in the shortwave radiation parameterization have the effect of enhancing the solar absorption by O₃, H₂O, O₂ and CO₂ such that a statistically significant increase of the low latitude stratospheric temperature occurs in the SKYHI GCM simulation (2.1.1).

          Using distributions of various aerosol species, as simulated by three-dimensional chemistry-transport models, and a well-calibrated solar radiative transfer parameterization, spatial patterns indicative of the presence of tropospheric aerosols are identified in the satellite clear-sky observations of reflected solar radiation over the oceans. While geographical signatures due to both natural and anthropogenic aerosol species are manifest, the naturally-occurring sea-salt is the leading aerosol contributor to the global-mean clear-sky radiation balance over the oceans (2.1.4).

          A cumulus parameterization which includes microphysics at the scales of deep convection and associated upper-tropospheric stratiform clouds has been implemented in SKYHI. SKYHI integrations with this parameterization reveal the important role in the climate system of mesoscale processes associated with deep convection, and demonstrate a physical basis for modeling interactions between convection, clouds, and radiation (2.2.1).

          Experiments with the GFDL limited-area non-hydrostatic (LAN) model have identified the roles of ice inflow and outflow, ice sedimentation, and ice-crystal size and shape in regulating the radiative energy balance of tropical convection. Both top-of-atmosphere and ocean surface radiative balances can be realistically modeled when these factors are taken into account appropriately. The LAN model has also been used to study sulfate transport and transformation in deep convective systems. Indirect radiative effects of aerosols in upper-tropospheric cloud systems can result from these processes (2.2.2).

          A new single column model using the Flexible Model System (FMS) framework has been prepared and compared to observations, as well as single column models and limited-area non-hydrostatic models from other institutions. The single column model has proven to be a useful tool for testing the parameterizations of global atmospheric models. New parameterizations of atmospheric vertical diffusion and cloud microphysics have also been incorporated into the FMS framework (2.2.3). A new parameterization of the effects of horizontal cloud inhomogeneity improves the physical basis of cloud microphysics in global atmospheric models (2.2.4).

          A database of sources of CO and NO_x related to biomass burning has been developed, evaluated in the GFDL GCTM, and found to have collective uncertainties much less than the factor of two suggested by previous estimates (2.3.3). A realistic CO distribution has been simulated by the GFDL GCTM and the sources responsible for its seasonal variation have been identified (2.3.4).

          The GFDL GCTM has successfully simulated the tropospheric column ozone maximum over the tropical South Atlantic Ocean. Transport of precursors from biomass burning and lightning are shown to be major contributors (2.3.5). The GFDL GCTM simulations of the impact of Asian emissions on western U.S. air quality find that surface ozone contributions in 1990 are a modest 5-6 ppb ozone during late spring and summer, but may jump to 10-20 ppb by the year 2020 (2.3.6).

          The simulated atmospheric distribution of carbonaceous aerosols from fossil fuel burning, as obtained using the SKYHI GCM, compares reasonably well with observations made from a variety of sites (e.g., Sable Island, Bondville, Mauna Loa). The substantial spreading of anthropogenic black carbon aerosol to oceanic regions gives warming radiative perturbations due to this aerosol species at considerable distance from the source regions (2.3.7).

          Analysis of the spectrum of kinetic energy has been performed using results from a very high resolution version (1/3° latitude) of the SKYHI model. The model kinetic energy spectra in the troposphere display the observed transition from steep spectral slopes at wavelengths greater than about 500 km to a shallower mesoscale regime at shorter scales. The detailed model results allow a diagnosis of the processes that maintain the eddy energy in the mesoscale. A key finding is that the nonlinear transfers of energy are downscale throughout the entire range of resolved scales (2.4.5).

          Stratospheric age-of-air calculations in the SKYHI GCM using centered second-order and fourth-order advective schemes more closely match observed ages than calculations using the diffusive semi-Lagrangian or Lin-Rood schemes. Short-term transport calculations, however, can be significantly distorted by the dispersive properties of the centered schemes. The optimal choice of advection schemes thus appears to be a sensitive function of the nature of the problem being addressed (2.4.7).

          Using a detailed microphysical model and observations from the Monterey Area Ship Track experiment, it has been found that aerosol-cloud interactions, and the impact of these interactions on the "indirect" aerosol radiative forcing, are more complex than the conventional "Twomey" effect in which cloud liquid water content is assumed to be fixed under perturbations of aerosol concentrations. In addition to the changes that liquid water can undergo when aerosol concentrations increase, the resulting changes in cloud drop-size distributions and cloud albedo are sensitive to the nature of the polluting source and the chemical composition of the aerosol species emitted (2.5.3).

          GFDL benchmark calculations of fluxes at 6.7 micron infrared wavelength have been used in an intercomparison exercise to identify biases in radiation models used for satellite retrievals, in particular those arising due to the treatment of water vapor absorption. A SKYHI GCM study of the radiative effects of the non-CO₂ well-mixed greenhouse gases using the new longwave radiation parameterization indicates that the temperature response of the tropical troposphere tends to be "dynamical" in nature, while that in the stratosphere tends to be "radiative" (2.5.5).

          Two new atmospheric dynamical cores, a large assortment of physical parameterizations, a comprehensive coupler, and numerous modeling support tools are now available as part of GFDL's new Flexible Modeling System (FMS). Long atmosphere-only integrations have been produced with FMS and results have been unexpectedly good. Coupled models for seasonal prediction have also been constructed and are being tested (3.1).

          A software package for supporting geophysical models on parallel computer architectures has been constructed and tested on a variety of parallel computer platforms. Parallel versions of GFDL's FMS dynamical cores have been constructed and tested, and good scaling behavior has been obtained (3.1.2).

          A statistical atmospheric model for use in the evaluation of ocean models destined for coupled model integrations has been completed. This statistical model provides a more flexible and economical mechanism for testing ocean models than coupling to a comprehensive atmospheric model (3.2.1.2).

          A new sea-ice model has been developed for coupling to oceanic and atmospheric models using the GFDL Flexible Modeling System framework. Sea-ice motion in the new model is a product of internal stresses within the ice, in addition to the motions forced by the wind and underlying ocean currents (4.1.4).

          A new shear- and stratification-dependent mixing scheme has been developed for simulating down-slope flows in isopycnal ocean models. The new scheme is able to reproduce the penetration of salty Mediterranean water into the interior of the Atlantic Ocean (4.1.5).

          A new version of GFDL's Modular Ocean Model (MOM 3) is nearly ready for release to the oceanographic community as public domain software. MOM 3 is an upgrade which allows MOM to be run on efficiently on the current generation of parallel computers. MOM 3 includes options for partial-depth bottom cells, a terrain-following bottom boundary layer, improved along-isopycnal diffusion, the Gent-McWilliams eddy-induced advection effect, and an explicit free surface (4.2.1).

          An inverse model has been developed for determining the fluxes of oxygen and CO₂ across the ocean's surface. The model uses ocean tracer distributions and the circulation fields of an ocean GCM to estimate O₂ and CO₂ fluxes (4.4.5).

          A global model of the oceanic cycling of silica exhibits strong sensitivity to the representation of lateral exchange and vertical mixing. Silica flux observations place a firm upper limit on low-latitude mixing and require eddy-induced advective fluxes at high latitudes (4.4.8).

          The mean diurnal cycle of satellite-observed upper tropospheric relative humidity, fraction of upper tropospheric cloud cover, and fraction of deep convective cloud cover has been documented. A fundamental difference in the diurnal relationship of deep convection and relative humidity was noted between tropical ocean and tropical land regions. Variations in humidity occurred in phase with deep convection over land, but nearly 12 hours out of phase with that over ocean. This difference was shown to be associated with differences in the temporal evolution of convection at various vertical levels over the tropical oceans (5.2.1).

          A team composed of representatives from various research groups at GFDL has been assembled to construct a web-based tool for displaying and comparing various observational and model datasets. A user-friendly package has been designed for visualizing and validating the output of the modularized GCMs currently under development. Considerable progress has been made in the selection of model variables to be diagnosed, as well as the organization and standardization of data processing procedures and the formulation of quantitative measures for assessing model performance (5.4).

          A combination of high-resolution numerical and analytical models have been used to test the feasibility of a proposed parameterization of total mountain drag due to unresolved terrain. The parameterization assumes steady, linear gravity waves and gradual horizontal variations of the basic flow. Given the low-level buoyancy frequency and topography, it yields a map of the divergent horizontal velocity due to the terrain. The total drag on the atmosphere can then be obtained by multiplying this velocity perturbation by the vertical velocity due to the resolved horizontal surface wind (7.3.3).

PLANS FY00

           A new spectral atmospheric model will be finalized for inclusion in the FMS. A T42 version of the model, with physics similar to that of the present atmospheric model, will be coupled to new land and ice models and a mixed layer ocean for sensitivity studies. Tests will also begin of the new MOM3 ocean model at several resolutions coupled with this spectral model. In a parallel development, the physics of the atmospheric model will be revamped, with initial focus on water vapor advection, surface fluxes, the planetary boundary layer, and the diurnal cycle.

           Analysis of solar radiative benchmark computations for more complex cloud systems such as multi-layered and non-plane-parallel cloud systems will continue. Additionally, calculation of longwave benchmarks will be continued to support development of parameterizations for non-gray absorbers in the thermal infrared.

           The chemistry-transport problem affecting the geographical distribution of various aerosols will continue. Plans call for simulation of the transport of sulfate aerosols, to be followed by dust, organics, and sea-salt. Simulations of the microphysics of stratospheric aerosol distributions following volcanic events will be pursued, and the accompanying radiative impacts will be evaluated, as will the radiative forcing and climatic effects due to non-well-mixed species in the troposphere and stratosphere.

           Aerosol-cloud interactions will continue to be investigated from a range of perspectives from microphysical and chemical bases, to cloud-resolving and GCM spatial scales. Both observational data and model simulations will be utilized in this investigation. Various methods for addressing the indirect climate effect of aerosols on the model's cloud microphysics will be evaluated.

           Cloud-radiation-climate interactions will be pursued using the cloud-resolving model and the laboratory's GCMs under the FMS framework. Emphasis will include examination of available observational data on water vapor, clouds, and the energy budget in order to test, evaluate, and develop the physical components that govern their spatial and temporal distributions.

           Interactions between deep convection and radiation will also be studied using the FMS framework, and large-scale cloud systems will be linked with deep convection. These studies represent ongoing applications of a GCM cumulus parameterization which includes microphysics and mesoscale processes. Processes governing the formation and decay of cloud systems in GCMs will be studied in order to determine the principal factors controlling cloud amounts. The impact of the new parameterizations of vertical diffusion and cloud microphysics on the climate of the new FMS will be documented.

           The spectral budget studies in the high-resolution SKHYI model simulations will be extended to investigate the mechanisms leading to the maintenance of spectra of conservative trace constituents.

           New models that are part of the GFDL FMS will become the primary tool for scientific research in seasonal/interannual prediction. The enhanced capabilities of these models and their accompanying diagnostic tools should accelerate research progress.

           A parallel version of the complete FMS will be completed. Remaining challenges are the coupler, diagnostic support tools, and physical parameterizations.

           A version of the MOM ocean model that uses the capabilities of the GFDL FMS will be constructed. This will assist in the design and analysis of coupled model experiments.

           A new version of MOM (MOM 4) is under development for use on the next generation of massively parallel computers.

           An improved determination of the elemental ratios in marine organic matter will be undertaken using a detailed analysis of oceanic tracer distributions. This project will allow for more accurate modelling of oceanic biogeochemical cycling.

           Observations from a network of four geostationary satellites will be combined to provide near-global coverage of the diurnal cycle in upper tropospheric relative humidity. An algorithm will also be developed to examine the diurnal variations in the total column water vapor. The relationships between diurnal changes in convection, total column water vapor and upper tropospheric water vapor obtained from satellite observations will be compared against the corresponding results based on atmospheric GCM simulations.

           Development and enhancement of a webtool for visualizing and comparing model-generated and observational datasets will proceed, with the goal of achieving faster access to the desired data fields and greater ease in selecting various data domain and graphics options. Upon adequate testing of the various functions of this tool, it will be made available to the general GFDL community, as well as interested external users.

           To celebrate the conclusion of the current form of the NOAA/Universities collaborative project for model diagnosis, a comprehensive review of the accomplishments of this decade-long effort will be conducted in 2000. Presenters of the reviews on various subject areas will organize the pertinent material in a written form. The resulting set of manuscripts will be published in a special issue of a journal, or a scientific monograph.

           The newly developed mountain drag parameterization will be tested in one or more additional GCMs being used within the laboratory. Of particular interest will be whether the scheme makes it easier to match the present climate in long-term experiments and whether the drag is more important for steeper mountain ranges like the Andes. Further high-resolution runs with the limited-area model will focus on this issue, as well as the effect of moist physics on the drag. An effort will be made to explain analytically the sharp differences in drag caused by allowing precipitation to fall from moist flow over orography.

*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).