This is a summary of the WOCE UOT/DAC meeting held November 1990 in Carmel, CA, USA. In order for the UOT/DAC to begin operations in January 1991, preparations need to be undertaken immediately (first data sets to regional centers early February). The meeting was held without the benefit of both Australian and Global center operators. Those present did review all comments sent by those not attending but all suggestions and questions may not be addressed below. If you have any comments, please send them to WOCE.UOT.DAC so all will stay informed. You will note there are differences in some procedures described below from those originally planned, particularly in the QC to be performed at the regional centers. We would like to emphasize that we are still striving for consistency in the QC procedures and would like all centers to standardize methodology as much as possible.
MEDS FORMAT
The format adopted for RT data within the GTSPP will be used for the delivery of data to the regional WOCE DACs. We refer to this as MEDS format throughout this paper. The format is described in Annex K of the report of the first session of the GTSPP Steering Group, 17-19 September 1990, Brest. Bob Keeley provided copies of data that had been processed and put into the MEDS format and walked us through the details of an XBT message pointing out the more important elements. The format and examples will be sent to all three components in Australia. The format is relatively straight forward and should not create any major problems in the WOCE UOT/DAC system. The next step is to make the regional systems compatible with MEDS by changing internally to MEDS or by converting the MEDS formatted data into the regional format before processing. This implies converting back once processing is complete. Regional centers will return the entire record to NODC along with the flags. Would each regional center please send Bob Keeley a description of your format immediately. Bob will look it over and suggest which way to proceed. He may be able to assist with the conversion software, for both ends.
QUALITY CONTROL
The QC of XBT data takes place at several levels, some of which WOCE will not be able to influence. For example National operational centers change some data before entering messages on the GTS. We will try to identify where this is happening, for example NMC in the USA, and discuss the alternatives. If changes are necessary for whatever the reason, the original data should be kept in a history record as planned by MEDS. The data are often QC'ed on board ship. I'm quite sure we will have difficulty influencing that, but shipboard QC should be better than QC done at a national facility in real time.
The GTSPP QC at MEDS is described in the GTSPP real-time QC manual which was "adopted" in Brest in September 1990. It reflects the comments provided by WOCE, primarily by Meyers. Please let us know if you do not have a copy, we will send you one. The data QC'ed by MEDS are not changed in any way, but header information that can be corrected will be. For example, ships are tracked for logical positions so when it shows up in the middle of a continent or other hemisphere, a change can be made. I believe we can accept that type of change. The basic checks, which have several elements each, are: Location and Identification; Profile (e.g. spikes); climatology (Levitus); Profile consistency, i.e. waterfall; and Visual. All problems are flagged and sent with the record to NODC and then the Regional Centers. MEDS will use the following flags:
1 - Good
2 - Inconsistent
3 - Doubtful
4 - Bad
These flags are consistent with IGOSS. We suggested that another flag (6) be introduced to indicate that some data in the file are useful even though a particular data set has bad elements. The practice is to place the lowest rating in the Quick look indicator. For example, if some parts of an observation are good and others bad, the Quick look indicator would be 6, not 4. Why didn't we select 5 instead of six? A 5 indicates that elements have been changed in the IGOSS system. Bob Keeley will send you a more complete description of the flags.
The regional center responsibility is to QC the data sets, replace those flags introduced by MEDS based on the scientific QC, place those flags replaced in the historical section of the file and return the whole lot to NODC. The regional centers will use the same flagging system (1-4, 6).
The QC suggested for the Regional centers is basically the same as discussed in Tallahassee February 1990; profile, histogram and mapping tests. Smith and Meyers had suggested that an objective method would be adequate for the RT data, i.e. profile visual inspection not necessary at this stage. The latter should be part of the delayed mode QC in their opinion. Profile inspection will remain part of the RT mode, each center doing the best they can in this regard. As delayed mode data are often slow to come in, 2 to 5 years to collect, and at 5 years we only have 80% of the RT replaced, the RT data sets need careful scrutiny as they will be used in many cases in the research data set. May we have your comments on this point, particularly if you have strong disagreement.
Each center should generate histograms from their own climatology. They should be monthly, 5 degree squares, based on 10 or more observations and for every 50 meter depth.
Each center uses its own objective analysis technique and it is suggested that we continue this practice but build in a scheme to compare results. The suggestion is that all three centers analyze the Indian Ocean for this comparison. White and Molinari will therefore analyze two basins each period. (Note this does not imply that all three centers will send Indian Ocean flags back to NODC, that remains an Australian responsibility.) Some of the near-real time data that comes in after the mapping could be used for an independent test of the results. The objective is to map each month. White will start bi-monthly and switch to monthly in about a year, Molinari will map monthly and we assume Smith will also. Levels to be mapped are: SST, depth averaged temp between 0-400 m, Temp at 150 m, Depth of 14 degrees C, and MLD (.5 difference) for the tropical belt. Suggested grid is 2 x 5 degrees.
SCHEDULE
Following is the present understanding of the schedule for the exchange, QC and storage of data starting with the January 1991 data set. Remember, 1990 was to be addressed entirely in the delayed mode.
MEDS collect GTS, NMC and FNOC files daily.
MEDS pass daily processed file to NODC in one or two days of the end of the month. White wants the file ASAP, Molinari a week or so. Australia should give its requirements and identify to whom the sets should be sent.
Regional centers pass to NODC bi-monthly QC'ed sets. Note this is a change from the previous monthly schedule but regional centers may still return data to NODC monthly if they wish. Data are to be passed back to NODC by the 15th. of each bi-monthly period. NODC will forward these same sets to MEDS.
NODC provides Brest sets of RT and delayed mode data at 6 month intervals. Brest may want to reconsider and receive the RT sets on a bi-monthly schedule as well. We did not talk about the Brest operation much in Carmel but we assume they are gearing up to expand to include all WOCE data.
OTHER ACTION ITEMS IN BRIEF
WOCE Operations Manual - IPO is putting together an operations manual. We will need details of the UOT/DAC activities. Thompson will organize but needs your inputs.
Late RT data QC - We are always going to be receiving data for our analysis or QC periods just after the close of the time period. This is not a new problem, but we must address this issue as we don't want bad data getting into the system nor have to wait until the delayed mode takes over so to speak. As a first step, Keeley will determine the percent of data that are late so we can assess the problem. The proposal is that late data be subjected to at least two QC tests: profile and histogram.
Converting to GTSPP means the regional centers could end up with historical data sets in one format, future sets in another. NODC will look into providing help for the conversion of your historical files.
Salinity QC - White will begin to develop a salinity QC programme. Any suggestions or information on this would be welcomed.
Data Availability - To WOCE and all GTSPP participants at any time from MEDS and NODC. Contacts are Withee and Hamilton at NODC and assume Wilson and Keeley at MEDS. This implies one can get the data even before the regional QC. That's OK, they just must be made aware of the processing it has gone through. While data will be available to all, they may not be free to all. Rebert, what is the Global enter distribution policy?
Levitus climatology - Being updated by Levitus at NODC. We should keep tabs of this to see how it can be of use to WOCE.
NODC will look into putting data sets in track form to simplify the processing at the regional centers.
Tracking XBT ops - IPO has started the tracking system and believe we are beginning to achieve results. You can see what is happening by accessing OCEANIC and asking to see the WOCE field programme, XBT section. We have a long way to go and will need your help. At Meyers' suggestion, IPO and DIU will issue a 1990 summary early in 1991.