Quality Control Procedures applied to Real-Time Data.

Introduction

The collection of Real-Time Data from inhospitable locations is not a simple task. Throughout the life-cycle of a mooring there are innumerable problems which may arise to thwart the aim of data collection. Real-Time data collection enables data to be secured and used while the mooring is still deployed, giving scientists the earliest access to the data and also giving a continual “health-check” on the moorings integrity.

Designing moorings which will survive in deep waters for 12 months is a complex task and adding the elements to make the data being recorded available in real-time adds a further degree of complexity.

The 3 years of the ANIMATE project and the continuation into the MERSEA project has generated both valuable multi-disciplinary scientific data and information on the performance of the ARGOS and ORBCOMM satellite systems in delivering data in an operational manner.

Data Delivery

Two satellite systems have been employed to deliver data from moorings to the home base: ARGOS and ORBCOMM. ARGOS provides world wide coverage but message structure and high rates of data corruption limit the amount of information which can be sent. ORBCOMM can accommodate larger messages and the messages are seldom corrupted, but its coverage is poor at high latitudes and the costs higher.

ARGOS messages are prone to data corruption so each message must contain a check sum to confirm that the message received is the same as the message sent. The method chosen for this checksum is a Cyclical Redundancy Check(CRC). The value is calculated at the mooring and sent as part of the data message. The value is then recalculated when the data block is received by the home station and compared with the value embedded in the message. 20% to 25% of messages fail the CRC check and are not processed further.

Due to size constraints within the ARGOS message blocks the CRC has a value between 0 and 255 which implies a 1/256 probability of calculated and transmitted values being the same by chance when there has been corruption in the message block. These errors are detected by checking the range of a value appearing at the end of the message block.

ORBCOMM messages are sent as email messages. The email protocol ensures the messages are not corrupted during transmission, but like any email the message may be delayed or fail to arrive at all. The messages sent from the DOLAN buoy at the ESTOC

site have a sequence number as part of the data and observation of these indicates message loss to be extremely rare.

The ARGOS telemetry mooring used on the ANIMATE moorings has been developed by IFM-GEOMAR. Each mooring design requires a specific data format within the ARGOS message to encode all the data which may not be defined very long before deployment. The programs receiving the data are adjusted as data starts to arrive to ensure the data is correctly interpreted. The telemetry software stores instantaneous data for 2hours before transmission time and transmission time and then repeatedly sends the blocks of data for 2 hours which should ensure that the buoy will successfully upload the data to a satellite and also preserve battery life.

The message blocks are time stamped with the satellite processing time, but within each message there is a message number and block identifier. The block number increments at each transmission cycle, which for ANIMATE moorings has been every 4 hours, so when the time of starting the transmitter and the starting block number are known the time of data collection may be calculated.

Processing Temperature, Salinity and Pressure.

The physical oceanographic data, temperature, salinity and pressure, have been recorded using Seabird MicroCAT sensors and TD-Loggers built by IFM-GEOMAR but only the MicroCATs have the capacity to transmit real-time data.

Pressure

In each deployment the mooring was designed to have a mixture of MicroCATs SBE 37-IMP that have an integral pressure sensor and SBE 37-IM without pressure sensors. The sensors are distributed in such a way that the pressure reading from the SBE 37-IMPs can be used to interpolate the pressure for the SBE 37-IMs. The ANIMATE moorings are designed to remain taunt and upright and plots of the pressure readings indicate that they do act in this manner so an algorithm of

Pressure at nominal depth D1= (measured pressure at nominal depth D2) – D2 + D1

The pressure reading of each SBE 37-IMP in air is observed immediately prior to deployment and this value is used to correct for any offset in the pressure sensors readings.

The pressure is checked to be greater than the nominal depth at which it is deployed and marked as quality 4 if it is more than 10m shallower than the nominal depth.

Temperature

Any predeployment calibration information is applied to the data, either from a CTD dip or from a pre-cruise laboratory calibration. Temperature sensors tend to be very stable and not prone to drift.

Salinity

This is calculated using the measured and corrected temperature and conductivity, and the measured, corrected pressure or interpolated pressure. The algorithm [1] requires the IPTS-68 temperature and this is derived from the ITS-90 scale temperature using the following formula T68 = T90 / 0.99976

The World Ocean Database 1998 (WOD98) which tabulate minima and maxima for various depths in various ocean basins have been used to provide figures to range check the temperature and calculated salinity. The figures for the North Atlantic Basin were used for all ANIMATE sites.

Bio-geochemical sensors

At CIS and PAP the biogeochemical sensors were deployed in the euphothic zone, at about 40m. The frame was designed to deploy the suite of sensors, for nutrients, carbon dioxide and fluorescence together with a MicroCAT measuring temperature, salinity and pressure, in close proximity to each other. During early deployments the sensor frame was deployed as a separate, independent sub sea mooring with no satellite communication, the data being collected in delayed mode only. Development of electrically conductive swivels and inductive links to the biogeochemical sensors has enabled the redesign of the moorings to combine the telemetry and sensor frame functionality and allow satellite communication These new design moorings have not yet successfully delivered real-time data from the biogeochemical sensors.

At the ESTOC site the DOLAN buoy allowed the nutrient and fluorescence sensors to be deployed at a fixed depth, in the euphotic zone at 90m, and the carbon dioxide sensor was deployed at 10m together with a MicroCAT with no pressure sensor.

Carbon Dioxide

This is measured by Sunburst Sensors LLC, SAMI Carbon Dioxide sensor. A modified sensor was deployed at ESTOC from June 2003 which enabled data to be sent to the Data Centre via the ORBCOMM system.

Each SAMI is set up for each site to ensure maximum sensitivity at the expected temperature range, then calibrated by Sunburst prior to deployment.

Sunburst calculate the calibration constants required to convert the sensor outputs to partial pressure of carbon dioxide in micro atmospheres. The raw sensor outputs are received at the NOCS data centre and processed in a Matlab version of the Excel spreadsheet that Sunburst issue with each instrument.

The data are then presented to Professor Dr. Arne Koertzinger of IFM-GEOMAR. If he approves the data it is then processed and presented as a graph on the web site.

Observation of the graph would indicate non “realistic” changes in data but current experience has been that the SAMI sensor has either performed reliably throughout a deployment or returned no reliable information and so the entire data set has been rejected.

Nitrate and Nitrite

This is measured by EnviroTech LLC NAS-2E or NAS-3X Nutrient Analyser.

The CIS5 deployment was the first to incorporate this data in the telemetry data stream but unfortunately a failure in the inductive swivel above the sensor frame meant that no data was sent.

The WOD98 Appendix G for Nitrate may be useful as a range indicator.

During ANIMATE and continuing in MERSEA the EnviroTech LLC NAS-2E Nutrient Analyser have been deployed with 2 standard solutions on board which periodically are used to generate new coefficients for the conversion of the sensor outputs to micro M/l of Nitrate + Nitrite.

Changes in these standard levels will be used to quality check the incoming data. This and any further quality control checks will be developed with reference to ANIMATE delayed mode procedures.[2]

Chlorophyll-a

This has been measured with HOBI Labs HydroScat-2 sensors but none of the data was received in real time. This instrument makes 11 measurements in a burst and these can be used eliminate instrument generated errors and erroneous readings, possibly caused by microbiology.

The instrument currently used at all sites is the WetLabs FLNTUSB Fluorimeter but no real-time data has yet been received. The procedures used on the delayed mode data set will be adapted to the real-time data.

Conclusion

The experience built up during ANIMATE has enabled a methodology to develop so that quality controlled data captured in real-time from remote moorings in the Atlantic may be made available to the scientific community within hours of the original recording of the data. The data already captured during the previous 3 years of ANIMATE can form the basis of localised climatology which can be used to ensure that only high quality data with associated quality control markers is released to the general community.

Quality Control processes continue to evolve with the mooring design and consequent changes in data form and availability.

References

[1] N.P.Fofonoff and R.C.Millard Jr. Algorithms for computation of undamental properties of seawater. Unesco technical papers in marine science 44 Unesco 1983

[2] Marimar G. Villagargia. ANIMATE: Calibration of Biogeochemical data from sensors – Nitrate.