Assessement of real-time QC procedure efficiency
1. The ARGO QC procedures
The Argo data system has three levels of Quality Control : the first level is the real-time system that performs a set of agreed checks on all float measurements. Real-time data with assigned quality flags are available to users within the 24-48 hrs timeframe ; the second level of quality control is the delayed-mode system ; the third level of quality control is regional scientific analyses of all float data with other available data.
Nonetheless in all these tests, the profiles are qualified individually (except for test 16 “Gross salinity or temperature sensor drift” where the previous profile of the same float is taken into account) and the experience has proven that individual profiles may look good while they are not coherent with the neighbours or far from climatology (i.e. pressure anomaly in a serie of Solo float in 2006-2007).
Therefore some consistency check procedures have been developed within Euro-Argo to assess the quality of the ARGO data in near real time in order to detect bad data that have not been flagged by real time quality control (RTQC) procedure. Such procedures when applied periodically, help improving the quality of the ARGO data for users before delay mode processing in applied on the data.
Some of these procedures have been developed or improved within Euro-Argo projects. Some are developed in collaboration with the MyOcean EU project.
2. Proposed Improvements
A few teams have been working on statistical methods that would look at the Argo data as a whole and would be able to detect profiles that are not coherent with either their neighbours , with other datasets either in-situ or satellite or with recent climatology.
2.1 - Comparison to a climatology :
Two methods have been developed to compare the global in-situ data to a reference climatology. These methods allow to detect gross errors and large deviations in the field which could be not detected with the methods during the RTQC.
- Objective Analysis and residual analysis
An objective analysis based on optimal estimation methods (Bretherton et al., 1976) is performed on the global ocean with all data available including Argo floats, XBT, CTD, drifters and fixed point stations. Residuals between the raw data and the gridded field are computed by the analysis and are then screened to detect gross errors (big differences between the fields). The resulting alerts are then checked visually. This method combines the advantage of a collocation method since it takes into account all neighbouring sensors, and the comparison with a climatology.
- Anomaly method
Anomalous Argo temperature and salinity profiles are detected by comparing all profiles against different climatologies, namely ARIVO (2003-2008, von Schuckmann et al., 2009), the World Ocean Atlas 2005 (1955-1998, WOA05, Locarnini et al., 2006; Antonov et al., 2006), as well as a gridded field of all the CORIOLIS data obtained during 3 months before the measurement. The profiles are alerted if the calculated bias from the anomalies is greater than 5 standard deviations of the ARIVO climatology. The method will finally produce an alert map showing the location of the profile and as well as a plot of the profile itself. This method is applied after the objective mapping method introduced in the section above. The difference of this method to the first one lies in the fact, that here an alert is given only if a minimum of 50% of the data points lie outside the defined ranges. With this method, deviations can be detected which are smaller compared to those alerted in the objective mapping method of the section above.
2.2 - Comparison with altimetry :
Satellite altimeter measurements are used to check the quality of the Argo profiling floats time series. The method is described in Guinehut et al., 2009 and compares collocated sea level anomalies from altimeter measurements and dynamic height anomalies calculated from Argo temperature and salinity profiles for each Argo float time series. Different kinds of anomalies (sensor drift, bias, spikes, etc.) have been identified on some real-time but also delayed-mode Argo floats. About 4% of the floats should probably not be used until they are carefully checked and reprocessed by the principal investigators (PIs). The method appears to be very complementary to the existing quality control checks performed in real time or delayed mode. It could also be used to quantify the impact of the adjustments made in delayed mode on the pressure, temperature, and salinity fields.
For this study, when available, delayed-mode fields are preferred to realtime ones and only measurements having pressure, temperature, and salinity observations considered ‘‘good’’ (i.e., with a quality flag numerical grade of ‘‘1’’) are used. The altimeter data used are Archiving, Validation, and Interpretation of Satellite Oceanographic data (AVISO) combined products, which provide maps of SLA obtained from an optimal combination of all available satellite altimeters (AVISO 2008). Dynamic heights relative to 900-m depth are first calculated from Argo pressure, temperature, and salinity profiles. The 900-m depth is chosen to keep the maximum of profiles in the open ocean, as
many floats do not profile deeper than 1000 m, particularly at low latitudes due to technical limitations. If the reference depth is taken to be at 1000-m depth, the number of profiles is reduced by 16.6%. Furthermore, the method could be easily extended to other reference levels in order to check the quality of the subset of floats that are drifting at lower levels, like the Mediterranean ones. Next, to calculate dynamic height anomalies consistent with altimeter SLA, a contemporaneous Argo climatology is used as the mean dynamic height. To compare SLA and DHA, SLA maps are then interpolated to the time and location of each in situ DHA measurement using a linear space/time interpolation. Finally, general statistics between the two datasets (correlation coefficient, rms of the differences) are generated for each Argo float time series. They are then compared to a priori knowledge of these statistics. Finally, those profiles are alerted where rms(SLA-DHA)/rms(sla)*100 < 30%.
2.3 - Comparison with Argo neighbours :
A new method has been tested to systematically compare floats with their neighbors in space and time. We use first methods used for delayed mode data processing and adapt them for a real-time context.
The objective was to implement a new and complementary method for Argo float quality control, based on floats inter-comparison. Every Argo profile is then compared to neighbouring floats available during the same period. The frame of the interpolation is derived from the method of Owens and Wong (OW, V1.0) used by PI’s for delayed mode corrections (Wong et al., 2003; Böhme and Send, 2005; Owens and Wong, 2009), which has been adapted in order to compare float tested measurements with data from neighbouring floats.
The results of this comparison are complementary to the classic OW method, in which the reference database is build from historical CTDs. This new method is then useful in regions of poor historical CTD covering, like for instance the Antarctic region. Furthermore, the method has been adapted in order to be run in near realtime and quasi-automatic mode: a salinity deviation ∆S is computed for each profile by averaging the differences between observed and interpolated salinity over the 10 most stable θ levels (those where salinity variance is the lowest).
This method has been applied to 210 Argo floats in the North Atlantic, for which PIs have already performed delayed mode analysis with the classic method.
For all those new methods, the plan is thus to implement these different techniques in the real time processing system and to generate a series of diagnostics (plots, warnings) on the float data quality. These methods will also be analyzed to see how they could improve the delayed mode data processing.
2.4 - Visual Quality Control
A visual QC is performed by an operator on all profiles with bad or probably bad data. This procedure is performed on all in-situ measurements which are on alert after the application of the validation procedures described in previous sections. The main functions performed at the Coriolis Visual Quality Control (CVQC) are :
• Display the profiles of a station
• Change profile quality flags
• Compare current profile to neighbouring profiles
• Display ancillary information of a station, meta-data
• Control of platform speed
• Control of the density profile
• Display of T/S diagrams
2.5 - COMPARISON OF ARGO QC & METEOROLOGICAL FORECAST CENTRE QC
This work is being carried out by BODC in collaboration with Alastair Gemmell, Keith Haines, and Jon Blower at the Environmental Systems Science Centre (ESSC), University of Reading. BODC work on this to be undertaken with EuroArgo and MyOcean resource.
During late 2009 a meeting of the Argo data assimilators including representatives from the UK RAPID project, ECMWF and the UK Met Office occurred. One of the questions raised was how good is the automated operational quality control (QC) when compared against the delayed mode QC undertaken by the Argo project?
To try and address this question Alastair Gemmell produced a database and web-portal for such a comparison (examples shown in Figure 6 and Figure 7). The operational data in the comparison is from data assembled by Jim Cummings (FNMOC) detailing accept and reject decisions on Argo profiles from the GTS by 4 meteorological assimilations namely those used by :
- BMRC: Australia's Bureau of Meteorology Research Centre
- FNMOC: US Navy's Fleet Numerical Meteorology and Oceanography Center
- MEDS: Canada's Marine Environmental Data Service
- UKMO: UK's Met Office
The data from the Argo project QC is taken from the Argo format netcdf profile files at from the Argo GDACs.
Because the source Argo data in from assimilations has come from the GTS it has already passed all the Argo automated real time data checks and comparisons such as the following are possible: Compare assimilation profile accepts/rejects to Argo files flags to try and identify systematic differences and thus potential enhancements to the Argo real time systems and/or assimilation checks. Identify suspicious Argo profiles not already found in the Argo QC. These results can then be checked against results from the Altimetry QC and Coriolis objective analyses. Comparisons by region e.g. Southern Ocean, North Atlantic etc. The portal is still in its early stages but initial results show significant differences in accept and reject decisions of profiles between the operational centres. The initial intention is to evaluate the portal with the aim of making the data public. It is hoped that the output from the Hadley Centre EN3 QC can also be added. The EN3 analyses is of ‘climate grade’ so will be suitable for a direct comparison with Argo delayed mode results. The addition of DAC information from the Argo data is possible and will also allow comparisons by DAC against the data assimilations.
Since the portal has the potential to check for any systematic differences in QC methods and assimilations and it is hoped to feedback any findings to the operational centres and the Argo data management team. The goal being to improving the operational assimilations and Argo project QC where appropriate.
Antonov, J., R. Locarnini, T. Boyer, A. Mishonov and H. Garcia (2006), World Ocean Atlas 2005, vol.2, Salinity, NOAA Atlas NESDIS, vol. 62, edited by S. Levitus, 182 pp., NOAA, Silver Spring, Md.
AVISO, 2008: Ssalto/Duacs user Handbook: (M)SLA and (M)ADT near-real time and delayed time products. SALP-MU-P-EA-21065-CLS ed. 1.9, 39 pp.
Böhme, L., and U. Send (2005), Objective analyses of hydrographic data for referencing profiling float salinities in highly variable environments, Deep Sea Research II, 52, 651–664, doi:10.1016/j.dsr2.2004.12.014.
Bretherton, F., R. Davis and C. Fandry (1976), A technique for objective analysis and design of oceanic experiments applied to mode-73, Deep Sea Res. Oceanogr. Abstr., 23, 559-582.
Guinehut, S., C. Coatanoan, A.-L. Dhomps, P.-Y. Le Traon and G. Larnicol (2009): On the Use of Satellite Altimeter Data in Argo Quality Control, Journal of Atmospheric and Oceanic Technology, 26, 395-402.
Locarnini, R., A. Mishonov, J. Antonov, T. Boyer and H. Garcia (2006), World Ocean Atlas 2005, vol. 1, Temperature, NOAA Atlas NESDIS, vol. 61, edited by S. Levitus, 182 pp., NOAA, Silver Spring, Md.
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 θ–S climatology, Deep Sea Research I, 56, 450-457, doi:10.1016/j.dsr.2008.09.008.
Von Schuckmann, K., B. Garau, H. Wehde, T. Gies, D. Durand and F. Reseghetti (2010), Real Time Quality Control of temperature and salinity measurements, MyOcean document.
Von Schuckmann, K., F. Gaillard and P.-Y. Le Traon (2009), Global hydrographic variability patterns during 2003-2008, Journal of Geophysical Research, 114, C09007, doi:10.1029/2008JC005237.
Wong, A. P. S., G. C. Johnson, and W. B. Owens (2003), Delayed-Mode Calibration of Autonomous CTD Profiling Float Salinity Data by θ−S Climatology, Journal of Atmospheric and Oceanic Technology, 20, 308–318.