Home

Argo quality controls manual, real-time & delayed-mode

1. PRES X to see whether the T S curve can be recovered Air bubbles in the pressure transducer can also cause erroneous pressure measurements that are visible as anomalous T S curves 10 Conductivity cell thermal mass error Salinity reported immediately after a float has crossed a strong thermal gradient can be in error as a result of conductivity cell thermal mass This error arises because the thermal inertia in the flow duct alters the temperature of water entering the conductivity cell thus inducing a conductivity error For details please refer to Johnson et al 2007 A float that transits from cold to warm water can result in fresh error and from warm to cold in salty error These errors can exceed 0 01 PSS 78 for strong thermal gradients and sometimes result in unstable fresh spikes at the base of the mixed layer This salinity error can be corrected if the ascent rate of the float is known A correction algorithm is available from Greg Johnson at Gregory C Johnson noaa gov Argo data management quality control manual version 2 9 1 T m 11 Abnormal salinity hooks at base of profiles In some floats salty or fresh salinity hooks may be observed at the base of the CTD profiles The hook appearance occurs when the two deepest measurements are taken at nearly identical pressures usually less than 5dbar apart but salinity from the first measurment looks abnormal relative to the second measurement The first deeper
2. Argo data management quality control manual version 2 9 1 T 2 Despike the SP time series to 1 dbar This is most effectively done by first removing the more conspicuous spikes that are bigger than 5 dbar as in the real time procedure then the more subtle spikes that are between 1 to 5 dbar by comparing the SP values with those derived from a 5 point median filter For standard Argo floats that sample every 10 days a 5 point filter represents a filter window of 40 days 20 days from a profile which is an appropriate time scale for retaining effects from the atmospheric seasonal cycle 3 Replace the spikes and any other missing SP values by interpolating between good neighbouring points If missing values occur at the ends of the SP time series extrapolate from the nearest good points 4 The resulting SP time series should then be inspected visually to make sure there are no more erroneous points The clean SP value from cycle i 1 is then used to adjust pressures from cycle i by PRES ADJUSTED cycle i PRES cycle i SP cycle i 1 The CTD profile and the associated SP is staggered by one cycle because the SP measurement is taken after the telemetry period and therefore is stored in the memory and telemetered during the next cycle The real time procedure does not match SP value from cycle i 1 with PRES from cycle 7 because real time adjustment cannot wait 10 days However in delayed mode it is important to
3. ADJUSTED ERROR 2 4 dbar is the recommended error to quote with 2 4 dbar being the manufacturer quoted accuracy of the pressure sensor Salinity should be re calculated by using PRES ADJUSTED and recorded in PSAL ADJUSTED Salinity error due to pressure uncertainty is negligible and can be ignored in the consideration of PSAL ADJUSTED ERROR Please use the SCIENTIFIC CALIBRATION section in the netCDF files to record details of the delayed mode adjustment Argo data management quality control manual version 2 9 1 30 Note to users The 1 dbar despiking threshold for SP assumes that spikes greater than 1 dbar represent noise in the SP measurement that should not be integrated into float pressures After despiking to 1 dbar the remaining SP values contain sea surface atmospheric pressure variations and variations due to other high frequency surface processes While sea surface atmospheric pressure variations affect the whole water column and therefore should be adjusted for high frequency surface processes do not affect the whole water column Therefore users should be aware that PRES ADJUSTED contains noise from high frequency surface processes that are of the order lt 1 dbar In addition other more subtle pressure errors such as those due to non linear hysteresis and other temperature and pressure dependent effects are not accounted for in PRES ADJUSTED Hence users should always heed the error bars quoted in PRES ADJUSTED ERROR 3 2
4. cell such as etching scouring or dissolution of the glass surface will result in erroneously salty salinity measurements 4 Conductivity cell geometry changes The geometry of conductivity cells can change thus causing electrodes to change distance This will result in either an increase or a decrease in salinity values 5 Conductivity cell circuit changes The circuit within the conductivity cell contains capacitors and resistors Changes to any of these electrical components will affect electrical conductivity and thus will give erroneous fresh or salty salinity measurements Electrical complication can result in sensor drifts that have varying drift rates e g drift rates can change from slow and linear to exponential Usually jumps in salinity measurements are an indication of electrical malfunction If electrical complication is suspected delayed mode analysts should check the shape of the vertical salinity profiles for adjustability Usually the vertical profiles after a measurement jump are wrong and so are unadjustable 6 Low voltage at end of float life and Energy Flu Some floats experience a sudden rapid decrease in available battery energy reserves This premature exhaustion of battery known as Energy Flu usually starts about 2 years after deployment The sharp drop in battery voltage related to Energy Flu as well as the low voltage towards the end of a float s natural life will produce low of corre
5. floats that used Druck pressure sensors with serial numbers greater than 2324175 or were deployed after 1 October 2006 if the Druck serial numbers are unknown the likelihood of them being affected by the microleak disease is elevated about 30 For these suspicious data an upper bound of the estimated error should be cited Since a negative 20 dbar pressure error will cause a positive 0 01 salinity error at which point T S anomalies will become observable and data should be flagged as 4 as described in Category 2 b below 20 dbar has been chosen as the upper bound of the data error for this group PRES ADJUSTED ERROR 20 dbar Note that SeaBird will eventually provide a list of serial numbers that represents Druck sensors that have been screened as healthy These healthy Druck sensors should be excluded from receiving the larger pressure error bar Moreover SeaBird has records that connect the Druck serial number to CTD number and Teledyne WRC can make the connection to the float hull number Argo data management quality control manual version 2 9 1 38 2 When float data show observable T S anomalies that are consistent with increasingly negative pressure drift after cycle n This means that the TNPD data have unknown negative pressure error and that the error becomes severe after cycle n a For the less severe part of the TNPD data before cycle n the adjusted variables should receive a dmqc flag of 2 PRES ADJUS
6. is an example AS Constant offset Discontinuity no transition phase Sensor drift with a constant drift rate Transition phase Stable time Argo data management quality control manual version 2 9 1 a 3 Choose length of sliding calibration window for each segment These can be Long window 6 months or greater for the stable regime or highly variable regimes where a long window is required to average over oceanographic variability to detect slow sensor drift or period of constant drift rate Short window can be as short as 10 days for the transition phase Zero length window for spikes That is adjust single profile 4 Select temperature levels for exclusion from least squares fit e g seasonal mixed layer highly variable water masses 5 Calculate proposed adjustment for each segment The assembled proposed adjustments for the entire float series should be continuous and piecewise linear within error bars except where the delayed mode operator believes there is a genuine discontinuity In general the delayed mode operator should aim to use as long a calibration window as possible because a long calibration window where the least squares fit is calculated over many cycles will average over oceanographic noise and thus give a stable calibration Hence splitting the float series into short segments is to be avoided short segments mean short calibration windows hence
7. match the CTD profile with the staggered telemetry of SP because SP values can contain synoptic atmospheric variations and because a missing CTD profile is often associated with an erroneous SP point By this scheme SP 1 which is taken before cycle and therefore before the float has had its first full dive is not used in delayed mode Note that the real time procedure does not adjust for pressure offsets that are greater than 20 dbar or less than 20 dbar This is because the real time automatic procedure cannot determine whether SP values greater than 20 dbar or less than 20 dbar represent true sensor drift or erroneous measurements Instead in real time floats that return SP values greater than 20 dbar or less than 20 dbar for more than 5 consecutive cycles are grey listed in consultation with the PI In delayed mode operators can inspect the SP time series visually when severe pressure sensor drift occurs Therefore there is no upper limit to the magnitude of delayed mode pressure adjustment After adjustment delayed mode operators should check that PRES ADJUSTED gt 0 If PRES ADJUSTED lt 0 delayed mode operators should check for decoding error in SP or in the CTD pressures In addition the following should be observed PRES should always record the raw data PRES ADJUSTED QC should be set appropriately For example floats that have had significant pressure adjustment should have PRES ADJUSTED QC 2 PRES
8. measurement is taken at the end point of descent the second shallower measurement is the first deep sample taken during ascent The abnormal deeper salinity reading is caused by water in the conductivity cell carried from the surface or park level to deep profile level not being flushed out completely before the descent end point sample is taken Salty hooks are produced when surface or park level water is saltier than deep water fresh hooks are produced when surface or park level water is fresher than deep water These salinity hooks cannot be detected by the real time tests so delayed mode analysts are urged to examine carefully the base of CTD profiles for these abnormal salinity values and to flag them appropriately in delayed mode Argo data management quality control manual version 2 9 1 Sal 4 4 Criteria for CTD profiles to be retained in the reference database The following criteria are used to select CTD data as reference for delayed mode quality control of Argo salinity profiles in the open ocean 1 Use only data that have passed all NODC quality control tests for observed level data 2 Use all country codes 3 Use only profiles that sampled deeper than 900 dbar 4 Weed out all data points outside these ranges 24 lt S lt 41 0 01 lt P lt 9999 0 C lt T lt 40 C except for WMO boxes with latitudes north of 60 N or south of 50 S where 2 5 C lt T lt 40 C 5 For WMO boxes that contain
9. result is a series of measurements from very close together depth levels Progressively shallower profiles and close together measurements are therefore two ways to identify whether the Druck pressure sensor has been contaminated When the Druck pressure sensor has been contaminated pressure measurements become suspect and should be considered bad The corresponding temperature and salinity measurements are therefore also suspect and should be considered bad 8 Druck pressure sensor oil microleak problem Another pathology in Druck pressure sensors is oil microleak past the glass metal seal This oil leak leads to an internal volume loss which then exhibits itself as an increasing negative offset at all pressures At the early stages of oil microleak float measurements are still correctable and usable However as more and more oil is leaked the flexible titanium diaphragm will dip so far down the oil chamber that it will short the electrical parts causing erratic behaviour in float measurements This is the end stage of oil microleak and the data at this point are bad and uncorrectable 9 Incorrect pressure sensor coefficient Incorrect scaling coefficient in the pressure sensor will give anomalous T S curves at depth The T S relation will look acceptable but at depth it will look as if the float is sampling an anomalous water mass relative to nearby floats Delayed mode analysts should try to re scale pressure measurements e g PRES
10. sample up to the sea surface with the pump off or with the pump on closer to the sea surface or carry auxiliary modules for high resolution near surface sampling These specialised near surface data are focused on the top 5 dbar of the ocean They may extend deeper than 5 dbar so as to overlap with the primary CTD profiles for the purpose of cross calibration They are stored as additional profiles N PROF gt 1 in the single cycle core Argo profile files and are identifiable by VERTICAL SAMPLING SCHEME Near surface sampling Note the full character string is Near surface sampling averaged discrete mixed pumped unpumped optional description Please refer to Table 16 in the Argo Users Manual for details of the various vertical sampling schemes and their full character strings The following tests are applied in real time to these specialised near surface data 6 Global range test Same as in Section 2 1 2 7 Regional range test Same as in Section 2 1 2 8 Pressure increasing test Same as in Section 2 1 2 9 Spike test Same as in Section 2 1 2 11 Gradient test Same as in Section 2 1 2 19 Deepest pressure test Same as in Section 2 1 2 21 Near surface unpumped CTD salinity test Details described in this section 22 Near surface mixed air water test Details described in this section Argo data management quality control manual version 2 9 1 24 21 Near surface unpumped CTD salinity test W
11. test This test can detect a float that reproduces the same profile with very small deviations over and over again Typically the differences between two profiles are of the order of 0 001 PSU for salinity and of the order of 0 01 C for temperature A Derive temperature and salinity profiles by averaging the original profiles to get mean values for each profile in 50 dbar slabs Tprof T previous prof and Sprof S_previous_prof This is necessary because the floats do not sample at the same level for each profile B Substract the two resulting profiles for temperature and salinity to get absolute difference profiles e deltaT abs Tprof T previous prof e deltaS abs Sprof S previous prof C Derive the maximum minimum and mean of the absolute differences for temperature and salinity e mean deltaT max deltaT min deltaT e mean deltaS max deltaS min deltaS D To fail the test require that e max deltaT 0 3 Argo data management quality control manual version 2 9 1 E e min deltaT lt 0 001 e mean deltaT lt 0 02 e max deltaS lt 0 3 e min deltaS lt 0 001 e mean deltaS lt 0 004 Action if a profile fails this test all measurements for this profile are flagged as bad data flag 4 If the float fails the test on 5 consecutive cycles it is inserted in the grey list 19 Deepest pressure test This test requires that the profile has pressures that are not higher than DEEPEST
12. unstable calibrations 3 4 4 The PI evaluation part The PI PI means Principal Investigator or responsible persons assigned by the PI should first check that the statistical recommendations are appropriate This is because the semi automatic methods cannot distinguish ocean features such as eddies fronts and water mass boundaries Near such ocean features semi automatic statistical methods are likely to produce erroneous estimations The associated uncertainties reflect the degree of local variability as well as the sparsity of reference data used in the statistical estimations However these associated uncertainties are sensitive to the choice of scales Hence the PI also needs to determine that the associated uncertainties are realistic The PI then determines whether the proposed statistical adjustment is due to sensor malfunction or ocean variability Care should be taken to not confuse real ocean events with sensor drifts and offsets This is done by inspecting as long a float series as possible and by evaluating other independent information Some of the diagnostic tools are nspecting the trend of AS over time Trends that reverse directions or oscillate are difficult to explain in terms of systematic sensor malfunction These are often caused by the float sampling oceanographic features e g eddies fronts etc that are not adequately described in the reference database Visually checking the float trajectory with refere
13. where AS is the difference in salinity between float measured values and statistical recommendations If AS a bt where t is time then a is the offset and b is the drift rate Note that these drifts and offsets can be sensor related or they can be due to real ocean events Evaluation by salinity experts is needed to distinguish between sensor errors and real ocean events OFFSET DRIFT AS AS time time Argo data management quality control manual version 2 9 1 r 3 4 3 Splitting the float series and length of calibration window If a float exhibits changing behaviour during its lifetime the time series of AS should be split into separate segments according to the different behaviours so that one time segment does not contaminate the other during the least squares fit process e g the slowly fouling segment does not contaminate the stable segment The following is a step by step guide on how to deal with float salinity time series with changing behaviours 1 Identify different regimes in the time series These can be Stable measurements no sensor drift including constant offsets Sensor drift with a constant drift rate Spikes Transition phase where drift rate changes rapidly e g elbow region between stable measurements and constant drift or initial biocide wash off 2 Split the time series into discrete segments according to these different regimes or when there are too many missing cycles Here
14. 2 Truncated negative pressure drift TNPD in APEX floats APEX floats with Apf 5 Apf 7 or Apf 8 controllers that set negative SURFACE PRESSURE SP to zero then add an artificial 5 dbar present a challenge to delayed mode qc because information from SP on any negative pressure offset is lost thus making the pressure data unadjustable The problem with some of these APEX floats having unknown negative pressure error escalated with the discovery of the oil microleak defect in Druck pressure sensors The Druck oil microleak defect manifests itself as increasingly negative offset at all pressures and will eventually end the useful life of the float During delayed mode qc of pressure measurements from APEX floats with Apf 5 Apf 7 or Apf 8 controllers operators should first remove erroneous SP values and any isolated spikes in the time series by following the procedure described in Section 3 2 1 The delayed mode operator should then examine the resulting valid and despiked SP time series and determine whether there are long periods of zero SP readings after removing the artificial 5 dbar that qualify as Truncated Negative Pressure Drift which has the following definition Truncated Negative Pressure Drift TNPD refers to the part of a float s time series from which valid and despiked SP after removing the artificial 5 dbar reads continuously zero without reverting back to positive values or containing any occasional positive valu
15. 4175 occurred after October 2006 Since July 2009 SeaBird has begun screening Druck pressure sensors in order to identify those transducers that have microleaks One way to identify affected floats is by T S analysis since severe pressure error will lead to observable T S anomalies Anomalies associated with severe negative pressure drift include a Positive salinity drift e g pressure error of 20 dbar will cause a positive salinity error of approximately 0 01 PSS 78 Statistical comparison methods that are used to determine conductivity sensor drift e g WJO BS OW can be used as diagnostic tools for these cases Please refer to Section 3 4 1 for descriptions of these statistical comparison methods b Cold temperature anomaly whose size depends on vertical temperature gradient c Float derived dynamic height anomalies significantly lower than satellite derived sea level anomalies d Shoaling of isotherm depths independent of time space migration of the float In addition to T S analysis delayed mode operators should also observe when a float begins telemetering highly erratic data This is a sign that it may be suffering from the Druck oil microleak problem and that the pressure sensor may be about to fail completely Note that symptoms of this failure are very similar to those of the Druck snowflakes problem which affected floats that were manufactured in 2002 and 2003 Please refer to Appendix 4 3 for a brief des
16. ATION DATE should have non FillValue entries in every N PARAM dimension and should have format YYYYMMDDHHMISS seconds must be O to 59 10 DATE UPDATE should be equal to or later than any CALIBRATION DATE HISTORY DATE DATE CREATION JULD and JULD LOCATION 11 There should be at least one HISTORY record 12 All dates must be after 1st Jan 1997 and before GDAC file time 13 All dates must be 14 digit strings in the format YYYYMMDDHHMISS seconds must be 0 to 59 14 Character strings should not contain the NULL character Argo data management quality control manual version 2 9 1
17. Argo data management DOI http dx doi org 10 13155 33951 argo data management Argo Quality Control Manual Version 2 9 1 18 November 2014 ARGO part of the integrated global observation strategy Argo Quality Control Manual Authors Annie Wong Robert Keeley Thierry Carval and the Argo Data Management Team How to cite this document Annie Wong Robert Keeley Thierry Carval and the Argo Data Management Team 2014 Argo Quality Control Manual http dx doi org 10 13155 33951 OT Table of contents 1 INTRODUCTION 7 2 REAL TIME QUALITY CONTROLS 8 2 1 ARGO REAL TIME QUALITY CONTROL TEST PROCEDURES ON VERTICAL PROFILES 8 2 1 1 INTRODUCTION 8 2 1 2 QUALITY CONTROL TESTS ON VERTICAL PROFILES 9 2 1 3 TESTS APPLICATION ORDER ON VERTICAL PROFILES 16 2 1 4 QUALITY CONTROL FLAG APPLICATION POLICY 16 2 24 ARGO REAL TIME QUALITY CONTROL TEST PROCEDURES ON TRAJECTORIES 17 2 3 ARGO REAL TIME ADJUSTMENTS ON VERTICAL PROFILES 20 2 3 1 REAL TIME PRESSURE ADJUSTMENT FOR NON AUTO CORRECTING FLOATS 20 2 3 2 REAL TIME SALINITY ADJUSTMENT 21 2 3 3 REAL TIME FILES WITH DATA MODE A 21 2 4 FEEDBACK FROM STATISTICAL TEST AT CORIOLIS 22 2 5 ARGO REAL TIME QUALITY CONTROL TEST PROCEDURES ON NEAR SURFACE DATA 23 2 6 27 3 DELAYED MODE QUALITY CONTROLS 28 3 1 EDITING RAW QC FLAGS IN DELAYED MODE 28 3 2 DELAYED MODE PROCEDURES FOR PRESSURE 28 3 2 1 DELAYED MODE PRESSURE ADJUSTMENT FOR APEX FLOATS 28 3 2 2 T
18. Argo data management quality control manual version 2 9 1 T The decision to insert a float parameter in the grey list comes from the PI or the delayed mode operator A float parameter should be put in the grey list when sensor drift is too big to be corrected adequately in real time or when the sensor is judged to be not working correctly The grey list only concerns real time files R files When an anomalous float is dead and has been adjusted in delayed mode it should not appear in the grey list When an anomalous float is active and has been partially adjusted in delayed mode it should remain in the grey list only if real time adjustment is not adequate 16 Gross salinity or temperature sensor drift This test is implemented to detect a sudden and significant sensor drift It calculates the average salinity on the deepest 100 dbar on a profile and the previous good profile Only measurements with good QC are used Action if the difference between the two average values is more than 0 5 PSU then all measurements for this parameter are flagged as probably bad data 3 The same test is applied to temperature if the difference between the two average values is more than 1 C then all measurements for this parameter are flagged as probably bad data 3 17 Visual QC Subjective visual inspection of float values by an operator To avoid delays this test is not mandatory before real time distribution 18 Frozen profile
19. CIENTIFIC CALIB COMMENT Sensor drift detected Adjusted float salinity to statistical recommendation as in WJO 2003 with WOD2001 as the reference database Mapping scales used are 8 4 4 2 Length of sliding calibration window is 20 profiles Regardless of whether an adjustment has been made or not the date of delayed mode qc for each measurement parameter e g PRES TEMP PSAL CNDC DOXY should be recorded in SCIENTIFIC CALIB DATE in the format YYYYMMDDHHMISS 3 5 3 Other variables in the netcdf file A history record should be appended to the HISTORY section of the netcdf file to indicate that the netcdf file has been through the delayed mode process Please refer to the Argo User s Manual 85 Using the History section of the Argo netCDF Structure on usage of the History section The variable DATA MODE should record D The variable DATA STATE INDICATOR should record 2C or 2C The variable DATE UPDATE should record the date of last update of the netcdf file in the format Y Y Y YMMDDHHMISS Lastly the name of the single cycle profile file is changed from R nc to D nc Argo data management quality control manual version 2 9 1 T 4 Appendix 4 1 Reference Table 2 Argo quality control flag scale This table describes the Argo qc flag scales Please note that this table is used for all measured parameters This table is named Reference Table 2 in the Argo User s Manual Flag Mean
20. Cs contains the DATE UPDATE information along with other information for every file on the GDACs and can be used to monitor updates The profile index file is maintained in the top level GDAC directory and is named ar_index_global_prof txt index files also exist for the meta data and trajectory files 3 4 8 References Bohme L and U Send 2005 Objective analyses of hydrographic data for referencing profiling float salinities in highly variable environments Deep Sea Research IT 52 3 4 651 664 Johnson G C J M Toole and N G Larson 2007 Sensor corrections for Sea Bird SBE 41CP and SBE 41 CTDs Journal of Atmospheric and Oceanic Technology 24 1117 1130 Owens W B and A P S Wong 2009 An improved calibration method for the drift of the conductivity sensor on autonomous CTD profiling floats by 0 S climatology Deep Sea Research Part I Oceanographic Research Papers 56 3 450 457 Wong A P S G C Johnson and W B Owens 2003 Delayed mode calibration of autonomous CTD profiling float salinity data by 0 S climatology Journal of Atmospheric and Oceanic Technology 20 308 318 Argo data management quality control manual version 2 9 1 A 3 5 Compulsory variables to be filled in a D file This section lists the compulsory variables that must be filled in an Argo netCDF profile file that has been through the delayed mode process 3 5 1 Measurements for each profile The following are compulsory measurem
21. JUSTED FillValue PSAL ADJUSTED ERROR FillValue PSAL ADJUSTED QC 4 quality control manual version 2 9 1 T ll 3 4 7 Timeframe for availability of delayed mode salinity data The statistical methods used in the Argo delayed mode process for checking sensor drifts and offsets in salinity require the accumulation of a time series for reliable evaluation of the sensor trend Timeframe for availability of delayed mode salinity data is therefore dependent on the sensor trend Some floats need a longer time series than others for stable calibration Thus delayed mode salinity data for the most recent profile may not be available until sufficient subsequent profiles have been accumulated The default length of time series for evaluating sensor drift is 12 months 6 months before and 6 months after the profile This means that in general the timeframe of availability of drift adjusted delayed mode salinity data is 6 months after a profile is sampled Users should also be aware that changes may be made to delayed mode files at any time by DACs and delayed mode operators For example delayed mode files may be revised when new CTD or float data become available after the original delayed mode assessment and adjustment The date of latest adjustment of a parameter can be found in SCIENTIFIC_CALIB_DATE Anytime an Argo file is updated for any reason the DATE UPDATE variable will reflect the date of the update The profile index file on the GDA
22. PRESSURE plus 10 DEEPEST PRESSURE value comes from the meta data file of the float Action If there is a region of incorrect pressures all pressures and corresponding measurements should be flagged as bad data flag 4 All pressures flagged as bad data and all of the associated temperatures and salinities are removed from the TESAC distributed on the GTS Argo data management quality control manual version 2 9 1 T 2 1 3 Tests application order on vertical profiles The Argo real time QC tests on vertical profiles are applied in the order described in the following table Order Test Number Test Name 3 2 jmpssbeDaeTest 3 Impossible Location Test Position on Land Test DE TET Speed Test Global Range Test Regional Range Test fo a Pressure Increasing Test 1o 9 Spike Test 1 m Gradient Test 13 Stuck Value Test Density Inversion 16 15 Greytist 47 46 Gross salinity or temperature sensor drift 18 48 Frozen profile 19 17 Msuaroc 2 1 4 Quality control flag application policy The QC flag value assigned by a test cannot override a higher value from a previous test Example a QC flag 4 bad data set by Test 11 gradient test cannot be decreased to QC flag 3 bad data that are potentially correctable set by Test 15 grey list A value with QC flag 4 bad data or 3 bad data that are potentially correctable is ignored by the quality control test
23. RUNCATED NEGATIVE PRESSURE DRIFT TNPD IN APEX FLOATS 30 3 2 3 SUMMARY FLOWCHART FOR UNADJUSTABLE APEX PRESSURES 36 3 3 DELAYED MODE PROCEDURES FOR TEMPERATURE 37 3 4 DELAYED MODE PROCEDURES FOR SALINITY 38 3 4 1 INTRODUCTION 38 3 4 2 QUALITY CONTROL AND THE SEMI AUTOMATIC PART 39 3 4 3 SPLITTING THE FLOAT SERIES AND LENGTH OF CALIBRATION WINDOW 40 3 4 4 THE PI EVALUATION PART 41 3 4 5 ASSIGNING ADJUSTED SALINITY ERROR ESTIMATES AND QC FLAGS 42 Argo data management quality control manual version 2 9 1 al 3 4 6 SUMMARY FLOWCHART FOR SALINITY 45 3 4 7 TIMEFRAME FOR AVAILABILITY OF DELAYED MODE SALINITY DATA 46 3 4 8 REFERENCES 46 3 5 COMPULSORY VARIABLES TO BE FILLED IN A D FILE 47 3 5 1 MEASUREMENTS FOR EACH PROFILE 47 3 5 2 SCIENTIFIC CALIBRATION INFORMATION FOR EACH PROFILE 47 3 5 3 OTHER VARIABLES IN THE NETCDF FILE 48 4 APPENDIX 49 4 1 REFERENCE TABLE 2 ARGO QUALITY CONTROL FLAG SCALE 49 4 2 REFERENCE TABLE 2A PROFILE QUALITY FLAGS 50 4 3 COMMON INSTRUMENT ERRORS AND FAILURE MODES 51 4 4 CRITERIA FOR CTD PROFILES TO BE RETAINED IN THE REFERENCE DATABASE 54 4 5 CRITERIA FOR ARGO PROFILES TO BE RETAINED IN THE REFERENCE DATABASE 55 4 6 CONSISTENCY CHECKS FOR D FILES FORMAT AT THE GDACS 56 Argo data management quality control manual version 2 9 1 History Date Comment dd mny yyyy Added inversion test Test 10 removed 07 10 2004 1 Real time and delayed mode manuals merged in Argo quality contr
24. STED FillValue PSAL ADJUSTED QC 4 PSAL ADJUSTED ERROR FillValue For float salinity that are affected by minor instrument errors and that are considered adjustable in delayed mode These are measurements that are affected by minor instrument errors whose effects can be rectified by delayed mode adjustments These include minor calibration offsets and minor sensor drifts where the data still retain a relative vertical T S shape that is close to good data When an adjustment has been applied PSAL ADJUSTED original value adjustment recommended by statistical analysis or adjustment provided by PI PSAL ADJUSTED QC l or 2 PSAL ADJUSTED ERROR maximum Yadjustment error 0 01 where adjustment error is the uncertainty from each type of adjustment applied to PSAL These can be statistical uncertainty from sensor drift adjustment uncertainty from conductivity cell thermal mass adjustment etc Argo data management quality control manual version 2 9 1 a Upper limit of delayed mode salinity adjustment 0 05 PSS 78 should be considered as the upper limit of good salinity adjustment in delayed mode If the magnitude of sensor drift or calibration offset exceeds 0 05 then either i salinity data are considered unadjustable in delayed mode PSAL _ ADJUSTED FillValue PSAL ADJUSTED QC 4 PSAL ADJUSTED ERROR FillValue or 11 salinity data are considered adjustable but confidence in adjus
25. TED QC 2 TEMP_ADJUSTED_QC 2 PSAL ADJUSTED QC 2 while the pressure error should increase to 20 dbar PRES ADJUSTED ERROR 20 dbar b For the severe part of the TNPD data after cycle n the adjusted variables should receive a dmqc flag of 4 PRES ADJUSTED QC 4 TEMP ADJUSTED QC 4 PSAL ADJUSTED QC 4 Please note that whenever PARAM ADJUSTED QC 4 PARAM ADJUSTED FillValue and PARAM ADJUSTED ERROR FillValue Note For the severe cases in Category 2 b delayed mode operators in consultation with float PIs should consider putting the real time data on the grey list Example e A complex case belonging to Category 2 where T S anomalies consistent with increasingly negative pressure drift are observed after cycle n part way through the TNPD portion of the time series SP dbar PARAM ADJUSTED QC 4 PARAM ADJUSTED QC o PRES ADJUSTED ERROR FillValue PRES ADJUSTED ERROR Adjustable 20 dbar Refer to Section 3 2 1 a T S anomalies consistent with TNPD status begins increasingly negative pressure drift are observable after cycle n Argo data management quality control manual version 2 9 1 T All TNPD data should receive a standard label in SCIENTIFIC CALIB COMMENT in the Argo single cycle netcdf files in the dimension corresponding to PRES The standard label consists of the character string TNPD APEX float that truncated negati
26. _ fields are filled in A mode the adjustments are applied in an automated manner in real time and are not checked by delayed mode operators Argo data management quality control manual version 2 9 1 m 2 4 Feedback from Statistical Test at Coriolis At Coriolis an objective analysis is performed on a daily basis on Argo temperature and salinity profiles that have been quality controlled during the previous 3 weeks As results of the comparison with climatology anomalies on Argo profiles are detected by this objective analysis For flag correction on those profiles daily automatic feedbacks in text files by email are sent to the appropriate DAC The email message contains the list of Argo profiles highlighted by the objective analysis and examined by a Coriolis operator with the recommended flag correction listed at the end The information is also available in a csv format file on the ftp site e ftp ftp ifremer fr ifremer argo etc Objective Analysis Warning Argo data management quality control manual version 2 9 1 BT 2 5 Argo Real time Quality Control Test Procedures on near surface data The near surface data described in this section are specialised data that are collected with vertical sampling methods different from the primary CTD profiles For most profiling floats the CTD pump is normally switched off at around 5 dbar during ascent to avoid contamination of the conductivity cell Several float types continue to
27. a management quality control manual version 2 9 1 286 Various types of specialised near surface data in Argo that can be found under VERTICAL SAMPLING SCHEME Near surface sampling Float type o operating mode Me Test21 Test 22 PROVOR ARVOR primary CTD unpumped T amp S averaged a a NOVA primary CTD unpumped T amp S averaged a a APEX Argos NST primary CTD unpumped T only discrete n a a APEX Iridium STS auxiliary module free flushing T amp S discrete n a a SOLO II S2A primary CTD pumped T amp S discrete n a n a Deep SOLO primary CTD pumped T amp S discrete n a n a a applicable n a not applicable Argo data management quality control manual version 2 9 1 7 Argo data management quality control manual version 2 9 1 T m 3 Delayed mode quality controls 3 1 Editing raw qc flags in delayed mode Delayed mode operators should examine profile data for pointwise errors such as spikes and jumps and edit the raw qc flags in PARAM QC when they are set incorrectly PARAM here refers to PRES TEMP CNDC and PSAL Examples where PARAM QC should be edited in delayed mode include a PARAM QC should be changed to 4 for bad and un correctable data that are not detected by the real time tests and b PARAM QC should be changed to 1 or 2 for good data that are wrongly identified as bad or probably bad by the real time tests 3 2 Dela
28. able PSAL QC Float salinity with PSAL QC 4 are bad data that are in general unadjustable However delayed mode operators can evaluate the quality and adjustability of these bad data if they have a reason to do so Please refer to Section 4 1 for definitions of the Argo qc flags in real time The delayed mode operators can edit lt PARAM gt QC if they consider that data are flagged inappropriately In delayed mode float salinity values that have PSAL QC 1 2 or 3 are further examined Anomalies in the relative vertical salinity profile such as measurement spikes and outliers that are not detected in real time are identified Of these anomalies those that will skew the least squares fit in the computation for drift and offset adjustments are excluded from the float series for evaluation of drifts and offsets These measurements are considered unadjustable in delayed mode Float salinity values that are considered adjustable in delayed mode are assembled into time series Sufficiently long time series are compared with statistical recommendations and associated uncertainties to check for sensor drifts and offsets These statistical recommendations and associated uncertainties are obtained by the accepted methods listed in Section 3 4 1 in conjunction with appropriate reference datasets These methods are semi automatic and have quantified uncertainties Drifts and offsets can be identified in the trend of AS over time
29. cription of the Druck snowflakes problem Argo data management quality control manual version 2 9 1 E In light of these events the following categories should be considered in assigning delayed mode qc flags and error bars for data classified as TNPD 1 When float data do not show observable T S anomalies that are consistent with increasingly negative pressure drift This means that the TNPD data may have unknown negative pressure error that is not severe For these less severe cases the adjusted variables should receive a delayed mode qc flag of 2 PRES ADJUSTED QC 2 TEMP ADJUSTED QC 2 PSAL ADJUSTED QC 2 Note that TEMP ADJUSTED QC and PSAL ADJUSTED QC can change to 3 or 4 if TEMP and PSAL contain additional errors that are independent of the pressure error e g pointwise temperature spike conductivity cell drift etc For these less severe cases two groups should be considered in assigning the pressure error bars in delayed mode a For TNPD data belonging to floats that used Druck pressure sensors with serial numbers less than 2324175 or were deployed before 1 October 2006 if the Druck serial numbers are unknown the likelihood of them being affected by the oil microleak problem is low about 3 Hence it is reasonable to cite the manufacturer quoted accuracy of 2 4 dbar as the pressure error for this group PRES ADJUSTED ERROR 24 dbar b For TNPD data belonging to
30. ct salinity values Towards the end of float life low voltage will result in large drift followed by death Energy Flu will cause spikes that get worse and more frequent also followed by death Argo data management quality control manual version 2 9 1 gt 7 Druck pressure sensor snowflakes problem About 4 of SBE41 CTDs that were manufactured in late 2002 through end of 2003 have experienced the Druck pressure sensor snowflakes problem SeaBird has fixed this problem in 2004 so this feature is only included in this section for identifying the historical profiles that have been affected The Druck pressure sensor snowflakes problem is due to internal electrical shorting by the growth of titanium oxide particles snowflakes in the oil filled cavity in the pressure sensor causing the pressure sensor to report erratic pressure measurements or going to full scale i e either report PRES 3000 dbar or 3000 dbar These erratic pressure measurements will preferentially report deeper than correct The firmware tries to adjust the piston according to the erroneous deeper pressures causing the float to park shallower The float will thus progressively become a surface drifter Erroneous deeper pressures will also result in the firmware placing the pointer at the deeper nominal sampling levels in the lookup table thus causing the float to take a sample everytime the firmware performs a lookup every 6 seconds The
31. d to be removed This is done in real time in two steps 1 Discard SP values greater than 20 dbar or less than 20 dbar then revert to the last valid SP 2 If the most recent SP value SP 1 is different from the last valid SP by more than 5 dbar that is if abs SP 1 last valid SP gt 5 dbar revert to the last valid SP When no valid SP value is available no real time pressure adjustment is available When there are valid SP values real time adjusted pressures will be recorded in the variable PRES ADJUSTED where PRES ADJUSTED PRES SP Zero is a valid SP value Therefore when SP 0 PRES ADJUSTED should be filled In this case PRES ADJUSTED PRES Argo data management quality control manual version 2 9 1 al PRES should always record the raw data PRES ADJUSTED QC will be filled with the same values as PRES QC PRES ADJUSTED ERROR and all variables in the SCIENTIFIC CALIBRATION section of the netCDF files will be filled with FillValue DATA MODE will record A There is no need to re calculate salinity data in real time by using the real time adjusted pressure values This is because the difference in salinity due to real time pressure adjustment is small Pressure adjustment of less than 20 dbar will result in salinity error of less than 0 01 When the SP value exceeds 20 dbar or 20 dbar for more than 5 consecutive cycles the float should be put on the grey list because of pressure error af
32. e TESAC message The appropriate actions to take are noted with each test Argo data management quality control manual version 2 9 1 T 2 1 2 Quality control tests on vertical profiles 1 Platform identification Every centre handling float data and posting them to the GTS will need to prepare a metadata file for each float and in this is the WMO number that corresponds to each float ptt There is no reason why except because of a mistake an unknown float ID should appear on the GTS Action If the correspondence between the float ptt cannot be matched to the correct WMO number none of the data from the profile should be distributed on the GTS 2 Impossible date test The test requires that the observation date and time from the float be sensible e Year greater than 1997 e Month in range 1 to 12 e Day in range expected for month e Hour in range 0 to 23 e Minute in range 0 to 59 Action If any one of the conditions is failed the date should be flagged as bad data and none of the data from the profile should be distributed on the GTS 3 Impossible location test The test requires that the observation latitude and longitude from the float be sensible Action If either latitude or longitude fails the position should be flagged as bad data and none of the data from the float should go out on the GTS e Latitude in range 90 to 90 e Longitude in range 180 to 180 4 Position on land test The test requires that the obs
33. e propensity of some floats to acquire TBTO contamination 9 No cycles in the 6 months prior to salinity drift adjustment AS gt 001 PSS 78 in bottom data 10 No cycles whose bottom data have PSAL ADJUSTED ERROR gt 0 015 Argo data management quality control manual version 2 9 1 56 4 6 Consistency checks for D files format at the GDACs The following is a list that is used at the GDACs for checking D files format 1 PARAM ADJUSTED and lt PARAM gt ADJUSTED ERROR must contain data except when lt PARAM gt ADJUSTED QC 4 or 9 Here lt PARAM gt PRES TEMP PSAL and CNDC 2 Where lt PARAM gt DOXY DOXY_ADJUSTED should be filled with the same values as DOXY Furthermore DOXY ADJUSTED QC should record 0 DOXY ADJUSTED ERROR FillValue and PROFILE DOXY QC 3 If PRES ADJUSTED QC 4 then TEMP ADJUSTED QC 4 and PSAL ADJUSTED QC 4 4 lt PARAM gt ADJUSTED QC cannot be 0 except when lt PARAM gt DOXY 5 POSITION QC and JULD QC cannot be 0 6 No variable should be filled with the netCDF value of IEEE NaN 7 In the Scientific Calibration section PARAMETER should have N PARAM entries equal to the number of measurement parameters recorded in the netCDF file 8 In the Scientific Calibration section SCIENTIFIC CALIB COMMENT should have non FillValue entries in every N PARAM dimension 9 In the Scientific Calbration section CALIBR
34. ent variables that must be filled in a D file lt PARAM gt ADJUSTED lt PARAM gt ADJUSTED QC lt PARAM gt ADJUSTED ERROR The variable PROFILE lt PARAM gt QC should be recomputed when lt PARAM gt ADJUSTED QC becomes available See Section 4 2 for definitions Here lt PARAM gt denotes all the measurement parameters that are reported in the netCDF file Currently lt PARAM gt PRES TEMP PSAL are the fundamental measurement parameters that are reported in every Argo netCDF file and have approved delayed mode qc procedures See Sections 3 2 3 3 3 4 on how to fill their related ADJUSTED variables For lt PARAM gt CNDC CNDC_ADJUSTED CNDC ADJUSTED QC and CNDC_ADJUSTED_ERROR can be their respective FillValues If they are not their respective Fill Values then CNDC_ADJUSTED must be calculated to be consistent with PSAL ADJUSTED TEMP ADJUSTED and PRES ADJUSTED CNDC ADJUSTED QC must be consistent with PSAL_ADJUSTED_QC and CNDC ADJUSTED ERROR must be consistent with PSAL ADJUSTED ERROR Some Argo netCDF files report DOXY There is currently no approved method for delayed mode qc on DOXY Therefore DOXY ADJUSTED original values recorded in DOXY DOXY ADJUSTED QC 0 DOXY_ADJUSTED_ERROR FillValue and PROFILE DOXY QC i e Blank the FillValue for PROFILE DOXY QC 3 5 2 Scientific calibration information for each profile Within each single cycle Argo netcdf profile file is a scientific calibration section t
35. ervation latitude and longitude from the float be located in an ocean Use can be made of any file that allows an automatic test to see if data are located on land We suggest use of at least the 5 minute bathymetry file that is generally available This is commonly called ETOPOS TerrainBase and can be downloaded from http www ngdc noaa gov mgg global global html Action If the data cannot be located in an ocean the position should be flagged as bad data and they should not be distributed on the GTS Argo data management quality control manual version 2 9 1 ol 5 Impossible speed test Drift speeds for floats can be generated given the positions and times of the floats when they are at the surface and between profiles In all cases we would not expect the drift speed to exceed 3 m s If it does it means either a position or time is bad data or a float is mislabeled Using the multiple positions that are normally available for a float while at the surface it is often possible to isolate the one position or time that is in error Action If an acceptable position and time can be used from the available suite then the data can be sent to the GTS Otherwise flag the position the time or both as bad data and no data should be sent to the GTS 6 Global range test This test applies a gross filter on observed values for pressure temperature salinity and dissolved oxygen It needs to accommodate all of the expected extremes encou
36. es The continuous valid zero reading period needs to span at least 6 months preferably longer This captures the microleakers whose oil leak rates are fastest and allows for seasonal variability from half of an annual cycle when surface pressure values may read just below zero For floats whose useful life is less than 6 months or when the continuous valid zero reading period is shorter than 6 months the qualifying time span is at the PI s discretion Examples a to d below illustrate some cases that should or should not be classified as TNPD Please note that in all of the following schematic examples SP represents valid and despiked values after removing the artificial 5 dbar This feature was corrected in the Apf 9 and later versions of the controller Argo data management quality control manual version 2 9 1 Example a 100 of the time series is TNPD TNPD if at least 6 months preferably longer If shorter than 6 months PI decision SP dbar p TNPD status begins Example b There are occasional valid positive SP readings in the first part of the time series followed by a continuous zero reading period that does not contain any occasional valid positive readings SP dbar Occasional valid positive readings TSS TNPD if at least 6 months preferably longer If shorter than 6 months PI decision TNPD status begins Example c The time series
37. etected by the real time qc tests and b vertical profiles that have the wrong shape Bad data points identified by visual inspection from delayed mode analysts are recorded with TEMP ADJUSTED QC 4 Please note that whenever PARAM ADJUSTED QC 4 PARAM ADJUSTED FillValue and PARAM ADJUSTED ERROR FillValue TEMP ADJUSTED TEMP ADJUSTED ERROR and TEMP ADJUSTED QC should be filled even when the data are good and no adjustment is needed In these cases TEMP ADJUSTED ERROR can be the manufacturer s quoted accuracy at deployment Please use the SCIENTIFIC CALIBRATION section in the netCDF files to record details of the delayed mode adjustment Argo data management quality control manual version 2 9 1 T m 3 4 Delayed mode procedures for salinity 3 4 1 Introduction Delayed mode qc for PSAL described in this section are specifically for checking sensor drifts and offsets Analysts should be aware that there are other instrument errors e g conductivity cell thermal mass error see Johnson et al 2007 contact Gregory C Johnson noaa gov for the related adjustment software and should attempt to identify and adjust them in delayed mode It is recommended that float salinity be adjusted for pressure offset and cell thermal mass error before sensor drift adjustment If a measurement has been adjusted for more than one instrument error analysts should attempt to propagate the uncertainties from all the adjustments The f
38. g distributed on the GTS Red Sea e Temperature in range 21 7 to 40 0 C e Salinity in range 2 to 41 0 PSU Mediterranean Sea e Temperature in range 10 0 to 40 0 C e Salinity in range 2 to 40 0 PSU Argo data management quality control manual version 2 9 1 T m 20 Questionable Argos position test For floats that use the Argos system to obtain position data this test can be used in lieu of Test 5 Impossible Speed Test This test identifies questionable Argos position data collected during surface drift of a float cycle by considering the float speed at the sea surface and Argos position errors Details of the method can be found in Nakamura et al 2008 Quality control method of Argo float position data JAMSTEC Report of Research and Development Vol 7 11 18 A brief description of the procedure is summarized here A Collect all Argos positions during surface drift of a float cycle The distance between two positions A and B is referred to as a segment A segment is considered questionable if i the float speed along the segment exceeds 3 ms and 11 the length of the segment is longer than the critical error length defined as 1 0x Er Er where Er and Erg are the radii of position error of the Argos system 150m 350m and 1000m for Argos class 3 2 and 1 respectively at 4 and B respectively B Ifa segment is not considered questionable then both positions 4 and B are good C If a segment is c
39. hat records details of delayed mode adjustments It is compulsory to fill the variables in the scientific calibration section at the completion of delayed mode qc In the scientific calibration section every measurement parameter recorded in the netCDF file should be listed in the variable PARAMETER For every measurement parameter listed in PARAMETER e g PRES TEMP PSAL CNDC DOXY there are four variables to record scientific calibration details SCIENTIFIC_CALIB EQUATION SCIENTIFIC CALIB COEFFICIENT SCIENTIFIC CALIB COMMENT SCIENTIFIC CALIB DATE Argo data management quality control manual version 2 9 1 T m In cases where no adjustment has been made SCIENTIFIC CALIB EQUATION and SCIENTIFIC CALIB COEFFICIENT shall be filled by their respective FillValues SCIENTIFIC CALIB COMMENT shall contain wordings that describe the evaluation E g 1 No adjustment is needed because no significant sensor drift has been detected E g 2 No approved method for delayed mode qc on DOXY is available In cases where adjustments have been made the PI is free to use any wordings he she prefers Please be precise and informative Examples of wordings for PSAL can be SCIENTIFIC CALIB EQUATION PSAL ADJUSTED PSAL AS where AS is calculated from a potential conductivity ref to O dbar multiplicative adjustment term r SCIENTIFIC CALIB COEFFICIENT r 0 9994 0 0001 vertically averaged AS 0 025 0 003 S
40. he computation should be taken from lt PARAM gt ADJUSTED QC if available and from lt PARAM gt QC otherwise n Meaning No QC was performed LA N 100 All profile levels contain good data 5 lt N lt 100 C 50 lt N lt 75 5 lt N lt 50 E o lt N lt 25 FIN 0 No profile levels have good data Argo data management quality control manual version 2 9 1 ST 4 3 Common instrument errors and failure modes This section describes some common instrument errors and failure modes that will cause error in float measurements 1 TBTO leakage TBTO tributyltinoxide is a wide spectrum poison that is used to protect conductivity cells from biofouling However accidental leakage of TBTO onto the conductivity cells can occur but it usually gets washed off in time This will result in fresh salinity offsets in float data that gradually return to normal Delayed mode analysts should pay special attention to salinity data at the beginning of the float s life because TBTO leakage usually affects the first few CTD profiles after deployment before the contamination gets washed off 2 Pollution events Any pollution on the conductivity cell will result in erroneously fresh salinity measurements When pollution washes off reversal of sensor drift trend can occur Delayed mode analysts need to be careful in splitting the float time series in such cases 3 Ablation events Any ablation of the conductivity
41. he separate segments join Adjustment continuity between separate float segments can be achieved by making adjustment in the transition phase even though the adjustment is below the 2 times error threshold limit In the following example the float experiences salinity sensor drift after a stable period The time series has been split for calibration However the time series has no discontinuity so the final assembled adjustment should be continuous Adjustment continuity is achieved by using model a not b a b atl time time 3 4 5 Assigning adjusted salinity error estimates and qc flags After evaluating all available information the PI then assigns adjusted salinity values error estimates and delayed mode qc flags In Argo data files these are found respectively in the variables PSAL ADJUSTED PSAL ADJUSTED QC and PSAL ADJUSTED ERROR Please refer to Section 4 1 for definitions of the Argo qc flags in delayed mode Several Matlab based graphical user interface softwares are available for interacting with Argo netcdf files For examples of these softwares please contact John Gilson jgilson ucsd edu or Pelle Robbins probbins whoi edu Argo data management quality control manual version 2 9 1 o n When LATITUDE LONGITUDE JULD are missing operators should fill the missing values with interpolated x y and t wherever possible and record POSITION QC 8 JULD QC 8 The prof
42. hile unpumped temperature data from the primary CTD are of good quality unpumped salinity data from the primary CTD are of dubious quality because the flow rate through the conductivity cell is wrong when the pump is switched off This test specifies that unpumped or partially pumped salinity data returned by the primary CTD should be flagged as probably bad data in real time Action Unpumped or partially pumped salinity data returned by the primary CTD are flagged as probably bad data in real time That is PSAL QC 3 a PROVOR ARVOR Data returned by PROVOR ARVOR floats are bin averaged and are not separated into pumped and unpumped types This separation is done during data processing at the DACs by checking when the CTD pump is switched off In addition for the older versions of PROVOR ARVOR floats some bin averaged data can contain a mixture of pumped and unpumped measurements if the pump cut off pressure falls in the middle of the bin Therefore near surface data from PROVOR ARVOR floats are identified as unpumped or partially pumped when PRES lt Pcutoff Poverlap Poverlap bin size 2 for PROVOR floats with software versions lt 5816400 and ARVOR floats with software versions lt 5605A00 Poverlap 0 5 dbar for PROVOR floats with software versions gt 5816A00 and ARVOR floats with software versions gt 5605A00 Pcutoff is the pressure at which the CTD pump is switched off This is stored in the
43. ile can then be evaluated and adjusted if necessary by using the interpolated x y t The ADJUSTED fields can then be filled accordingly The following is a set of guidelines for assigning values to PSAL ADJUSTED PSAL ADJUSTED QC and PSAL ADJUSTED ERROR in Argo data files For float salinity that are considered good and need no adjustment in delayed mode These are measurements that are considered to be unaffected by any instrument errors such as sensor drift or calibration offset For these good salinity data the recommended value for PSAL ADJUSTED ERROR is 0 01 or higher if statistical uncertainty exceeds 0 01 PSAL ADJUSTED PSAL original value PSAL ADJUSTED QC 1 or 2 PSAL ADJUSTED ERROR maximum statistical uncertainty 0 01 Note that a subset of these good salinity data is selected for inclusion in an Argo based reference database In cases where a float is considered to need no salinity adjustment in delayed mode but that the data may not be suitable for use as reference because statistical uncertainty is high PSAL ADJUSTED ERROR should increase to gt 0 015 Please refer to Section 4 5 for the list of selection criteria used in compiling the Argo reference database For float salinity that are considered bad and unadjustable in delayed mode For example large spikes or extreme behaviour where the relative vertical T S shape does not match good data These measurements are bad and unadjustable PSAL ADJU
44. ing Real time comment Delayed mode comment NO OG was No QC was performed No QC was performed performed The adjusted value is statistically consistent and a statistical error estimate is supplied Test 15 or Test 16 or Test 17 failed and all other real time QC tests passed These data are not to be used without scientific correction A flag 3 may be assigned by an operator during additional visual QC for bad data that may be corrected in delayed mode Probably bad data that are potentially An adjustment has been applied but the value may correctable still be bad Data have failed one or more of the real time QC tests excluding Test 16 A flag 4 may be Bad data Not adjustable assigned by an operator during additional visual Data replaced by FillValue QC for bad data that are uncorrectable 5 Value changed Not used Not used Not used 7 Notused ___Notused 5 5 Nouse o O o Missigvaue Missing vaus Missing value Bad data Argo data management quality control manual version 2 9 1 T 4 2 Reference Table 2a profile quality flags Please note that this table is used for all measured parameters This table is named Reference Table 2a in the Argo User s Manual N is defined as the percentage of levels with good data where QC flag values of 1 2 5 or 8 are GOOD data QC flag values of 9 missing are NOT USED in the computation e All other QC flag values are BAD data T
45. l for dissolved oxygen DOXY The real time QC tests 6 9 11 and 13 are applied to DOXY 04 11 2013 2 5 addition of real time quality control test procedures for near surface data 04 11 2013 3 4 5 added upper limit of delayed mode salinity adjustment as 0 05 PSS 78 rem Argo data management quality control manual version 2 9 1 1 1 Introduction This document is the Argo quality control manual Changes from the previous version of the manual are highlighted in yellow The Argo data system has three levels of quality control e The first level is the real time system that performs a set of agreed automatic checks on all float measurements Real time data with assigned quality flags are available to users within the 24 48 hrs timeframe e The second level of quality control is the delayed mode system e The third level of quality control is regional scientific analyses of all float data with other available data The procedures for regional analyses are still to be determined This document contains the description of the Argo real time and delayed mode procedures Please note that at the present time quality control procedures exist only for the parameters JULD LATITUDE LONGITUDE PRES TEMP PSAL and DOXY Argo data management quality control manual version 2 9 1 s 2 Real time quality controls 2 1 Argo Real time Quality Control Test Procedures on vertical profiles 2 1 1 Introduction Because of the req
46. lue exceeds 0 9 PSU for pressures less than 500 dbar or e the test value exceeds 0 3 PSU for pressures greater than or equal to 500 dbar Dissolved oxygen The V2 value is flagged when e the test value exceeds 50 micromol kg for pressures less than 500 dbar or e the test value exceeds 25 micromol kg for pressures greater than or equal to 500 dbar Action Values that fail the spike test should be flagged as bad data and are removed from the TESAC distributed on the GTS If temperature and salinity values at the same depth both fail they should be flagged as bad data and the values for depth temperature and salinity should be removed from the TESAC being distributed on the GTS 10 Top and bottom spike test obsolete 11 Gradient test This test is failed when the difference between vertically adjacent measurements is too steep The test does not consider differences in depth but assumes a sampling that adequately reproduces changes in temperature salinity and dissolved oxygen with depth The algorithm is used on vertical profiles of temperature salinity and dissolved oxygen Argo data management quality control manual version 2 9 1 Do ad Test value V2 V3 V1 2 where V2 is the measurement being tested as a spike and V1 and V3 are the values above and below Temperature The V2 value is flagged when e the test value exceeds 9 0 C for pressures less than 500 dbar or e the test value exceeds 3 0 C for pressures g
47. m to top if the potential density calculated at the lesser pressure is greater than that calculated at the greater pressure by more than 0 03 kg m both the temperature and salinity values should be flagged as bad data Bad temperature and salinity values should be removed from the TESAC distributed on the GTS 15 Grey list This test is implemented to stop the real time dissemination on the GTS of measurements from a sensor that is not working correctly The grey list contains the following 7 items e Float WMO Id e Parameter name of the grey listed parameter e Start date from that date all measurements for this parameter are flagged as bad or probably bad e End date from that date measurements are not flagged as bad or probably bad e Flag value of the flag to be applied to all measurements of the parameter e Comment comment from the PI on the problem e DAC data assembly center for this float Example Float WMO ld Parameter Start date End date Flag Comment DAC 1900206 esa 20030005 3 Tr Each DAC manages a grey list sent to the GDACSs The merged grey list is available from the GDACs e Grey list format ascii csv comma separated values e Naming convention xxx greylist csv xxx DAC name e g aoml greylist csv coriolis greylist csv jma greylist csv e PLATFORM PARAMETER START DATE END DATE QC COMMENT DAC e g 4900228 TEMP 20030909 3 AO e g 1900206 PSAL 20030925 3 IF
48. meta data variable CONFIG _CTDPumpStopPressure dBAR The sum Pcutoff Poverlap is stored in the meta data variable CONFIG CTDPumpsStopPressurePlusThreshold dBAR b NOVA Some NOVA floats collect near surface data by using the primary CTD in the unpumped mode with the CTD pump switched off at 4 dbar Near surface data from NOVA floats are bin averaged and are identified as unpumped when PRES lt 4 dbar c APEX Test 21 does not apply to near surface data returned by APEX floats APEX Argos floats with the NST firmware do not report unpumped salinity from the primary CTD APEX Iridium floats equipped with the STS module return unpumped salinity from the auxiliary module The STS module is designed to measure salinity in a free flushing mode and so is different from the primary CTD Argo data management quality control manual version 2 9 1 o 22 Near surface mixed air water test Most near surface profiles extend all the way to the sea surface Therefore the shallowest part of a near surface profile will contain some mixed air water measurements This test identifies broadly the pressures at which this shallowest part of a near surface profile takes place and specifies that data in that pressure range are probably bad data Action Data from the shallowest part of a near surface profile which may contain mixed air water measurements are flagged as probably bad data in real time a PROVOR ARVOR F
49. meter if temperature is flagged as bad then salinity is flagged as bad This change was decided during ADMT8 in Hobart 82 2 test 6 minimum salinity set to 2 PSU instead of O PSU 82 2 test 7 minimum salinity set to 2 PSU instead of 0 PSU 04 11 2008 82 1 1 SigmaO0 specified in density inversion test 82 3 1 added a section on Real time pressure adjustment for APEX floats 83 3 5 updated Delayed mode section based on DMQC 3 Workshop 19 10 2009 82 1 2 amp 82 2 test 6 minimum P set to 5 dbar 83 1 added a section on Editing raw qc flags in delayed mode 83 2 2 updated delayed mode treatment for APEX TNPDs after DMQC 4 15 7 2010 82 4 added a section on Feedback from Statistical Test at Coriolis 05 11 2010 83 2 2 revised definition for TNPD after ADMT11 in Hamburg Argo data management quality control manual version 2 9 1 Date Comment dd mny yyyy 01 12 2011 82 1 2 added threshold of 0 03 kg m to RT Test 14 Density Inversion Test for profile data following ADMT 12 in Seoul Test to use potential density referenced to mid point pressure between the two levels to be compared 82 2 Added RT Test 20 Questionable Argos position test from JAMSTEC as a new real time qc test for trajectory data following 3 Trajectory Workshop and ADMT12 in Seoul 03 01 2013 2 3 1 clarified real time pressure adjustment for non auto correcting floats when SP 0 03 01 2013 2 1 2 addition of real time quality contro
50. more than 10 000 profiles only select profiles that are post 1995 6 Eliminate nearby duplicates T Do objective residual analysis using previously qc d reference data to identify anomalies Then do visual inspection of anomalies 8 Identify each reference profile with a unique ID e g under the variable SOURCE It is recommended that in regions with adequate reference data that delayed mode qc for salinity should use CTD data only If CTD data are too sparse bottle data BOT may be included Argo data management quality control manual version 2 9 1 SST 4 5 Criteria for Argo profiles to be retained in the reference database The following criteria are used to select Argo data as reference for delayed mode quality control of Argo salinity profiles in the open ocean 1 No real time data 2 No floats that fail within 1 year of deployment 3 No cycles within 6 months of end of record 4 No cycles that have salinity drift adjustment AS gt 001 PSS 78 in bottom data to distinguish from thermal lag adjustment at shallower levels 5 No floats whose deepest sampling level is shallower than 800 dbar 6 No cycles following ones that have salinity drift adjustment AS gt 001 PSS 78 in bottom data 7 No cycles where less than 90 of P T S values are good PRES ADJUSTED QC 1 TEMP ADJUSTED QC 1 PSAL ADJUSTED QC 1 8 No cycles 18 first 6 months to be used due to th
51. nagement quality control manual version 2 9 1 n 8 Pressure increasing test This test requires that the profile has pressures that are monotonically increasing assuming the pressures are ordered from smallest to largest Action If there is a region of constant pressure all but the first of a consecutive set of constant pressures should be flagged as bad data If there is a region where pressure reverses all of the pressures in the reversed part of the profile should be flagged as bad data All pressures flagged as bad data and all of the associated temperatures and salinities are removed from the TESAC distributed on the GTS 9 Spike test The difference between sequential measurements where one measurement is significantly different from adjacent ones is a spike in both size and gradient This test does not consider differences in depth but assumes a sampling that adequately reproduces changes in temperature salinity and dissolved oxygen with depth The algorithm is used on vertical profiles of temperature salinity and dissolved oxygen Test value V2 V3 V1y2 V3 VD 2 where V2 is the measurement being tested as a spike and V1 and V3 are the values above and below Temperature The V2 value is flagged when e the test value exceeds 6 0 C for pressures less than 500 dbar or e the test value exceeds 2 0 C for pressures greater than or equal to 500 dbar Salinity The V2 value is flagged when e the test va
52. nce to oceanographic features such as eddies and rings that can introduce complications to the semi automatic methods nspecting contour plots of float salinity anomaly time series Systematic sensor malfunction should show up as salinity anomalies over several water masses Using other independent oceanographic atlases to anticipate water mass changes along the float trajectory that can be misinterpreted as sensor malfunction nspecting residuals from objective maps Cross checking with nearby stable floats in cases of suspect sensor calibration offset Argo data management quality control manual version 2 9 1 e If the PI is confident that sensor malfunction has occurred then the recommended threshold for making an adjustment is when AS is greater than 2 times the error from the statistical methods but the PI can provide an alternative estimate of uncertainty if they have a basis for doing so Note that this guideline is to help the PI in deciding whether a slope or offset is statistically significant and so should be used to evaluate the entire float segment being fitted and not to single points In cases where the time series has been split into separate segments the PI must ensure that the assembled adjustment for the entire time series is continuous within error bars except where the PI believes there is a genuine discontinuity see Step 5 in Section 3 4 3 This is to ensure that no artificial jump is introduced where t
53. ntered in the oceans e Pressure cannot be less than 5 dbar e Temperature in range 2 5 to 40 0 C e Salinity in range 2 to 41 0 PSU e Dissolved oxygen in range 0 5 to 600 micromol kg Action If a value fails it should be flagged as bad data and only that value need be removed from distribution on the GTS If temperature and salinity values at the same depth both fail both values should be flagged as bad data and values for depth temperature and salinity should be removed from the TESAC distributed on the GTS 7 Regional range test This test applies to only certain regions of the world where conditions can be further qualified In this case specific ranges for observations from the Mediterranean and Red Seas further restrict what are considered sensible values The Red Sea is defined by the region 10N 40E 20N 50E 30N 30E 10N 40E and the Mediterranean Sea by the region 30N 6W 30N 40E 40N 35E 42N 20E 50N 15E 40N 5E 30N 6W Action Individual values that fail these ranges should be flagged as bad data and removed from the TESAC being distributed on the GTS If both temperature and salinity values at the same depth both fail then values for depth temperature and salinity should be removed from the TESAC being distributed on the GTS Red Sea e Temperature in range 21 7 to 40 0 C e Salinity in range 2 to 41 0 PSU Mediterranean Sea e Temperature in range 10 0 to 40 0 C e Salinity in range 2 to 40 0 PSU Argo data ma
54. ol manual Frozen profile real time qc test 17 proposed at ADMT5 in Southampton 3 Deepest pressure real time qc test 18 proposed at ADMT5 in Southampton 4 Order list for the real time qc tests 5 Regional Global Parameter Test renamed Regional range test test 7 6 Grey list naming convention and format test 15 7 Real time qc on trajectories 4 delayed mode quality control manual from Annie Wong 17 11 2005 2 3 added a section on Real time salinity adjustment 3 1 added usage of SURFACE PRESSURE from APEX floats 83 3 5 added some more guidelines for PSAL ADJUSTED QC 2 83 3 8 clarified that PROFILE lt PARAM gt QC should be recomputed when lt PARAM gt ADJUSTED QC becomes available 82 1 2 test 19 deepest pressure delta set to 10 82 1 2 test 14 density inversion test applied downward and upward 14 11 2007 82 1 2 test 6 minimum salinity set to 2 PSU instead of 0 PSU 82 1 2 test 7 minimum salinity set to 2 PSU instead of 0 PSU This change was decided during ADMT8 in Hobart 83 3 1 use known pressure drift instead of delta P gt 5dbar This change was decided during AST8 meeting in Paris 83 3 2 delayed mode operators can edit real time QC flags This change was decided during ADMT8 in Hobart 81 2 4 values with QC flag 4 are ignored by quality control tests This change was decided during ADMT8 in Hobart 14 11 2007 82 1 4 when salinity is calculated from the conductivity para
55. onsidered questionable then ifthe Argos class at 4 and B are different then the position with the less accurate Argos class is flagged as 3 ifthe Argos class at 4 and B are the same and there is one good position before and one good position after A and B i e there are 4 positions for the check then the position that gives the higher speed along the segment from the previous good position to the later good position is flagged as 3 ifthe Argos class at A and B are the same and there is one good position either before or after 4 and B i e there are 3 positions for the check then the position that gives the higher speed along the segment either from the previous good position or to the later good position is flagged as 3 ifthe Argos class at A and B are the same but there are no other good positions around A and B i e there are 2 positions for the check then both 4 and B are flagged as 3 Argo data management quality control manual version 2 9 1 2 2 3 Argo Real time Adjustments on vertical profiles 2 3 1 Real time pressure adjustment for non auto correcting floats There are many float types in Argo each has its own way of treating the pressure measurements For example while PROVOR and SOLO floats internally correct for pressure offsets APEX floats do not make any internal pressure correction All Argo float types and their treatment of pressure are summarized in the Surface Pressu
56. or PROVOR ARVOR floats if the first bin closest to the sea surface has PRES lt dbar then the temperature value from that first bin is suspected to contain averages of mixed air water measurements and should be flagged as probably bad data That is TEMP QC E b NOVA For NOVA floats if the first bin closest to the sea surface has PRES lt 1 dbar then the temperature value from that first bin is suspected to contain averages of mixed air water measurements and should be flagged as probably bad data That is TEMP QC 3 c APEX The shallowest part of near surface profiles collected by APEX floats will contain pressure readings that are not necessarily monotonic with time For APEX Argos floats with the NST firmware when near surface data have pressures shallower than 5 dbar check difference in pressure between two successive measurements If the difference is less than 0 5 dbar from the previous measurement then data from that level and all levels after that should be flagged as probably bad data That is PRES QC 3 TEMP QC 3 For APEX Iridium floats equipped with the STS module when near surface data have pressures shallower than 5 dbar check that pressure readings decrease monotonically with time Data from the level when monotonicity stops and all levels after that should be flagged as probably bad data That is PRES QC 3 TEMP QC 3 PSAL QC 3 Argo dat
57. re Correction Table which can be found at http www argodatamgt org Documentation Floats that do not adjust their pressure data on board before data telemetry are referred to as non auto correcting floats Pressure measurements from non auto correcting floats need to be treated in real time whenever valid surface pressure values are available The section below details the real time pressure adjustment procedures for APEX floats which is the largest group of non auto correcting floats in Argo Real time pressure adjustment for APEX floats APEX floats return raw pressures which are stored in the variable PRES in the Argo netCDF files Pressure adjustment should be applied in real time to all APEX floats by using SURFACE PRESSURE SP values returned by the APEX floats The SP measurement is taken while the float is at the sea surface just before descent and hence is different from the shallowest measured pressure in the vertical profile which is taken on ascent and while the float is beneath the sea surface These SP values are stored in the Argo technical files in the variable PRES SurfaceOffsetTruncatedPlus5dbar dBAR or PRES SurfaceOffsetNotTruncated dBAR depending on the type of APEX controller used Subtract 5 dbar from the SP values in PRES SurfaceOffsetTruncatedPlus5dbar dBAR SP values in PRES SurfaceOffsetNotTruncated dBAR are used as they are without needing to subtract 5 dbar Then erroneous outliers in SP nee
58. reater than or equal to 500 dbar Salinity The V2 value is flagged when e the test value exceeds 1 5 PSU for pressures less than 500 dbar or e the test value exceeds 0 5 PSU for pressures greater than or equal to 500 dbar Dissolved oxygen The V2 value is flagged when e the test value exceeds 50 micromol kg for pressures less than 500 dbar or e the test value exceeds 25 micromol kg for pressures greater than or equal to 500 dbar Action Values that fail the test i e value V2 should be flagged as bad data and are removed from the TESAC distributed on the GTS If temperature and salinity values at the same depth both fail both should be flagged as bad data and then values for depth temperature and salinity should be removed from the TESAC distributed on the GTS 12 Digit rollover test Only so many bits are allowed to store temperature and salinity values in a profiling float This range is not always large enough to accommodate conditions that are encountered in the ocean When the range is exceeded stored values rollover to the lower end of the range This rollover should be detected and compensated for when profiles are constructed from the data stream from the float This test is used to make sure the rollover is properly detected e Temperature difference between adjacent depths gt 10 C e Salinity difference between adjacent depths gt 5 PSU Action Values that fail the test should be flagged as bad data and are removed from
59. ree moving nature of profiling floats means that most float salinity measurements are without accompanying in situ ground truth values for absolute calibration such as those afforded by shipboard CTD measurements Therefore Argo delayed mode procedures for checking sensor drifts and offsets in salinity rely on reference datasets and statistical methods However since the ocean has inherent spatial and temporal variabilities these drift and offset adjustments are subject to statistical uncertainties Users therefore should include the supplied error estimates in their usage of Argo delayed mode salinity data Three methods are available for detecting sensor drifts and offsets in float salinity and for calculating adjustment estimates and related uncertainties 1 Wong Johnson Owens 2003 estimates background salinity on a set of fixed standard isotherms then calculates drifts and offsets by time varying weighted least squares fits between vertically interpolated float salinity and estimated background salinity This method suits float data from open tropical and subtropical oceans For the related software please contact Annie Wong at awong ocean washington edu 2 Boehme and Send 2005 takes into account planetary vorticity in its estimates of background salinity and chooses a set of desirable isotherms for calculations This method suits float data from oceans with high spatial and temporal variabilities where multiple water masses e
60. s For floats where salinity PSAL is calculated from the temperature TEMP and conductivity CNDC parameters if temperature is flagged 4 or 3 then salinity is flagged 4 or 3 Argo data management quality control manual version 2 9 1 ET 2 2 Argo Real time Quality Control Test Procedures on trajectories The following tests are applied in real time on trajectory data 1 Platform identification Every centre handling float data and posting them to the GTS will need to prepare a metadata file for each float and in this is the WMO number that corresponds to each float ptt There is no reason why except because of a mistake an unknown float ID should appear on the GTS Action If the correspondence between the float ptt cannot be matched to the correct WMO number none of the data from the profile should be distributed on the GTS 2 Impossible date test The test requires that the observation date and time from the float be sensible e Year greater than 1997 e Month in range 1 to 12 e Day in range expected for month e Hour in range 0 to 23 e Minute in range 0 to 59 Action If any one of the conditions is failed the date should be flagged as bad data and none of the data from the profile should be distributed on the GTS 3 Impossible location test The test requires that the observation latitude and longitude from the float be sensible Action If either latitude or longitude fails the position should be flagged a
61. s bad data and none of the data from the float should go out on the GTS e Latitude in range 90 to 90 e Longitude in range 180 to 180 4 Position on land test The test requires that the observation latitude and longitude from the float be located in an ocean Use can be made of any file that allows an automatic test to see if data are located on land We suggest use of at least the 5 minute bathymetry file that is generally available This is commonly called ETOPOS TerrainBase and can be downloaded from http www ngdc noaa gov mgg global global html Action If the data cannot be located in an ocean the position should be flagged as bad data and they should not be distributed on the GTS Argo data management quality control manual version 2 9 1 S Sm 5 Impossible speed test Drift speeds for floats can be generated given the positions and times of the floats when they are at the surface and between profiles In all cases we would not expect the drift speed to exceed 3 m s If it does it means either a position or time is bad data or a float is mislabeled Using the multiple positions that are normally available for a float while at the surface it is often possible to isolate the one position or time that is in error Action If an acceptable position and time can be used from the available suite then the data can be sent to the GTS Otherwise flag the position the time or both as bad data and no data should be sent to
62. starts with continuous valid positive readings then becomes continuously zero with no occasional valid positive readings SP dbar adjustable _ _ TST TNPD if at least 6 months preferably longer If shorter than 6 months PI decision TNPD status begins Argo data management quality control manual version 2 9 1 3 1 oo Example d The time series starts with a continuous zero reading period then reverts back to valid positive values The initial zero reading period does not qualify as TNPD This is because pressure drifts are typically monotonic and therefore a reversal back to positive values indicates that the pressure sensor is not likely to have developed a negative drift adjustable SP dbar Unadjustable but not TNPD After determining which part of the time series qualifies as TNPD the delayed mode operator should then determine the probability of the TNPD data being affected by the Druck oil microleak problem According to SeaBird the date of manufacturing change at Druck that led to the oil microleak defect occurred sometime in mid 2006 The microleak failure rate jumped from 3 before 2006 to 30 in 2007 Any Druck pressure sensor with serial number greater than 2324175 falls into the group that has 30 likelihood of being affected by oil microleaks Cross checking between the various APEX groups within Argo indicated that deployment of floats with Druck serial number greater than 232
63. ter consultation with the PI When available real time adjusted values are distributed to the GTS instead of the raw values 2 3 2 Real time salinity adjustment When delayed mode salinity adjustment see Section 3 3 becomes available for a float real time data assembly centres will extract the adjustment from the latest D nc file as an additive constant and apply it to new salinity profiles If a better correction is available in real time DACs can use that instead In this manner intermediate quality salinity profiles will be available to users in real time The values of this real time adjustment will be recorded in PSAL ADJUSTED PSAL ADJUSTED QC will be filled with the same values as PSAL QC PSAL ADJUSTED ERROR and all variables in the SCIENTIFIC CALIBRATION section of the netCDF files will be filled with FillValue DATA MODE will record A When available real time adjusted values are distributed to the GTS instead of the raw values 2 3 3 Real time files with DATA_MODE A When real time files have DATA MODE A it means real time adjustments are available for one or more parameters All PARAM ADJUSTED variables should therefore be filled where PARAM PRES TEMP PSAL CNDC DOXY etc PARAM ADJUSTED real time adjusted values or PARAM if no real time adjustment is available PARAM ADJUSTED QC PARAM QC PARAM ADJUSTED ERROR FillValue Users should be aware that even though the ADJUSTED
64. the TESAC distributed on the GTS If temperature and salinity values at the same depth both fail both values should be flagged as bad data and then values for depth temperature and salinity should be removed from the TESAC distributed on the GTS 13 Stuck value test This test looks for all measurements of temperature salinity and dissolved oxygen in a vertical profile being identical Action If this occurs all of the values of the affected variable should be flagged as bad data and are removed from the TESAC distributed on the GTS If temperature and salinity are affected all observed values are flagged as bad data and no report from this float should be sent to the GTS Argo data management quality control manual version 2 9 1 ooo 14 Density inversion This test compares potential density between valid measurements in a profile in both directions 1 e from top to bottom and from bottom to top Values of temperature and salinity at the same pressure level P should be used to compute potential density p or o p 1000 kg m referenced to the mid point between P and the next valid pressure level P A threshold of 0 03 kg m should be allowed for small density inversions Action From top to bottom if the potential density calculated at the greater pressure is less than that calculated at the lesser pressure by more than 0 03 kg m both the temperature and salinity values should be flagged as bad data From botto
65. the GTS Note Floats that use the Argos system to obtain position data can use Test 20 instead 6 Global range test This test applies a gross filter on observed values for pressure temperature and salinity It needs to accommodate all of the expected extremes encountered in the oceans e Pressure cannot be less than 5 dbar e Temperature in range 2 5 to 40 0 C e Salinity in range 2 to 41 0 PSU Action Ifa value fails it should be flagged as bad data and only that value need be removed from distribution on the GTS If temperature and salinity values at the same depth both fail both values should be flagged as bad data and values for depth temperature and salinity should be removed from the TESAC distributed on the GTS 7 Regional range test This test applies to only certain regions of the world where conditions can be further qualified In this case specific ranges for observations from the Mediterranean and Red Seas further restrict what are considered sensible values The Red Sea is defined by the region 10N 40E 20N 50E 30N 30E 10N 40E and the Mediterranean Sea by the region 30N 6W 30N 40E 40N 35E 42N 20E 50N 15E 40N 5E 30N 6W Action Individual values that fail these ranges should be flagged as bad data and removed from the TESAC being distributed on the GT S If both temperature and salinity values at the same depth both fail then values for depth temperature and salinity should be removed from the TESAC bein
66. tment is low PSAL ADJUSTED original value adjustment gt 0 05 PSAL ADJUSTED QC 2 PSAL ADJUSTED ERROR uncertainty provided by PI but has to be gt 0 015 The following are some cases where PSAL ADJUSTED QC 2 should be assigned e Adjustment is based on unsatisfactory reference database Adjustment is based on a short calibration window because of sensor behaviour transition or end of sensor life and therefore may not be stable Evaluation is based on insufficient information e Sensor is unstable e g magnitude of salinity adjustment is greater than 0 05 or sensor has undergone too many sensor behaviour changes and therefore data are inherently of mediocre quality When a float exhibits problems with its pressure measurements e g APEX TNPD Argo data management quality control manual version 2 9 1 r 3 4 6 Summary flowchart for salinity Argo salinity sensor drift amp offset QC procedures PSAL_ADJUSTED FillValue PSAL_ADJUSTED_ERROR FillValue PSAL ADJUSTED QC 4 PSAL ADJUSTED PSAL original value PSAL ADJUSTED ERROR max Fadjustment error2 2 0 01 or uncertainty provided by PI PSAL ADJUSTED QC 1 20r3 Argo data management PSAL_ADJUSTED value recommended by statistical analyses or value provided by PI PSAL ADJUSTED ERROR max Fadjustment error2 2 0 01 or uncertainty provided by PI PSAL ADJUSTED QC 1 20r3 OR PSAL AD
67. uirement for delivering data to users within 24 hours of the float reaching the surface the quality control procedures on the real time data are limited and automatic The test limits are briefly described here More detail on the tests can be found in IOC Manuals and Guides 22 or at http www meds sdmm dfo mpo gc ca ALPHAPRO gtspp gcmans MG22 guide22 e htm Note that some of the test limits used here and the resulting flags are different from what is described in IOC Manuals and Guides 22 If data from a float fail these tests those data will not be distributed on the GTS However all of the data including those having failed the tests should be converted to the appropriate netCDF format and forwarded to the Global Argo Servers Presently the TESAC code form is used to send the float data on the GTS see http www meds sdmm dfo mpo gc ca meds Prog Int J COMM J COMM e htm This code form only handles profile data and reports observations as a function of depth not pressure It is recommended that the UNESCO routines be used to convert pressure to depth Algorithms for computation of fundamental properties of seawater N P Fofonoff and R C Millard Jr UNESCO Technical Papers in Marine Science 744 1983 If the position of a profile is deemed wrong or the date is deemed wrong or the platform identification is in error then none of the data should be sent on the GTS For other failures only the offending values need be removed from th
68. ve pressure drift The delayed mode operator may append to the end of this character string any other comments regarding PRES that he she wishes to make For the portion of the time series that contains occasional valid positive SP readings Example b it is the PI s decision based on the frequency of occurrence of the valid positive SP readings on whether or not to adjust those profiles For the unadjustable but non TNPD data Example d and cases in Example b where the PI decides to not adjust any negative pressure offset is likely to be less than the manufacturer quoted accuracy of 2 4 dbar For these unadjustable but non TNPD data if no additional error is found then PRES ADJUSTED QC 1 TEMP ADJUSTED QC 1 PSAL ADJUSTED QC I PRES ADJUSTED ERROR 2 4 dbar Argo data management quality control manual version 2 9 1 3 3 2 3 Summary flowchart for unadjustable APEX pressures Procedures for processing unadjustable APEX pressure data in delayed mode Argo data management quality control manual version 2 9 1 6 T 3 3 Delayed mode procedures for temperature Delayed mode qc for TEMP is done by subjective assessment of vertical profile plots of TEMP vs PRES and PSAL vs TEMP This assessment should be done in relation to measurements from the same float as well as in relation to nearby floats and historical data The assessment should aim to identify a erroneous data points that cannot be d
69. xist on the same isotherm and where water mass distribution is affected by topographic barriers For the related software please contact Lars Boehme at 1b284 st andrews ac uk 3 Owens and Wong 2009 improves the objective mapping scheme of WJO based on the method suggested by BS and performs an optimal linear piecewise continuous fit in potential conductivity space This method suits float data from the global ocean For the related software please contact Breck Owens at bowens whoi edu or Annie Wong at awong ocean washington edu All three methods require an adequate reference database and an appropriate choice of spatial and temporal scales as well as input of good adjusted float pressure temperature position and date of sampling Therefore analysts should first check the reference database for adequacy and determine a set of appropriate spatial and temporal scales before using these methods Operators should also ensure that other float measurements PRES TEMP LATITUDE LONGITUDE JULD are accurate or adjusted before they input them into the statistical tools for estimating reference salinity See Sections 3 2 and 3 3 for delayed mode procedures for PRES and TEMP Argo data management quality control manual version 2 9 1 39 l 3 4 2 Quality control and the semi automatic part The real time qc procedures described in Section 2 issue a set of qc flags that warns users of the quality of float salinity These are found in the vari
70. yed mode procedures for pressure Delayed mode qc for PRES is done by subjective assessment of vertical profile plots of TEMP vs PRES and PSAL vs PRES This assessment should be done in relation to measurements from the same float as well as in relation to nearby floats and historical data The assessment should aim to identify a erroneous data points that cannot be detected by the real time qc tests and b vertical profiles that have the wrong shape Bad data points identified by visual inspection from delayed mode analysts are recorded with PRES ADJUSTED QC 4 For these bad data points TEMP ADJUSTED QC and PSAL ADJUSTED QC should also be set to 4 Please note that whenever PARAM ADJUSTED QC 4 both PARAM ADJUSTED and PARAM ADJUSTED ERROR should be set to FillValue 3 2 1 Delayed mode pressure adjustment for APEX floats Similar to the real time procedure pressures from APEX floats should be adjusted for offsets by using SURFACE PRESSURE SP values in delayed mode These SP values are stored in the Argo technical files in the variable PRES SurfaceOffsetNotTruncated dBAR or PRES SurfaceOffsetTruncatedPlus5dbar dBAR depending on the type of APEX controller used The SP time series is examined and treated in delayed mode as follows 1 Subtract 5 dbar from PRES SurfaceOffsetTruncatedPlusSdbar dBAR SP values in PRES SurfaceOffsetNotTruncated dBAR are used as they are without needing to subtract 5 dbar

Argo quality controls manual, real-time & delayed-mode

Contents

Download Pdf Manuals

Related Search

Related Contents

Argo quality controls manual, real-time &amp; delayed-mode

Contents

Download Pdf Manuals

Related Search

Related Contents

Argo quality controls manual, real-time & delayed-mode