SUMMARY:

TOGA COARE Turbulence and Large-Scale Flux Planning Meeting

San Diego, March 23-25, 1994

INTRODUCTION

1) to follow up issues concerning the bulk formulae computation of fluxes raised at the first flux meeting in Boulder, Colorado in September, 1993;

2) to review the status of turbulent measurements made on ships and aircraft during COARE;

3) to review the computation of turbulent fluxes by inertial dissipation and eddy covariance methods, including agreement on methodology, test cases, and intercomparisons needed prior to the Toulouse workshop;

4) to relate the platform-specific (buoy, ship, and plane) near-surface observations and flux estimates to other data sets, including precipitation, satellite sea surface temperature, and estimates of fluxes and their area-average from the sounding array;

5) to plan for the activities of the flux group in the lead-up to and during the Toulouse workshop.

Dave Rogers and his staff made excellent local arrangements.

Attendees included Steve Anderson, Frank Bradley, Sean Burns, Richard Chinman, Carol Anne Clayson, Meghan Cronin, Peter Coppin, Lynn deWitt, Chris Fairall, Carl Friehe, Alan Grant, Denise Hagan, Phil Hignett, Hiroshi Ishida, Dick Johnson, Djamal Khelif, Guosheng Liu, Bob McMillen, Mike McPhaden, Clayton Paulson, Rob Pinkel, David Rogers, Paul Spyers-Duran, Osamu Tsukamoto, Ed Walsh, Alastair Williams, Bob Weller, Gary Wick, Xiashua Yang, and Guang Zhang.

Bulk Formulae Fluxes

Frank Bradley updated the results of the first intercomparison day (November 28, 1992). He pointed out that this day was somewhat unrepresentative of the few days that preceded it, being characterized by very light wind and strong diurnal warming of the sea surface temperature (SST). In these conditions, mean air temperature measurements from Franklin, Moana Wave, the WHOI mooring, Hakuho-maru, the 2 P3 aircraft, Wecoma and other platforms showed a spread of 3C. Unaspirated sensors on stationary platforms were particularly susceptible to radiative heating. Some differences, however, may be attributable to spatial variability, as evidenced by differences in sea surface temperature between the WHOI mooring and Wecoma at 12 km separation. (Carl Friehe points out that the spread in temperature measurement among the three turboprop aircraft was around 0.3C - they are free of solar heating problems, but may suffer other errors due to the dynamic heating factor).

Humidity agreement was good. The Wecoma dewpointer was 1 gm kg-1 low, probably an error in calibration which may be removed when the instrument is calibrated at WHOI. Among the other measurements the scatter was approximately 0.5 g kg-1.

The SST intercomparison included infrared radiometers on Franklin, the 2 P3's, and Electra in addition to the ship and buoy measurements. Hakuho-maru's SST suggested a bias which is being investigated. Differences between the radiometric and in-situ sensors indicated a difference between bulk and skin SSTs on average 0.2 to 0.25C; more discussion of the differences between skin and bulk SSTs was held later in the meeting. The intercomparison also highlighted the differences in SST to be expected when sensors are located at different depths in the strong near-surface temperature gradient caused by penetrating solar radiation. Differences between temperatures measured at a depth of 2.4 m and skin temperature were as high as 3C. Bradley pointed out that such uncertaities in SST can have a large effect on Qnet when the effects of net longwave, latent and sensible heat flux are combined.

The differences in incoming longwave radiation observed during the intercomparison can be largely explained by the way the pyrgeometer is used. The stock Eppley precision infrared radiometers (PIR) employ a battery and circuit to internally correct the sensor output for dome and body heating. This is less reliable than directly measuring thermopile output and body and dome temperatures, as was done by Chris Fairall on Moana Wave. His measurements were therefore taken to be the standard and other pyrgeometers can be reasonably well corrected for dome heating using the procedure of Alados-Arboledas et al (1988). Radiative transfer models of incoming longwave were run by Steve Cox and by Tom Ackerman to check the absolute longwave measured on Moana Wave; they were within a few Wm-2 .

Differences in incoming shortwave during the November 28 intercomparison were reported at the Boulder planning meeting. Cross-checks between the Franklin, Moana Wave and IMET pyranometers had also been performed at WHOI and confirmed 8% differences. A second set of radiometer comparisons have just been completed in Australia by Bradley. It included an instrument from the equatorial, 156E Proteus mooring and the Wecoma pyranometer, which was calibrated against the Aust. Bureau of Meteorology pyrheliometer at the end of the tests. Since the result agreed exactly with the pre-cruise factory calibration, Wecoma's instrument will be taken as reference and through the intercomparisons all other pyranometers can be adjusted to about 1%.

Comparison of Franklin and Moana Wave wind speeds showed Franklin consistently low by 0.4 ms-1. Moana Wave and the WHOI mooring were in agreement while Hakuho-maru was closer to Franklin; so the discrepancy remains unresolved. Peter Coppin has looked at flow distortion effects using a model of Franklin in a wind tunnel and will pursue these studies further. Bradley pointed out that for the fluxes, the winds should be those relative to the water. Weller's current meter at 5 m depth at the WHOI mooring showed a peak ocean current of 80 cm s-1 and typical speeds of 30 - 40 cm s-1 . These surface currents are not negligible compared to the low winds observed during COARE, and Weller will approach Charlie Erikson to prepare surface current maps for our use in time for Toulouse. Wind direction measurements among the platforms need to be compared more accurately, and will be done by Bradley.

Steve Anderson reported on the steps taken to correct the WHOI mooring meteorological time series for instrument errors noted in the intercomparison. The unaspirated VAWR air temperature suffered from radiative heating. The November 28 intercomparison period showed the highest error, approximately 1.5C in the afternoon. The aspirated IMET air temperature sensor on the buoy worked for several weeks and provides data to verify the correction algorithm. Specific humidity is being recomputed using the corrected air temperature. Anderson examined the sensitivity of bulk fluxes to errors in the observables. The chart he showed will be circulated separately.

Meghan Cronin reported on the Hakuho-maru and PROTEUS buoy data intercomparisons at 0N/156E. The intercomparison was for a 3-week period in November 1992 when the ship was within 5-10 km of the buoy. Adjustments for the different sensor heights were made. On average the ship winds were 0.3 ms-1 lower than the buoy winds; ship air temperatures were 0.4C warmer than the buoy even at night; ship specific humidities were 1 g kg-1 higher than the buoy; and ship SST was 0.2C warmer than the buoy SST. These differences, and particularly those of specific humidity, caused the buoy bulk latent heat flux to be 22 Wm-2 higher than that on the ship. Since, in addition, the ship turbulent latent heat flux was 12 Wm-2 lower than its bulk estimate, the buoy bulk flux was 34 W m-2 higher than the ship's direct measurement. Various explanations for the differences in observables and estimated fluxes are being investigated. Cronin also reminded the group that there were approximately 15 other ATLAS and PROTEUS buoys within the larger COARE domain and that, if at all possible, these buoys should be included in the intercomparison studies.

Chris Fairall reviewed the Version 1.0 COARE bulk flux algorithm. Based on comparison with the Moana Wave experiment-average turbulent latent heat fluxes, the transfer coefficients have been reduced by 15% from those in the original LKB model. This factor is supported by the Franklin data which have been processed so far. Version 1.1 will include parameterization of the cool skin developed by Fairall, Bradley and Godfrey. This parameterization is supported by the data of Coppin et al (1990), on the dependence of the skin-bulk temperature difference on wind speed, but disagrees with the earlier data of Schluessel et al (1987) (see comments by Gary Wick below). Version 1.1 is anticipated in mid - April (later it was decided to include a switch to disable cool skin). Work is also in progress on ways to adjust non-surface SST observations for the near-surface temperature gradient associated with diurnal heating by penetrating solar insolation. The surface temperature can be up to 3C warmer than that at 2 m depth during the early afternoon. However, Fairall is not yet ready to include a warm layer routine in the algorithm, because:

a) the warm layer is really an integral over fluxes since sun-up, not a pure function of the current conditions;

b) it is not clear if the total delta-T should be applied linearly or as a step function - Stuart Godfrey has sent some discussion about this;

c) he is not convinced that the current model has been verified with the Moana Wave data - there remains a question about the sampling of his floating sensor on hot , sunny days.

We need actual temperature profile measurements in the ocean (e.g. from Steve Anderson) to check it. Such a routine should be a goal for Toulouse, but a) particularly is a stumbling block.

Gary Wick also reported on efforts to parameterize the cool skin. A recent paper by Schluessel et al. (1990) concluded that the skin-bulk temperature difference increases with higher wind speed. However, Wick's data from the ships Baldwin and Malcolm Baldridge showed the opposite, and his re-examination of Schluessel's data showed that the increase with wind was confined to only some sections of the data, while the rest showed a slight decrease. A Mellor-Yamada (1982) 2nd order turbulence closure model is being used to get detailed vertical profiles of ocean temperature. A surface renewal model and skin-bulk temperature difference parameterizations were developed starting with the work of Saunders (1967) and compared with more recent work of Soloviev and Schluessel (1994) and Schluessel et al. (1990). Emery and Wick obtained data from the Vickers during the Central Equatorial Pacific Experiment (CEPEX), which followed immediately after COARE. The goal of their effort is to develop a way to produce time series of bulk-skin temperature difference.

A general discussion of the accuracy of radiative flux measurements followed. Fairall feels that the intercomparison work has justified an accuracy of 1% for shortwave and 2-5 Wm-2 for longwave. Piotr Flatau, visiting the meeting from SIO, suggested that the uncertainty in longwave measurements was closer to 10 Wm-2. Vickers had a spectral radiometer, and with this data the radiation transfer model results (such as the estimates provided by Cox and Ackerman for the November 28 comparison) could be checked. The Cessna had calibrated radiometers. The C-130 had a different design of longwave radiometer that does not suffer from the dome heating error of the Eppleys. Hignett cautioned that they have seen problems with recently purchased silicon domes for the Eppley sensors, with leakage of up to 20-30 W m-2 of shortwave radiation. The effect of clouds on the radiation budget was discussed, and the question raised as to whether or not radiometer calibrations were a function of cloud cover. The recent shortwave radiometer calibrations in Canberra indicated that three different types of cloud cover (clear sky, diffuse, overcast) produced slight differences in calibration factor.

*Action items arising from discussion included:

1) the need to look for more information about the ocean surface currents and their spatial and temporal variability in order to correct winds relative to the sea surface via a) broad- scale ocean data (Weller), b) comparison of IMET 5m, 10m and 20m currents (Anderson/Weller) and c) examination of large-scale coherence of surface current using the Wecoma ADCP and TOGA TAO (Paulson, McPhaden);

2) the need to complete intercomparison of all available data with the Proteus and Atlas buoys at the equator and at 2S along 156E (Cronin, Anderson);

3) the inclusion of the cool skin effect in the bulk algorithms, resulting in Version 1.1 in the near future (Fairall);

4) the need to intercompare wind directions, including checks on whether the same values for magnetic deviation are being applied to instruments with compasses (Bradley. Mike McPhaden will post the values used at PMEL across the TAO array);

5) need for a subroutine for solar albedo based on the Payne tables, to include a sun angle calculator (Coppin);

6) follow up estimates of uncertainty in SW and LW radiation measurements using CEPEX experience (Rogers, Flatau);

7) look into absorption profiles for shortwave radiation, especially red-light absorption (Bradley);

8) investigate the value for longwave emissivity of the sea surface, currently taken to be 0.97 (Paulson);

9) the need for a standard routine based on the the COARE v 1.0 algorithm to adjust wind speed, air temperature and humidity measurements to specified heights (Cronin).

Aircraft Reports

Friehe discussed the data from the 2 P3's. GPS has been used to correct the inertial navigation system (INS) for errors including the Schuler oscillation. Data were collected at 1 Hz rate all the time and at 40 Hz on demand. An offset in dewpoint, resulting in an offset of 0.5 g kg-1 was noted; work remains to be done to explain biases in dewpoint (0.5C) and air temperature (0.33C). SST was obtained with PRT-5 radiometers. Data from the two P3's are being compared to the Electra for the period that the three planes flew over the WHOI surface mooring. The wind speeds from the planes differ by approximately 0.75 ms-1. As a preliminary to the processing of the turbulent data, the ability to correct vertical velocity for platform motion has been checked. The Lyman-alpha was calibrated against the General Eastern chilled mirror dewpointer. Turbulence processing is being done using the Ogive technique ((Friehe et al, 1991 give a good description of this method) to verify that the time series was long enough to include all frequencies contributing to the covariance. Comparisons between the two P3 aircraft on November 28 show differences of 30-40 Wm-2 and a mean value of approximately 75 Wm-2 in latent heat flux (the Electra gave about 65 Wm-2). The differences in sensible heat flux were 1-2 Wm-2 , and the mean was 6 Wm-2. Stress did not agree so well, which is not unexpected considering the low winds on the 28th. At this time all the 1 Hz data from the P3's have been worked up, and plots will be made available. The 40 Hz data are is now being worked on.

David Rogers discussed the data collected on the Electra. Bulk and eddy fluxes have been worked up for selected days when intercomparisons were possible (Nov 16, Nov 26, Nov 28, Dec 2, Dec 3, Dec 5, Jan 16). Comparison of these shows differences that are attributed to uncertainties in the bulk observables. The 10 m value for Ce was estimated to be 1.2 x 10- 3. Cd estimates showed more scatter. Rogers questioned how the aircraft investigators should proceed to reconcile bulk and turbulent flux results. Fairall's approach in developing the COARE Version 1.0 bulk flux algorithms was to refine the bulk formulae assuming the Moana Wave turbulent fluxes were correct. Rogers noted that eventually the aircraft fluxes could be treated the same way, but that present problems with the mean aircraft observables make such analysis premature. Provided spatial variability is not too severe, the aircraft data have the potential of tying many of the platforms together as the boundary layer flights repeatedly overflew Moana Wave, Franklin, the WHOI mooring, and other platforms. During these overflights wind speeds ranged from 0.5 m s-1 to 7 m s-1.

Alastair Williams reported on the status of the data collected by the Cessna. The Cessna used a combined GPS/INS navigation system and had 95% data recovery rate for its 18 missions. First version processing was completed late last year. Intercomparison with other platforms point out the need to reprocess the data to correct some disagreement. A second version of the data will be ready by the Toulouse meeting. Intercomparisons with Franklin, Electra, and Malaita are being looked at. Flights over Kavieng provide comparison of aircraft and ground-based radiation measurements. Williams noted one unique aspect of the Cessna's operation; because it was economical for the Cessna to transit at low levels, data were collected during long (300 - 400 km) flights at 200 foot elevation between Rabaul and the Intensive Flux Array (IFA). The methodology for processing the Cessna data are being examined. Of particular interest is how to separate turbulent scale fluxes from mesoscale influences. Turbulent scales go up to 2-3 km, and Williams suggested that long runs be broken up into shorter (10 - 15 km) runs. More discussion about mesoscale variability in the fluxes followed. Fronts were common during COARE and provide one example of such variability.

Alan Grant spoke about the data collected on the UK C- 130. They just obtained the P3-42 data for three 20-minute intercomparison flights. Differences in potential temperature were 0.32C, winds differed by 0.25 ms-1, dewpoint by -0.75C, and SST by 0.13C for one intercomparison. The air temperature and SST biases varied among the intercomparisons, but the dewpoint bias was repeatable and requires investigation. SST errors are being examined. A humidity correction for wind speed has not yet been applied. Fluxes from the C-130 have not yet been compared with other platforms. The C-130 eddy covariance and bulk stress estimates indicate neutral 10 m Cd of 1.18 x 10-3, Ce of approximately 6.8 x 10-4, but adjustments to the dewpoint could increase this to 7.4 x 10 -4. The data are processed taking 30 km long legs, detrended. Cospectra indicate that such legs include all the signal.

*Action items (added in press by EFB on the suggestion of Carl Friehe, who is prepared to take on item 1); I suggest 2) is a prelude to the series of intercomparison papers listed below, and should be tackled by the co-ordinators plus Dave Rogers):

1) aircraft investigators to compare and resolve the generally small offsets in mean variables, e.g. SST, air and dew point temperatures (Friehe).

2) compile a complete listing of all possible intercomparison periods for aircraft- aircraft, aircraft-ship, ship-ship, ship-buoy (Bradley, Coppin, Anderson, Cronin, Khelif, Rogers)

Ship reports

Chris Fairall worked with George Young and Jim Edson to collect turbulent fluxes from Moana Wave. The data were processed in real time. Mean observables have been adjusted on the basis of intercomparison with other ships and the WHOI mooring. The basic averaging interval for fluxes is 10 minutes without detrending. 50-minute summary data sets have been produced by averaging these. At present Fairall has no plans to reprocess the data. There was discussion about this method of processing as opposed to the Ogive technique more commonly used by the aircraft investigators. One concern was whether or not in low wind speed the averaging of five 10-minute blocks resolved all the covariance. Ship/aircraft spectra comparisons would be desirable. At high wind speeds, covariance fluxes were slightly lower than inertial dissipation values, but inertial estimates compare well with covariance at wind speeds down to 1 - 2 ms-1.

Peter Coppin reported on progress with the data collected on the Franklin. Both a boom off the bow and the foremast were instrumented with sonic anemometers and fast-response hygrometers. Selected sections of the data have now been processed. Turbulent flux data were acquired at 20 Hz. Six centrally located and 3 additional accelerometers on the bow provide data for motion correction. An Ashtec differential GPS was also on the ship, but did not always perform well. Flow distortion studies have been done since COARE using a model of Franklin in a wind tunnel. These indicate a 5% decrease in wind speed at the bow mast for relative winds between +/- 20 deg. off the bow; the distortion was less for larger angles. More tests with a smaller model that can be rotated in the wind tunnel are planned to examine flow distortion effects on turbulence. Data from the Lyman-alpha instruments have been compared with data from the infrared hygrometers produced by OPHIR, Oak Ridge National Laboratory and the Japanese E009. The OPHIR was an very reliable instrument, and humidity spectra from the Oak Ridge instrument were extremely clean.

Covariance and inertial dissipation fluxes agree well, one indication that the motion correction to the Franklin data is correct. Coppin raised the issue of whether the group should consider adopting and documenting standard methodology for computation of fluxes by the inertial dissipation technique, and there was agreement to revisit this topic later. Cessna data broken into 10 km segments were compared with Franklin 15-minute average fluxes. Overall they agreed well, but scatter was considerable as expected. An important finding was that scatter was the same for segments adjacent to Franklin as it was for segments when the two platforms were well separated. Comparisons are now being done with Electra data.

Osamu Tsukamoto discussed the status of the turbulent flux measurements made on the Hakuho-maru. The ship was at the equator 11-26 November, but was in the IFA from 27- 28 November. An eddy correlation system was located on the foremast at 16 m with sonic anemometer, IR hygrometer, fine wire wet and dry bulb temperature sensors. The IR hygrometer was a new system and its calibration is being examined by comparison with a slower response hygrometer. Tsukamoto inquired how many other platforms measured air temperature via sound speed and if other investigators used the corrections for wind speed and humidity suggested by Kaimal. Corrections are made for the ship's motion. Turbulent flux data were collected at 10 Hz; ship motion data were collected at 2 Hz. Processing to do the motion correction has resulted in a combined 2 Hz data set. Eddy correlation data were collected over 60-minute intervals several times a day, processed as 30, 20, or 15-minute segments. Comparison of eddy and bulk fluxes suggest Ch = 1.32 x 10-3 and Ce = 1.162 x 10-3. The average sensible heat flux calculated with the Brook (1978) correction was 14 Wm-2; that without was 5 Wm-2. The validity of this correction was discussed; the general consensus of the group (except Tsukamoto) was that it should not be applied (see Businger, 1982). The average latent heat flux was 88 Wm-2 and the net radiative flux 145 Wm-2. Thus, the average net heat flux observed by the Hakuho-maru was 52 Wm-2. However, the Nov. 28 intercomparison showed that the Hakuho shortwave radiometer underestimated by about 12% , which would increase this value slightly.

Carl Friehe noted that both he and Frank Bradley are anticipating visits from investigators that worked on the Chinese ships collecting turbulent flux data, and it is hoped that at least two Chinese ships will contribute fluxes estimated by inertial dissipation techniques. (Later discussion revealed that both groups appear to be hosting the same Chinese scientist, Xilong Song, simultaneously. An approach from TCIPO to Dunxin Hu has resolved the situation; Song will spend a year at UCI).

Bob McMillen discussed the status of the data collected on Malaita, which was equipped with eddy correlation sensors, gimballed radiometers, and an optical rain gauge. Malaita collected data in the IFA at 2S, 154E from February 9 - 22, 1993. Because ship motion was extremely large, correction of the eddy flux data is proving difficult. Differences between eddy and dissipation estimates result. The data collected in the IFA in February gave inertial dissipation estimates of averaged sensible and latent heat fluxes of 14 and 76 W m-2 while the eddy estimates were 24 and 143 Wm-2. The difficulties of correction for platform motion are being examined, and also some problems with air temperature and SST measurements. Two overflights by the Cessna will provide data for comparison with the Malaita.

Turbulent flux working discussion

*Action items (with responsible person in parentheses):

1) establish location on TCIPO storage (Richard Chinman);

2) post 20 minutes of aircraft data and resulting fluxes (Carl Friehe);

3) post ship data and fluxes (Peter Coppin);

4) post file of mean observables and fluxes (Frank Bradley).

The need to standardize averaging intervals to permit intercomparison was identified. Some of the issues included: selection of averaging intervals to include homogeneous conditions and exclude transient events like squalls; should the full length record be linearly detrended or broken down into pieces; what standard flux products should be produced and circulated among the groups during the intercomparisons - ogive plots, statistics including moments up to and including 4th order, structure function parameters, cospectra.

*Action items:

1) to facilitate intercomparisons a chart of when the planes flew during COARE would be distributed (Richard Chinman);

2) time and position files for each platform would be created and posted to the /incoming directory on tcdm.coare.ucar.edu with latitude and longitude given in decimal degrees to 4 decimal places (all); [once notified via e-mail, TCIPO will then place these files in the appropriate directory for anonymous access (/pub/COARE_DATA/cross_group_data/platform_pos )]

3) aircraft position data files would be posted with 1 minute time resolution to keep the size of the files reasonable, and include track and height (all aircraft investigators);

4) data files would be ASCII and tab-delimited (all);

5) time information in data files would follow the YYMMDDHHMMSS.SSSS convention agreed upon in Boulder in September (all).

It was recognized that averaging issues would need to be revisited during the scientific analyses that would follow the intercomparison effort. In this case, the particular scientific focus could either determine or depend on the averaging scales used in the processing. Dick Johnson and Denise Hagan gave evidence of convective cells in the atmosphere with spatial scales of 30 km. We were reminded of the U. of Wisconsin video of convective development over the IFA which could be helpful in selecting cases.

Corrections to remove platform motion from the eddy covariance data were discussed. There is a need to document the algorithms, to be explicit about the reference frame used for flux calculation, and to consider flow distortion correction related to platform attitude and angles of attack. It was agreed that fluxes exchanged for intercomparison in preparation for Toulouse would be in kinematic units, mKs-1 for sensible heat and gm-2s-1 for water vapour. It was also agreed that vector fluxes would be computed in both wind (along, across) and geographic (north- south, east-west) coordinate systems to facilitate comparison with aircraft, ship, and buoys. Further, aircraft data files should indicate the aircraft track relative to the wind direction; geographic surveys were common because winds were often light and variable in direction.

*Action item:

1) specialists to discuss motion correction schemes, instrument (mis)alignment, INS performance and report (Friehe, Fairall, Coppin).

COARE inertial dissipation flux algorithms will be developed to standardize and permit comparison of dissipation fluxes. Inertial dissipation estimates will be included in flux intercomparisons whenever available.

*Action item:

1) an algorithm will be provided to calculate dissipation flux estimates using the structure function parameters CT2, Cq2, Cu2, Cv2, Cw2 (Fairall, Coppin).

Other items were raised which may require attention. The need to consider depth of the surface layer during intercomparison of aircraft fluxes with surface platforms - the vertical stacks by the planes can be used to examine flux divergence and the vertical structure of the boundary layer. Concern was raised about salt contamination on turbulence temperature probes - Friehe noted that no salt was seen on the P3 and Electra 150 micron diameter thermistor beads during occasional inspections. The thermistors are contained in a Rosemount housing which may have helped separate out the salt aerosols. The overall salt aerosol loading was also light. Hagan, however, saw some salt deposition on the window of her radiometer, and Bradley commented that salt contamination is evident on fine wires and psychrometer wicks even on very calm spray-free days.

Restictions on the applicability of Monin-Obukhov similarity in the strongly convective atmospheric boundary layer was discussed, and particularly the depth of the constant flux layer. It was agreed that looking for similarity in cospectra and other tests would be a good topic for the group in Toulouse. An overall goal for the flux group that arose from the discussion was to make available to COARE investigators a time series of the fluxes representative of the IFA, with documentation of the methodology used to compute the fluxes, and with quantification of the error bars associated with the fluxes.

It was concluded that every effort should be made to push the intercomparisons to completion prior to Toulouse. A series of intercomparison papers were identified to provide foci for this effort - they are listed below along with the investigators participating in their preparation. The investigator(s) who will coordinate and oversee the work is identified with an asterisk:

1) Aircraft-aircraft: Friehe* (and U. C. Irvine (UCI) team), Grant, Hagan, Hignett, McMillen, Rogers, Williams.

2) Ship-ship: Bradley*, Coppin, deWitt, Fairall, Fujitani, Ishida, Paulson,Tsukamoto.

3) Aircraft-ship: Coppin*, Bradley, Chinese ship investigators, deWitt, McMillen, Paulson, Rogers, UCI, Tsukamoto, Williams.

4) Buoy-ship: Anderson*, Bradley, deWitt, Cronin*, Fairall, Fujitani, McPhaden, Paulson,Tsukamoto, Weller.

5) Aircraft-buoy: Anderson*, Cronin, Hagan, Khelif*, McPhaden, Rogers, Williams.

Large-scale flux estimates and space/time variability

Dick Johnson discussed the status of the data from the large scale sounding array and illustrated results to date. The IFA-averaged 0000UTC winds showed the westerly wind burst in late December to early January. At the top of the troposphere strong vertical shear and downward-propagating Kelvin waves were evident. Vertical motion inferred from the time-varying pressure signal showed deep convection occurring some days ahead of the westerly burst - satellite images also show an explosion of deep convection around December 12. Heat and moisture budgets over the IFA have been evaluated using the 0000 UTC soundings to look at surface precipitation, evaporation, and heat flux. Sensible and latent heat fluxes using the lowest data points from the two Chinese ships, via the COARE v1 algorithm, are in remarkably good agreement. Rain rate has been estimated as a residual of the moisture budget. In general, satellite estimates of rainfall are large compared to the moisture budget, e.g. on December 14, a day characterized by substantial precipitation. An average rainfall rate of 10 mm day-1 over the IFA was inferred compared with the average of 11 mm day-1 measured on Moana Wave. Considering that a latent heat flux of 100 W m- 2 corresponds to an average evaporation rate of only 3 mm day-1, considerable advection of moisture is implied.

Guosheng Liu discussed satellite-based values for precipitation. He contrasted the use of Special Sensor Microwave/Imager (SSMI) and infrared satellite sensors for rainfall estimation and the coverage available over the COARE domain with these sensors. He works with the SSMI data which provides roughly twice per day coverage falling within the time slots: 6- 7, 10-11, 18-19, or 22-23 UTC. The SSMI and GPI (IR-derived) data show disagreement. Liu has also compared SSMI rainfall with Moana Wave rain gauge data. He indicated that a method that combines SSMI and IR satellite data may provide better estimates of rain rate.

Frank Bradley presented results from the MIT ship-based radar forwarded to the meeting by Steve Rutledge. Radar reflectivity maps had been prepared for a period of intercomparison with Moana Wave in late January. The radar data have been binned into 2 km x 2 km regions at 1 km above the sea surface. The reflectivity maps for the comparison show frontal lines and, often, indicate wide-spread rain. The radar rain rate estimates at the site of the WHOI mooring were, however, roughly a factor of ten smaller than that observed at Moana Wave nearby. Discussion failed to advance a plausible reason for such a large discrepancy. There is some doubt that rain was as wide-spread as suggested by the radar, but an overestimation of the area in which rain was falling would not resolve the matter, which will be referred to Steve Rutledge in due course.

Clayton Paulson showed a comparison of rain data from the TAO array collected by Mike McPhaden with satellite data (GPI). He noted, however, that calibration work on the optical rain gauges (ORG) continues, with early work suggesting that they overestimate rain rate by 20-30%. Wecoma had optical, siphon, and traditional (manual observation of volumetric gauge) rain gauges as well as instrumentation to measure the salinity at a depth of a few cm in the ocean and the temperature and salinity at a number of shallow ocean depths. The rain observed on Wecoma by the traditional gauge averaged to 10.5 mm day-1; the siphon gauge agreed well but the two ORGs were 20% and 30% high.

Rain falling on the sea surface was cooler than the ocean. The rain temperature was close to the wet bulb temperature, though on several occasions Wecoma recorded rain temperature colder than the wet bulb by up to 1C. Rainstorms cause temperature and salinity (T/S) variability in the warm pool, but advection and surface heating are other sources of variability. The bow sensors on Wecoma show very shallow, fresh and cool pools of water associated with the rain. Many such observations were made coincident with rainfall, but at least one fresh pool was observed without coincident rain. There was a discussion of the persistence of fresh pools caused by rain. Paulson said that earlier work indicated the time- scale to erode a freshwater lens was approximately 11 hours (Bahr and Paulson, 1991). Further work will be done on the Wecoma rain data, including statistics and the inference of rain rate as a residual of the upper ocean T/S budget.

Bradley reminded the group that there were all sky cameras on Malaita, Franklin, and Wecoma, recording cloud images every half-hour throughout the COARE IOP. The images have been processed manually at Sydney University of Technology, resulting in data for cloud amount and type in equal solid angle sectors of the sky. Bradley asked if the TCIPO should consider supporting the archiving of the cloud imagery to CDROM.

*Action items:

1) identify other sources of rainfall data not yet being brought into the intercomparison studies e.g. the two acoustic instruments deployed by David Farmer, a WOTAN (Weather Observation Through Ambient Noise) on the WHOI surface mooring and an instrument called ELSI on Farmer's mooring. The status of this data set will be checked and attempts made to bring it into the rain rate comparison studies (Weller).

2) Other COARE PI's will be consulted about the usefulness of the cloud imagery (Chinman)

3) If this produces a positive response, a memorandum will be drafted to TCIPO indicating the usefulness of the cloud imagery and asking for TCIPO support to make it readily available to all on CDROM (Bradley).

SST observations and variability

Ed Walsh compared observations of mean square surface slope made on November 28 with a scanning radar on one of the P3 aircraft flying together with the Electra carrying Hagan's radiometer. Smoother sea surface correlates well with higher SST. In some data, regions up to 8 km in size are observed to be very smooth. A discussion of the cause of the correlated surface-wave/SST variability followed. One possibility is internal waves. Walsh discussed this later with Paulson and Pinkel and will seek collaboration with the ocean mixing group.

Gary Wick discussed the analyses of High Resolution Picture Transmission (HRPT) data collected from NOAA 11 and 12 principally at Townsville with some other data coming from Guam. 4 km resolution maps of brightness and reflectance are available roughly 5 times a day. SST has been determined from this data using the operational NOAA procedures. Two-week composite maps of 4 km MCSST (buoy data having been used to generate the MCSST coefficients) have been produced from NOAA 11 data. The resulting night time images from October 1992 to March 1993 are available from Emery's group at Colorado via anonymous ftp. This group is presently working on 10-day and monthly composite maps of 4km skin SST from nighttime passes of NOAA 11 and 12. They are also working up a 3-year (1990-1992) climatology of SST for the COARE region and high resolution maps will be prepared. One product will be 10-day and monthly skin SST with 1 km resolution. Nighttime images will be composites of NOAA 11 and 12. Daytime composites will use only NOAA 12 data. Processing algorithms will be verified and corrected using all available ship and aircraft radiometer data. The second product will be 10-day merged maps of SST from satellite data and from ship and aircraft radiometer data. If there is sufficient aircraft data, maps will also be produced from that data set alone. Finally, they will compare the satellite skin temperature measurements with in-situ bulk SST measurements. Using skin temperature parameterization and their numerical mixed layer model they will attempt to extrapolate the bulk measurements to the surface and thus provide additional data points to quality check the SST products.

The CU group will be busy until Toulouse doing the navigation of the satellite data. They will bring 4km data to Toulouse. In addition, they can bring a selection of 1 km data.

*Action item:

1) identify times for intercomparison of satellite, aircraft, ship, and buoy SST data, such as November 28 (all).

Carol Anne Clayson discussed the project to produce fluxes from satellite data. A trial was done, comparing satellite fluxes with data collected during Chris Fairall's TOGA-COARE pilot cruise. Data from the International Satellite Cloud Climatology Project (ISCCP) provided cloud climatologies, radiative fluxes, and SST. SSMI data were available roughly every day with 25-30 km resolution. SST is derived from brightness temperatures using a radiative transfer model and compared with in-situ data. There are fewer surface observations to compare with satellite inferred surface shortwave flux. Wind speed comes from SSMI using the Wentz (1986) algorithm; this product is widely used. Surface air temperature and humidity are determined from TIROS Operational Vertical Sounder (TOVS) profiler data (which, for example, provides a measure of column integrated water vapor) using the algorithm of Miller and Katsaros(1992).

The satellite latent heat flux had a bias of 4 Wm-2 compared to the shipboard data collected in the COARE pilot cruise; however, it also had a standard deviation of 50 Wm- 2. The average total outgoing heat flux (sensible, latent, longwave) from the satellite differed from that of the ship by 10 W m-2. The error in the incoming shortwave was approximately 30 Wm-2.

The goal for Toulouse is to produce satellite-derived heat fluxes, wind stress, and freshwater flux on 25-30 km resolution, every 3 hours, for the region between 20N and 20S. The maps would be at 0, 3, 6, 9, 12, 15, 18, 21 hours UTC. Carol Anne will be trying to find intercomparison data for satellite, aircraft, ships and buoys to validate the bulk parameters and derived fluxes.

Planning for Toulouse

Clayton Paulson and Lynn deWitt are interested in working on air-sea flux comparisons with an emphasis on rainfall and on the effect of rain on the upper ocean. They will also work with the Ocean Mixing group. They will bring a laptop PC but will need network connections, a colour monitor (VGA) and access to printers capable of producing transparencies. They plan to arrive by July 30.

The UCI investigators, Friehe, Burns, Khelif and Chinese visitor (Xilong Song), will work on intercomparison papers and on studies of boundary layer structure and surface flux variability. The UCI group can arrive by July 30 with the P-3 data sets; earlier thoughts about sending their own computer systems to Toulouse may be reversed because of possible difficulties and delays with customs.

Alan Grant and Phil Hignett indicated that one of them will probably attend the workshop and focus on developing the COARE flux algorithm for use in computing fluxes from aircraft data. They plan on bringing fluxes calculated from the C-130 data.

Gary Wick (but not Bill Emery) will attend, bringing the 4 km satellite SST data and select periods of 1 km data for intercomparison. He will arrive by July 30.

Bob McMillen and Tim Crawford will attend and work at linking the Malaita measurements to other flux data, such as that from the Cessna. They will bring a PC but will need connection to the local network and a printer.

Guosheng Liu indicated that the work on precipitation by different types of cloud using satellite data will continue.

Alastair Williams and Jorg Hacker will attend to work on the Cessna data. They may bring the data on DAT tapes and will need a means to read the data as well as computing facilities. Their programs produce graphics using Calcomp plotting calls and need that software supported in Toulouse. The science issues to be pursued include boundary layer turbulence and parameterization, surface flux variability, and links between the fluxes and the mesoscale structures. The Cessna group will arrive by July 30.

Peter Coppin is interested in contrasting temporal and spatial scales of variability in the fluxes and in the small-scale air-sea energy budget focusing on the Lagrangian drifter studies carried out by Franklin.

Frank Bradley is interested in heat budgets, spatial structures in the fluxes, and on the cool skin and warm layer of the upper ocean. Bradley and Coppin will bring laptop PCs but need network connections, colour monitors (VGA), and colour printers capable of producing transparencies. Both will arrive by July 30.

Due to conflicts with other experiments it appears unlikely that Tsukamoto, Ishida or Fujitani will attend the Toulouse meeting. They will be sending some data with Cronin and McPhaden.

Meghan Cronin indicated that she, Mike McPhaden, and Nancy Soreide will attend. They will bring the TAO data and install the TAO display software on the workstations at Toulouse. They require display hardware and software, NetCDF, plot+ and MATLAB. They will continue work on rainfall and air-sea fluxes and seek to collaborate with the ocean mixing group.

Paul Spyers-Duran, who is from the Electra group at NCAR, will attend the Toulouse workshop. Friehe will make sure that NOAA AOC (P3's) are aware of the Toulouse workshop as well, although since it is the hurricane season they may not be available to attend.

Denise Hagan will continue work on variability in SST as observed by the Electra radiometer and on the effect of humidity structure on the radiometric data. She plans on arriving early.

David Rogers and Yolande Serra will be at Toulouse working on the spatial variability of the fluxes, linking the aircraft fluxes with the ship fluxes, and looking at convective structures. They plan to work with Frank Marks and Dave Jorgenson. They anticipate needing MATLAB, Fortran, and internet access.

Carol Anne Clayson said that she and Judy Curry would continue work on satellite- derived air-sea fluxes, working together with the ocean mixing and large-scale atmospheric circulation groups. They would like to see a common display format in use in Toulouse that could handle satellite data as well as other fields such as their fluxes. They will have a representative at the meeting by July 30 and will bring the flux data on tapes.

Ed Walsh will continue working with the mean square slope and sea state data collected from 8 flights of the NOAA aircraft up to and including December 1992. He is interested in collaboration in Toulouse with ocean mixing investigators who can work with him to explain the spatial patterns he observed in surface roughness.

Chris Fairall's interests include bulk, inertial dissipation, and eddy fluxes, the warm layer and cool skin, and issues associated with averaging fluxes on larger scales. He has not yet worked much with wind profiler and cloud liquid water vapor collected during COARE, but this may be something to be pursued jointly with Steve Rutledge. He will require a Magneto-optical disk drive which is PC/Unix compatible. Fairall noted that George Young is planning to attend the Toulouse workshop, examining mesoscale variability in the fluxes, developing an average surface signature for convective events, and looking at the recovery of the boundary layer after the passage of storms.

Bob Weller and Steve Anderson will arrive by July 30 and bring the surface mooring data. They are interested in intercomparison of the bulk fluxes from the WHOI surface mooring with fluxes from other platforms, in the temporal and spatial variability of the fluxes, and in the oceanic response to the fluxes. They plan on also working with the ocean mixing group. They will bring Macintosh laptops and need an Appletalk to Enet connection, color monitors, and access to printers.

Following these brief statements, the group discussed the need to keep in contact with other investigators prior to Toulouse. McIntosh and Carswell, who fielded a scatterometer, LeMone, Grossman, Lagerloef, Schlussel, the flux investigators on the Chinese ships, numerical weather prediction modelers at BMRC, NMC, and ECMWF, John Young, and Ian Barton were among those mentioned. Jim Edson has turbulence data from the Wecoma and Moana Wave - it is not known what his plans are.

The next agenda item was the discussion and development of a format for the Toulouse workshop. The group all favoured getting to Toulouse early, on the weekend prior to the August 2 formal start, on July 30 or 31. On the weekend computers would be set up and work begin. Monday morning August 1 there would be a flux group meeting. The results of the intercomparison and algorithm development work being done in preparation for the workshop would be presented and discussed and work on the scientific topics summarized below would begin. At the Tuesday plenary sessions bringing all the COARE investigators together, Bradley and Weller would report on behalf of the flux group. Thereafter, the group favoured a cycle in which there were plenary meetings only on the afternoons of Wednesday 3rd, Friday 5th, Monday 8th and Wednesday 10th. On the days in between plenaries, the flux group would hold their own meeting at 11 am. This would allow for two 1.5-day uninterrupted working sessions. (Our scheme has since been adopted by the workshop convenors).

It was felt strongly that there should be a central area where posters could be displayed, and that there should be a COARE-wide poster session every day, perhaps just prior to dinner, to bring people together informally so that collaborations would spring up. The flux group identified the requirement for a central flux working area or classroom. The room should have space for posters and plots to be pinned or taping up to display and share results. There should also be network connections, printers (including capability to make colour transparencies) and good colour monitors to hook to laptops. Several members of the group reported that they do not use Unix-type work stations, and that hook-ups to laptop PC's and Mac's will be required.

Chris Fairall's experience has been that computers shipped to France can take one month to clear through customs. Presumably lap-tops brought in as luggage are OK.

*Action items:

1) TCIPO to survey PC, Mac, network and Monitor needs.

2) TCIPO to investigate customs aspects of shipping computers to Toulouse.

Scientific foci (with coordinator in parentheses) for the flux group's activities at Toulouse were identified as:

1) Space/time scales of freshwater flux, encompassing buoy, ship, radar, satellite, and possibly, aircraft data (Paulson).

*Action item: check on status of aircraft PMS and other data relevant to freshwater flux (Friehe).

2) Spatial flux fields and flux divergences, using data from 2 P3s, Electra, buoys, ships (Rogers and Friehe).

3) Modulation of fluxes by mesoscale atmospheric variability (Williams).

4) Representivity of single point flux assessments (Coppin, Anderson).

5) Space/time variability of SST, cool skin, warm layer (Fairall).

6) Surface layer similarity theory (Friehe).

7) Short term response of boundary-layer and fluxes to storm events (Bradley).

An important issue that will be visited regularly during the workshop is the accuracy of the fluxes. The goal is to develop a quantitative statement of the accuracy of the observational methodologies used in COARE, of the COARE flux algorithms, and their physical correctness (do they properly represent the physics of air-sea exchange?), and of the flux products which will be made available to other investigators. To address this goal, the group will begin now to compare the ship-based flux algorithms with aircraft flux results. This requires that the aircraft investigators resolve the remaining measurement problems, compute turbulent fluxes, and compare them with bulk fluxes computed with the COARE Version 1 algorithms.

Polling of the group and discussion of how best to work together made it clear that very diverse software and hardware will be needed in Toulouse. To facilitate preparation for Toulouse it was deemed essential that TCIPO soon establish in Boulder an identical twin of the hardware/software workstation installation to be used in Toulouse Log-on accounts and disk space should be provided so that the flux investigators can begin now to install and debug the software they need operational in Toulouse. In addition, investigators could begin now to load TCIPO disks with the data needed in Toulouse. With this trial workstation available, there is good chance that work by the flux group can begin, as planned, on the weekend of July 30, 31.

*Action item:

1) Richard Chinman to carry this request to TCIPO.

References

Alados-Arboledas, L., J. Vida and J.I. Jimenez (1988): JTech, 5, 666-670
Bahr, F. and C.A. Paulson (1991): TOGA Notes, No. 3, 9-13
Brook, R.R. (1978): Boundary-layer Met., 15, 481-487
Businger, J.A. (1982): JAS, 39, 1889-1892
Coppin, P.A., E.F. Bradley, I.J. Barton and J.S. Godfrey (1991): JGR, 96 Supp., 3401- 3409
Friehe, C.A. et al. (1991): JGR, 96 No. C5, 8593-8609
Kudrayavtsev, V.N. and A.V. Soloviev (1990): JPO, 20, 617-628
Mellor, G.L. and T. Yamada (1974): JAS, 31, 1791-1806
Mellor, G.L. and T. Yamada (1982): Rev. Geophys. and Space Physics, 20, 851-875
Saunders, P.M. (1967): JAS, 24, 269-273
Schluessel, P., H-Y Shin, W.J. Emery and H. Grassl (1987): JGR, 92, 2859-2874
Schluessel, P., W.J. Emery, H. Grassl and T. Mammen (1990): JGR, 95, 13341- 13356
Soloviev, A. and P. Schluessel (1994): JGR-Oceans, in press.
Wentz, F.J., L.A. Mattox and S. Peteherych (1986): JGR, 91(C2), 2289-2307

(Bob Weller and Frank Bradley - 22 April 1994)

 

http://www.ncdc.noaa.gov/oa/coare/meetings/Turb_LS_flux.3.94.html
Downloaded Thursday, 08-Jan-2009 19:58:13 EST
Last Updated Tuesday, 31-May-2005 12:25:39 EDT by David.P.Smith@noaa.gov
Please see the NCDC Contact Page if you have questions or comments.