The first Workshop on Ocean Surface Turbulent Fluxes was organized by the GEWEX Radiation Panel, with cooperation of the JSC/SCOR Working Group on Air/Sea Fluxes and several members of the U.S. CLIVAR Panel and sponsorship by U.S. NASA and NOAA. A total of 35 scientists attended the workshop, representing the U.S., U.K. Australia, Japan, France, and Germany. While all components of surface heat, freshwater, and momentum fluxes are of great interest to the global climate community, this Workshop focused specifically on the turbulent fluxes (note: there are separate projects in the GEWEX Radiation Panel that address the surface radiation fluxes and precipitation).
The goals of the first Workshop were to:
A summary of the workshop findings and recommendations is given below. Further details on the workshop, including summaries of the individual panels and the working group reports, can be found at http://paos.colorado.edu/~curryja/ocean/home.html (see Workshop 1).
Requirements for surface flux datasets have been articulated by a number of different WCRP and SCOR programs, including CLIVAR, the GEWEX Radiation Panel, the JSC/SCOR Working Group on Air/Sea Fluxes. The GEWEX Radiation Panel and the U.S. CLIVAR Committee require a 1o spatial resolution, 3-6 h time resolution, and accuracy of 5 W m-2 for individual components of the surface heat budget. These high frequency flux values are needed because of the nonlinearities in the coupled system: for example, diurnal modes have been shown to impact intraseasonal variability, which must be simulated accurately to predict interannual variability. Particularly for coupled models, accurate observations of high-resolution surface flux components are needed to identify and resolve problems such as the model drift that characterizes most coupled models. Achieving the objectives of regional field experiments and the GEWEX Cloud System Study experiments also requires high-resolution surface fluxes. While a long-term global flux dataset is desirable, particularly for studies of interannual variability, it is probably only feasible to go back as far as 1987, given the availability of appropriate satellite sensors.
Bulk flux algorithms link turbulent fluxes to mean values of surface temperature, wind, and surface air temperature and humidity. A variety of different bulk flux algorithms are presently used, with the most recently developed research-quality algorithms showing fairly good agreement with each other and with observations in conditions of moderate wind speeds and neutral or unstable conditions. The following are outstanding issues in bulk algorithms: conditions of light wind and stable stratification, influence of sea spray, treatment of sea state (swell, directional effects), appropriate averaging scales, parameterization of mesoscale gustiness, and behaviour of scalar sublayer transfer.
The importance of high quality in situ turbulent flux reference measurements was emphasized. At present, high quality turbulent flux measurements are available from research ships during field experiments, and increasingly from flux buoys; however the amount of data available is small. Until recently, the technology has not been available to obtain direct flux measurements during precipitating or sea spray conditions. Existing flux datasets from the research ships require considerable processing and formatting before they are suitable for use by the community. Operational buoys and voluntary observing ships provide better data coverage, but poorer quality data and no direct flux measurements. More high quality flux reference measurements are needed. Flux buoys are recommended in regions with high winds, strong currents, and/or highly variable atmospheric surface temperature and humidity. Preferred locations for flux reference sites include: water mass transformation regions in high latitude seas; the Gulf Stream; and the Gulf of Mexico.
Several global wind datasets are available, including satellite, ship and NWP products. Passive microwave radiometers provide the foundation for several global data sets of wind speed. Scatterometers provide the most promising data stream for vector winds. Algorithms for determining winds from scatterometers are highly empirical and realistic error models are needed, or algorithms will be tuned incorrectly. Parametric issues in scatterometer wind determinations include: retrievals at high and low wind speed extremes, stratification extremes (stable, unstable), retrievals in precipitating conditions, and effects of surface currents and waves. Three scatterometers (QuikSCAT, ADEOS-II and ASCAT) may be operating simultaneously during the 2001-2006 time period; more research is needed to determine the time and space resolution for which vector winds can be determined from multiple scatterometers. For scalar winds, combination of scatterometer and passive microwave measurements should provide the needed time/space resolution. The scatterometer community is conducting some in situ evaluation of wind products, but there is no systematic comparison of different products with a reference dataset of in situ measurements. It is not straightforward to evaluate vector wind products using in situ measurements.
Operational sea surface temperature products estimate bulk sea surface temperature, not surface skin temperature, on weekly time scales. Skin temperature is the parameter required by research-quality bulk turbulent flux algorithms, so operational SST products do not meet the requirements. A new SST product is needed for this project, with the following enhancements relative to operational products: skin SST; diurnal cycle; and values under cloudy conditions. New sensors (ASTR2, TMI) and new analysis techniques that use result from ocean mixed layer models can provide the foundations for this new SST dataset. Increased field observations of skin SST data globally are required to evaluate such a dataset.
Because the surface air temperature (Ta) and near surface humidity (qa) are not directly retrievable from space with currently available sensors, all existing methods determining both quantities rely on indirect methods. To determine qa, almost all existing methods make use of a vertically integrated water vapor content, typically derived from passive microwave instruments. Determinations of Ta are even less satisfactory, typically invoking a specified value of relative humidity that is used with the qa value. It is hoped that new profiling instruments like those proposed for NPOESS will provide better accuracy for the boundary layer. To improve the accuracy and time resolution of Ta and qa, it has been suggested to use available profile information, indirect diagnostic relationships determined for different climatic regimes and 4DVAR assimilation available satellite data. The bottom line is that the Ta, qa problem is difficult; however it hasn't been worked as hard as the wind and SST fields, so further efforts may prove fruitful. For independent verification of existing methods, more high-quality buoys are needed as reference sites.
In addition to NWP products and flux climatologies derived from in situ measurements, there are several satellite-derived global and regional products of the vector momentum flux and sensible and latent heat fluxes. The satellite products have not been systematically compared or evaluated with in situ data, although for all datasets the accuracy of the flux products improves with decreased space/time resolution. A comprehensive regional satellite analysis during TOGA COARE highlighted the capability of the satellites to capture much higher resolution surface fluxes than did the ECMWF analyses; for example, satellites were able to capture the impact of a major westerly wind burst during the IOP that was largely missed by ECMWF analyses. While researchers are encouraged to use the existing satellite flux products, further improvements seem feasible and should be forthcoming as a result of the proposed activities of this Working Group.
Because the available flux algorithms and satellite retrievals of input variables have not been thoroughly tested in all meteorological regimes, comparisons of different methods against each other using common input data and against in situ measurements are needed to evaluate the current capability and support development of improved analysis capability. We plan to proceed with intercomparison of flux algorithms, input datasets and flux products in the following way:
Conclusions:
Recommendations:
The second Workshop on Ocean Surface Turbulent Fluxes will be held in New York at NASA GISS, sometime during Autumn 2000, at which initial comparison results will be examined and plans for a global processing approach discussed.
