Workshop 2




Here is draft of the SEAFLUX workshop report. Let me know if you have any corrections or additions. I imagine that I have left out some things that people have agreed to provide, please remind me if this is the case. Key input from all is needed on the cases (quantity and selection) selected for the intercomparison studies.

Thanks for your input.


Participants


Status Overview

Programmatic Issues

Bill Rossow, new chair of the GEWEX Radiation Panel, reiterated the support of GRP for SEAFLUX. The GRP will be meeting in Nov 2001 and will review SEAFLUX at that time. GRP recognizes that applications of SEAFLUX data will involve surface radiation fluxes and precipitation, and SEAFLUX is encouraged to support applications that include surface turbulent fluxes along with radiation fluxes and precipitation. A new web page for the GRP is being prepared.

Joerg Schulz represented the WCRP/SCOR Working Group on Air/Sea Fluxes. A brief overview was given of the terms of reference of the group, and a summary of the WGASF report, agenda for the workshop in Maryland May 21-24, and furture plans can be found at http://www.soc.soton.uk/JRD/MET/WGASF/index.html

The SEAFLUX web page is undergoing continual updates, and major additions to the page will be made during June 2001.


In situ Measurements

Updates were provided on the COAPS/WOCE, WHOI, ETL data sets. A presentation was made by Duane Waliser on an assessment of the accuracy of various buoy measurements of radiation fluxes. Additional data sets were identified with skin SST measurements and direct turbulent flux measurements. There was considerable discussion of error bars on the data, and consistency among the in situ data sets. There are relatively few direct comparisons of different data sets and measurement techniques (e.g. TOGA COARE). While each of the major data providers have undertaken error analyses on the instruments and measurement method, the methods and criteria are not consistent. Comparison with NWP and satellites may have some limited use in assessing consistency among the different data sources. No particular action items on accuracy and consistency of the in situ data sets were proposed. An additional source of error is in spatial sampling, which is a key issue in using point measurements to validate products on the satellite pixel or GCM grid cell. Low level aircraft measurements over "point" surface-based measurements seem the best way to address this issue.

Action items to obtain new in situ data sets:


Bulk Flux Algorithms

Xubin Zeng and Michael Brunke have begun undertaking a comprehensive evaluation of a variety of bulk flux algorithms against in situ data sets with direct turbulence measurements. The "best" algorithms have errors that are less than the errors for the measured fluxes under conditions of moderate wind speeds and near-neutral stability.


Winds

Presentations on scatterometer winds and wind stress were made by Dudley Chelton and Mark Bourassa. Some debate was made evident regarding the optimal resolution for gridded scatterometer products. Largest errors are in the subtropics, where the satellite coverage is poorest.

Presentations were also made by William Emery on winds from GPS reflectance and by Will Perrie on SAR-derived winds. Kristina Katsaros described a product developed by the French group (Katsaros or Bentamy.). (http://www.ifremer.fr/cersat/english/index.html) that merges SSM/I, scatterometer, and SAR winds (note scatterometer and SAR have less interference from rain than does SSM/I). A presentation was made by Tim Leudtke on requirements for winds to assess surface biogeochemical fluxes.



Skin SST

Presentations were made by Dethloff Stammer and Gary Wick on comparing TMI, Reynolds SST, and IR skin SST. TMI SST accuracy is better than 0.5o C, but accuracy degrades at T< 10o C. Biases of 0.5o C in Reynolds vs TMI SST (and occasional errors exceeding 1o C) were found in the Reynolds SST. Note TRMM does not measure skin SST, but the true bulk SST (about 1 mm depth). TMI retrievals are degraded at wind speeds > 12 m/s. Wick has plans to produce a tropical SST product by merging IR and TMI, including a diurnal cycle that is a function of wind speed. GODAE (subsequent conversation with Craig Donlon) also has plans to produce a skin SST product with resolution of 3 hours and 0.5o lat/lon.



Flux Products

The following flux products were described and assessed:

Some issues that were raised include:

Discussion focussed on determination of the sensible heat flux, Ta, Ta-Ts. The following methods are being used:


Intercomparison Study

A number of specific intercomparison studies were identified. A brief description of the intercomparison protocol is given below, along with the selected cases and periods. An initial list of participants is given, along with a coordinator(s) for each intercomparison study. Note, case coordinators for several studies have not yet been identifed. Volunteers, please step forward. In the absence of volunteers, someone from Webster's NASA EOS IDS project will serve (Curry, Rossow, Emery, Wick, Zhang., Reeder, Rothstein).

Basic valuation and comparisons of all products submitted for a given intercomparison study will be coordinated by the case coordinator. Statistical methods for conducting the evaluations can be discussed online. Individuals are encouraged to conduct in-depth evaluations of their own products.


Bulk Flux Algorithms
Case coordinator: Zeng/Brunke

The study will focus on the following in situ cases, which were selected for geographical scope.

The following algorithms have been incorporated so far into the Zeng/Brunke study.

Others who want their algorithms included in this study should contact Zeng directly ( xubin@gogo.atmo.arizona.edu ) so that your algorithm can be included in the intercomparison.

For each data set and algorithm, basic univariate comparison statistics will be evaluated (e.g. bias, rmse). The bias in the algorithms will also be evaluated for wind speed and stability dependence.


Sea Surface Temperature
Case coordinator: TBD (volunteers?)

An intercomparison of skin SST algorithms will be led by Wick, Emery. This intercomparison will focus on the in situ data sets with skin SST.

Additionally, ETL data sets will be included that measure 1-5 cm temperatures from which skin SST will be calculated using Sandra Castro's bulk-skin model.

In addition to the TMI and AVHRR data, the Reynolds product will also be supplied for these cases.

Identified participants include:

Evaluation and comparison will include basic univariate statistics, plus evaluation of problems of the individual methods (e.g. cloud clearing, precipitation, wind, etc).


Winds
Case coordinator: M. Bourassa

Substantial evaluation of wind products is underway under the auspices of other programs/projects. Hence SEAFLUX will conduct a focused evaluation of scatterometer wind stress products vs direct measurements of wind stress.


Pixel Fluxes, Ta, Qa
Case coordinator: TBD (volunteers?)

Evaluation and intercomparison of surface latent and sensible heat fluxes plus Ta and Qa will be conducted for a select set of data sets that include direct turbulent flux measurements:

Calculations will be made using each of the pixels within a 2 degree range of the in situ measurements.

The following participants have been identified:

Participants are requested to submit their retrievals at full pixel resolution. Participants are encouraged to submit results from their standard flux product as well as to produce new fluxes with alternative input variable data sets and newly developed methods for determining Ta, Qa.


Global Comparison
Case coordinator: TBD (volunteers?)

Evaluation and intercomparison of global surface latent and sensible heat fluxes will be made to assess the spatial variability of the products and to assess the utility of the products in various applications. The year 1999 has been selected for this global comparison. It is realized that products requiring scatterometer winds will only be done for the 2nd half of the year, and that products using TMI can only be produced for the tropics.

The following participants have been identified.

Participants are requested to submit results from both their standard flux product, as well as results from their best "new" method resulting from the pixel intercomparison. Participants are requested to submit their results at full pixel resolution, so we can explore the minimum useful resolution for these products.

In addition to the global comparison, pixel evaluations will be conducted using the large number of ETL data sets collected during 1999, which will provide an evaluation data set independent of those used in the pixel evaluation.


Future Plans

Intercomparison Schedule (near term)

July 1: Bulk flux algorithms submitted to X. Zeng

Aug 1: Web pages for SST, flux pixel intercomparison studies established, with relevant data sets (Rothstein/Reeder) (include calc of skin SST for ETL in situ cases)

Sep 1: New skin SST data sets obtained (Donlon, Wick/Castro, Bradley)

Oct 1: Web page and data prepared for 1999 global intercomparison (Rothstein/Reeder)

Whenever: Obtain new direct turbulent flux data sets (Fairall, Renfrew, Taylor)


BAMS Article

An overview article on the SEAFLUX project will be prepared for submission by end of summer 2000 (draft to be prepared by Curry). The intent of the article is to describe the objectives of SEAFLUX, the data set, and the intercomparison project. All who are contributing to the project (providing data, flux products, etc) will be included as coauthors. A subsequent BAMS article will be considered after the intercomparison is complete.


Newsletter

Approximately once per month, a brief newsletter for the project will be sent to participants. This will serve to update participants on progress, and will remind participants to work on the project .


Next Meeting

The next SEAFLUX Workshop is tentatively scheduled for summer 2002.