FixO3 Best Practices
The Fixed point Open Ocean Observatory network (FixO3) seeks to integrate European open ocean fixed point observatories and to improve access to these key installations for the broader community. This handbook collects the “best practices” in all phases of the system covering the entire infrastructural chain of data acquisition. It includes recommendations on how to produce high quality data aiming towards common methodologies and protocols within the FixO3 network. Find out more Web: www.fixO3.eu Twitter: @fixO3Project FixO3 is a project financed by the European Commission through the 7th Framework Programme for Research, Grant Agreement 312463 Handbook
This wiki page extracts sections of the FixO3 handbook which collects the “best practices” in all phases of the system covering the entire infrastructural chain of data acquisition. Find out more Web: www.fixO3.eu Twitter: @fixO3Project FixO3 is a project financed by the European Commission through the 7th Framework Programme for Research, Grant Agreement 312463
Maintenance plays a critical part in building a successful network of ocean platforms, and requires careful planning, correct use of resources, that include not only highly qualified personnel, specialized equipment, but also, many times, utilization of costly ship time. Specific characteristics referring both to local conditions and equipment requirements have to be taken into account. Platform maintenance procedures can be split into those that can be carried out on site, and those that can be carried out on land. Rationale for choosing one of these methods and the respective limitations will be discussed in the next paragraphs.
Although carrying out maintenance on site has the obvious advantage that the instrument array can be re-deployed almost immediately, it is limited by both the facilities available and the extent of maintenance procedures that can be offered, but also the strict time limitations related to possibly involved ship-time, weather conditions etc. Thus, any on-site 33maintenance procedure, periodicity, and means involved has to be carefully defined and planned beforehand, especially if a ship and its crew are involved in the operation. Recommended maintenance procedures than can take place on board the maintenance vessel are:
-
Visual inspection of the buoy or station hull and external housing for damage
-
Data downloading from the instruments internal memory
-
Cleaning the station housing and external components
-
Replace by the spare platform or
-
Check and replace if necessary:
- power related components (power line, batteries, solar panels, etc.)
- cabling or connectors
- station telemetry modules (antennae, positioning systems, etc.)
- navigation lights, radar reflectors, etc.
- sensors (specific procedures regarding sensors in the following paragraph)
- broken or damaged parts and components
- mooring line rope or cable mooring line components
- shackle axis and tapered pins.
- mooring line anchoring system
-
Checking and testing the functionality of all the components and subsystems
-
Applying antifouling coatings on the parts in contact with seawater.
-
Visual inspection of the mooring line for damage
-
Cleaning the underwater components from shells and algae
-
Check the free movement of each swivel around its head. If any swivel head sticks it must be replaced.
Recommended maintenance procedures that can take place on site with an ROV (or divers in shallow areas) for seabed observatories:
-
Data downloading with underwater wireless connection
-
Visual inspection of the seabed stations
-
Connection of replaced instruments or new instruments
-
Connection of new battery packs
-
Removal and deployment of associated autonomous instruments (OBSs,temperature loggers, landers, etc.)
-
Complete check of the connection to shore by a communication between the ROV team onboard the ship and the onshore remote maintenance team
-
Checking and testing the:
- release system of mooring lines and standalone nodes
- underwater communication systems (acoustic modems, inductive modems)
-
Applying antifouling and anti-corrosion coatings and anodes (discussed in greater extend in following paragraphs)
Station recovery and subsequent maintenance procedures carried out on land have the obvious disadvantage of instrument downtime, and possibly a second ship involvement for re-deployment, but they provide for ample time and facilities to carry out detailed work, maintenance and calibration, that would not be possible on site. The concept of long term monitoring of Eulerian observatories requires the management of spare components and spare sensors following the calibration protocols. It has the advantage to avoid a ship waiting forinstrument repair onboard, and the need to bring all experts onboard. Furthermore, new components and upgrades can be installed and all the modules can be tested in detail.
Components that would benefit from maintenance on land are instruments (sometimes individual sensors), actuators and mobile systems, front end electronics and loggers. Since each sensor has dedicated maintenance procedures, described in the manufacturer's manual, and many times demanding specific spare parts and equipment or it can be time consuming, so it can only be carried out on land.
It is a good practice to have available a second set of serviced and calibrated sensors, so the ones operating in the field can be easily and swiftly replaced. General maintenance guidelines differ for sensors operating in the air and sensors operating underwater. But when robustness is concerned, all FixO3 equipment will also be possibly submerged by seawater during their lifetime (storms, transfer for maintenance from small boats) and must be considered as underwater equipment.
Most important recommended maintenance procedures are:
- Applying antifouling and anti-corrosion coatings and anodes (discussed in greater
- extend in following paragraphs and Appendix C)
- Visual inspection
- Data downloading
- Cleaning
- Replacement of broken parts
- Servicing moving parts (e.g. anemometers)
- Replacement of connectors and cables if necessary
- Installing fresh batteries if necessary
- Checking memory if necessary
- Applying manufacturer's instructions and application notes
- Checking sealing and replacement of O-rings if necessary
- Leakage test (checking inside pressure of the housing when such system is
- implemented)
- Testing
The maximum sensor life on the shelves should be considered before its deployment. Servicing of the sensors should be made so the sensor doesn’t stay stored longer than recommended. For example, nitrate sensors need to be store with a wet cap protecting the probe. However, once the small volume of liquid in the wet cap has reached equilibrium of ionic transfer the depletion of gel is lessened. Thus, the sensor values may start drifting sooner than expected during long term deployments. The general procedures are valid and for the sensors operating inside the seawater, but in this case it is strongly recommend performing a leakage test. Especially for the sensors operating attached on a mooring line or a bottom frame a pressure test should be performed if possible. The majority of underwater sensors are sensitive to biofouling and their data are significantly affected usually weeks or even days after the field deployment. The techniques and materials for the removal of fouling are usually described in the sensor manual and should be included in the maintenance procedures. A summary of those procedures, for the main categories of oceanographic sensors, is presented in the table below.
Table 2.1.3.2a: Maintenance procedures for oceanic sensors with cleaning methods
FixO 3 standards requires that all the biofouling and corrosion protection methods will be completely presented and detailed in the Deployment document. This will include the estimate of the degree of protection of all parts of the observatory and sensors as well as a description of the systems that are not protected and the expected consequences. The descriptions must be compared with examples of in-situ tests and previous deployment experiences of the observatory or other observatories or submitted to corrosion experts (see Appendix C).
The environment of FixO3 observatories is one of the most corrosive environments on Earth. Unprotected steel structures continuously immersed for such long periods show corrosion rates of 100 - 200μm per year. This factor must be considered throughout the design of the observatory since a bad choice of materials could accelerate corrosion rates. There are various types of coating and they are often combined with cathodic protection. Coatings are the most common methods used for the protection of materials in offshore environments and the most common standards of offshore coating are:
- NORSOK M501-revision 5
- ISO 12944-1998/2007
- ISO 20340-2009
Material coating used for FixO3 observatories must comply with one of these standards. These standards define critical parameters of the coating selection such as the type and condition of the substrate, the environment and possible additional stresses, the surface preparation, the quality of the coatings, the selection of the coatings and the quality control of the substrate. For atmospheric zones it is important to select UV durable topcoats. Non- corrosive materials are strongly recommended instead of steel to improve the quality of the system such as Titanium and plastic. The use of composite materials with marine references is recommended.
The right choice of these materials apply to every element of the observatory including the structure, the connectors, the mechanical interfaces, the cables.... These elements must be carefully isolated to prevent corrosion and electrolysis. (see more details in Appendix C. The deployment document must specify all materials used in each of the parts of the system and any corrosion protection system of steel if these cannot be avoided.
Sensor protection from biofouling must be done with special care since it directly affects the area of measurement. This is especially important for optical sensors, membrane sensors and electrochemical sensors for which the interface between the measurement medium and the sensor sensitivity area must remain intact. Biofouling is a limiting factor in data acquisition at ocean observatories, especially during long term deployments. Figure 2.3 a shows an irradiance sensor after a one-year deployment during which the bioshutter broke.
Figure 2.3 a: Irradiance sensor after a one-year deployment during which the bioshutter broke.
Biofouling disrupts sensors in short-medium time (down to a few days) and therefore it quickly affects the quality of their data. Typically, bio-fouling creates a shift in the measurements that is particularly important for long term observations, shallow waters and near shore observatories. The appropriate techniques must be used according to the location of the observatory and the duration of the deployment. FixO3 observatories should ideally use proven solutions for sensor housing and sensing areas that could provide protection from biofouling while not interfering with the measurements. The protection must not affect the measurement or the environment while staying within the specifications of the observatory. For example, in order to endure the entire deployment, it should not consume too much energy and be reliable to aggressive conditions such as corrosion, sediments, hydrostatic pressure.... Today, the frequently used techniques are volumetric techniques such as bleaching or chlorine generation by seawater electrolysis and copper tape. Volumetric techniques act on small volumes surrounding the sensor area. Surface techniques on the sensing area are also possible and include wipers, copper shutter, water jets, ultrasonic sound, chlorine production.
Because observatories often look for limited cost, low power and easy to install on existing instruments solutions, other solutions that respond to the FixO3 requirements will be accepted if they are tested in the laboratory as well as in-situ where they are exposed to more harsh conditions. The time between maintenance is often directed by the biofouling protection, it is probably good for FixO3 policy to promote more efforts to adapt the most suited antifouling method to each sensor, and spend some budget for this aim.
Since bio-fouling may alter the oxygen, fluorescence and other measurements of the water, the housing of the instrument and the instrument frame must be protected from bio-fouling. Passive solutions to housing protection are to be chosen among the following options which has shown to be the most effective:
- Hull protection paint
- Antifouling paints with active biocides such as copper compounds, copper oxides and
- cobiocides chemicals
- Non-stick coatings paints based on silicon or fluorinated polymers
- Copper screen grid
- The addition of active anti-fouling devices may be a good choice for essential
- parameters in critical environment provided the time series of the active protection is
- recorded in the same data management system as the measurement for further QC.
- electromechanical biofouling protection devices
- chlorination by electrolysis or bleach injection
- UV radiation
For solutions such as paints and biocides, it must be demonstrated that the quantities used does not disrupt the environment of the observatory. This includes hazardous effects in the environment and interferences with the measurements. Among these solutions it is important to consider the high power consumption of electromechanical devices since they could have a strong impact on the duration of the observatory. Also, the non-stick coating solutions are usually biocide-free but they require of currents to remove the fouling, thus they are not appropriate for all locations of observatories.
FixO3 standards require a solution that protects the sensitivity of the sensor for the duration of the deployment. Passive biofouling prevention methods and active one are complementary in many cases.
The fixO3 observatories must include various elements allowing to successfully achieve their mission. They must carry sensors to allow measuring the chosen parameters for the specific location and power supply for the duration of the deployment. Additionally, depending on the type of the observatory, they can include meteorological payloads or data transmission systems for (quasi-)real data. Standardised payloads for all platforms is hard to achieve without the consideration of their goals and characteristics. The design should be optimized for each particular location and goals.
Meteorological payloads could be added to observatories with sea-surface access. These for instance, do not necessarily include bottom observatories. Meteorological payloads should measure parameters at the water-atmosphere surface such as temperature, humidity, air pressure, wind and position. Since these sensors are exposed to harsh conditions, they are vulnerable and they must be replicated when possible. It is recommended that these measurements are transmitted in quasi-real time as a feedback of the conditions of the observatory. Meteorological payload must be installed separately from the science payload in order to preserve the reliability of the other payload.
Oceanographic payload must respond to the science scope of the observatory. The observatory must always include sensors to measure basic parameters such as temperature, salinity and pH. Other sensors may include pCO2, oxygen sensors, cameras, fluorometer, plankton sensors, nitrate... All the parameters that the observatory intended to measure must be described in the Deployment document as well as the sensors that are used for this purpose. This description must include an estimate of the accuracy of these measurements and how the chosen sensors respond to these specifications. A description of the characteristics of various instruments will be provided by FixO3 to facilitate the choice of the sensors.
A scientifically powerful component of the seabed based observatories concept is the long time-series collection of multiple variables at a single location. All the bottom observatories should include sensors necessaries for climate system monitoring such as temperature, conductivity (salinity), pressure (depth), turbidity, dissolved oxygen, ocean currents, and passive acoustics suitable for all sites and depths. These variables, known as Essential Climate Variables (ECV), were defined to support the work of the UN Framework Convention on Climate Change (UNFCCC) and the IPCC. Moreover, other variables can be considered, such as the remaining ECV and other key chemical variables (e.g. Chl-a, pH, CO2, CH4, H2S, Eh, and hydrocarbons). In addition to an oceanographic payload, bottom observatories will also carry sensors specific to measurement of the seafloor environment. Biological payloads include cameras for visual monitoring of the seafloor (e.g. megafauna, phytodetritus input, bioturbation). Measurements of the optical characteristics of the water (light transmission, backscatter and fluorescence) are also recommended. Hydrodynamics and monitored with current meters and ADCP. Sediment traps can also be deployed from bottom observatories monitoring re-suspended as well as allochthonous material. Broadband ocean bottom seismometers, flowmeters and methane sensors are used for marine geosciences (e.g., seismic waves, gas seepage, etc.). Other sensors include hydrophone, sonar, tsunami pressure sensor, hydrophone array, precision range meter for crustal movement observations, potentiometer, magnetometer, tiltmeters....
Sampling rates of the sensors mainly depend on the temporal variations of the parameters that are to measured. However, it is important to consider the duration of the observatory as well as the power and data storage availability in order to define the sensor configurations. Ideally, fixO3 observatories will have remote communications that will not only allow data accessibility but also configuration changes of the sensors.
While the sensor configuration depends on the observatory framework, recommended values are as follow:
- A millisecond for seismic parameters
- High frequency sequences for acoustics (ADCP, passive or active)
- A fraction of hour for temperature, depth and salinity
- A fraction of a day for pH and CO2 sensors
- A fraction of a day for fluorometers
- A few hours for nitrate sensors
The choices of the sensor configuration must be stated in the deployment document. This should include a description of how the chosen configuration will take the appropriate measurements for the specific science goals in the observatory site. It will also have a technical report on the impact on the duration of the observatory due to data storage and power consumption.
In the chain of data acquisition, sensor calibration in the field and in the laboratory aims primarily at ensuring the measurements' accuracy. Regular, robust, well-documented calibration procedures will enhance inter-comparability of long-term data from the same or different infrastructures. Sensor calibration should be performed prior to and after deployment in order to apply corrections to the acquired data in case of drift of sensors. The sensor payload of FixO3 platforms is described in the “Review of the current marine fixed instrumentation, D2.2”. According to D2.2, most instruments of FixO3 platforms are used to measure temperature, salinity and oxygen. However other parameters such as currents, fluorescence, suspended matter, nutrients, seismology parameters, absolute pressure, chemical analyses, camera imaging, acoustic sound are successfully measured. Biofouling is reported as the most common problem in producing high quality data sets.
The strategy followed and the effort invested in the production of data for each parameter depends on the maturity, sensitivity to drift and the technical characteristics of the sensors. Overall, standardized methods exist for temperature and salinity sensors. With the exception of bio-fouling problems faced by some conductivity sensors, data derived from fixed-point observatories are of appropriate quality for many applications for both parameters. It is more challenging to acquire accurate measurements of dissolved oxygen and chlorophyll from automated sensors. At first, fluorescence, from which chlorophyll concentrations are derived, has a complex dependency on environmental conditions and phytoplankton community composition. Both artificial and natural calibration references exist but none of them is used as a commonly accepted standard. For the dissolved oxygen calibration, the Winkler method, which has very good accuracy and precision, is used as the calibration standard. It requires laboratory equipment though and in combination with the fact that, at least some of, the market’s automated oxygen sensors are prone to drift renders the monitoring of their performance challenging.
To reduce the costs and maintenance time of the sensors some of the observatory operators have established calibration facilities of the “core” parameters measured by fixed-point observatories. Best practice recommendations for the calibration of the “core” parameters are given below.
Temperature and conductivity calibration is performed by placing the sensors in a thermostatic calibration bath filled with seawater. The calibration points are selected according to the expected oceanographic range of temperature and salinity. For each calibration point the temperature reference is obtained by averaging a number of measurements taken with a calibrated Standard Platinum Resistance Thermometer (SPRT). The conductivity reference is obtained by inverting the salinity measurement of a water sample at each calibration point. The reference temperature is used for the inversion. A laboratory salinometer standardized with IAPSO standard seawater is used to obtain the reference salinity of the water samples.
Recommendations:
- Sensors should be visually inspected prior to calibration.
- Real-time monitoring of the conditions of the calibration bath can ensure the bath
- stability and homogeneity at the calibration points.
- Sensor handling and storage should follow the recommendations of the
- manufacturer. Sensor calibration should be performed prior to and after the
- deployment. The maximum period between two calibrations should not be more than one year.
- Reference instrumentation should be regularly sent to the certified laboratories or to the manufacturer for calibration.
- Calibration results should be accompanied by a declaration of uncertainty and information on reference material.
- The calibration and the deployment history of the sensors should be available for traceability. In the long-term this will provide useful information on the sensor performance and will reduce the risk of failures.
For the determination of chlorophyll fixed-observatories use optical sensors. Optical sensors sense a proxy, the fluorescent part of chlorophyll in the cells. The ratio of fluorescence and chlorophyll is variable and complex to determine. Also the response of the instruments to reference standards is sensor–specific. A common accepted method in calibrating optical chlorophyll sensors does not yet exist. The purpose of calibration of Chl-α fluorometers is to provide a reference to which all fluorescence measurements will be related through arbitrary fluorescent units. The use of fluorescence standard will ensure traceability. Most importantly it will provide comparability between sensors and deployments. Dissolved chemical standards, solid standards or algae cultures may be used as primary calibration standards. Manufacturer calibration can also be used but the operator should be able to track the instrument performance. Cultures should not be used as primary calibration standards because their fluorescence to Chl-α ratio varies. A review of standards for calibration of in situ fluorometers is given in Earp et al. (2011).
After the calibration the ratio of the fluorescence and the Chl-α may be determined, this is the validation process. The purpose of fluorescence validation is to explain the variability of fluorescence to Chl-α ratio. The validation process is composed of analysing natural phytoplankton samples (or cultures) assuming their optical properties are similar to those of the site the sensors operate. Chl-α is extracted form water samples and used as reference. Sensor measurements are then adjusted to the Chl-α concentration, usually through linear adjustment. Other auxiliary measurements, such as irradiance level or measurements from additional fluorescence channels, may be used to improve the fit.
Recommendations:
- Instruments should be cleaned and optically checked before calibration.
- Instruments should be sent on a regular basis to reference calibration lab or manufacturer for maintenance. Storage and handling should follow manufacturer recommendations.
- Calibration should take place in constant temperature. Instrumentation used should not cause background fluorescence.
- The effect of light on measurements should be taken into account. Background light should be minimum. PAR and irradiance sensors are useful to correct such effect in the time series.
- Keep trace of the procedure and information of reference materials used. The calibration coefficients and the variability of fluorescence to Chl-α ratio should for each sensor should be stored.
- Linearity of the sensors over the measuring range should be verified.
We address here two types of autonomous sensors for measuring dissolved oxygen, electrochemical and optical. The electrochemical sensors, such as SBE43 from Seabird Electronics, have high initial accuracy 2% of oxygen saturation and precision 1μmol/l. Oxygen measurements change the chemistry of the electrolyte of the sensor and result in a slow drift. The operation of optical sensors, such as the Aanderaa optode, is based on the fluorescence quenching principle. The response time of the optical sensors is slow compared to that of the electrochemical sensors. The advantages of the optical sensors are the long-term stability and the high accuracy, ~5μmol/l, provided sufficient time is given to equilibrate.
Both types of sensors are calibrated against the Winkler method (Winkler, 1988) that has high precision and accuracy (2μmol/l). High accuracy/precision results may be obtained following a multi-point calibration protocol. Multi-point calibration requires an apparatus/bench capable of creating different DO concentration. Temperature and salinity are measured during the calibration experiments. The effect of pressure on the DO sensors is not taken into account at present. The level of sophistication of such the benches, which are custom made, vary. IFREMER, Mediterranean Institute of Oceanography (CNRS) and Max-Planck Institute facilities are some examples of the state of the art facilities for oxygen calibration nowadays. Recommendations on specific equipment cannot be given at present. In general, the bench used should be thermally insulated and of high stability and homogeneity. An easy way of producing DO concentration gradient to perform multi-point calibration is to change the saturation of the water by changing its temperature.
The response of the optical sensors to the DO concentration can be linearized (Demas et al., 1999) and thus a multi-points calibration has to be performed. It is crucial that the temperature sensors be carefully calibrated before the oxygen calibration.
Recommendations:
- Only dedicated staff using specialized equipment should perform the DO calibration.
- The accuracy of the reference measurements (e.g. Winkler titration) depends strongly on the operator. Inter calibration experiments should be performed to eliminate that factor.
- Sensors should be visually inspected prior to calibration.
- The calibration facility should be monitored in real-time to ensure homogeneity and stability at the calibration points.
- Sufficient time should be given to the calibration facility to settle at the calibration points.
- Instrumentation should be handled and stored according to manufacturer recommendations. Detailed logbooks of calibration experiments should be archived for traceability.
- If temperature-conductivity sensors are used to monitor the state of the calibrating facility, they should have been carefully calibrated.
- Calibration results should be accompanied by a declaration of uncertainty and information on reference material.
Earp, A., Hanson, C. E., Ralph, P. J., Brando, V. E., Allen, S., Baird, M., Clemenson, L., Daniel, P., Dekker, A. G., Fearns, P. R. C. S., Parslow, J., Strutton, P. G., Thompson, P. A., Underwood, M., Weeks, S., and Doblin, M. A.: Review of fluorescent standards for calibration of in situ fluorometers: recommendations applied in coastal and ocean observing programs, OPTICS EXPRESS, 19, 16768-26782, 2011.
Calibration of the pH sensors will be performed ideally by a calibration lab but still by the manufacturer. Validation in the lab should be perform to check the calibration and possible drift of the measurements for sensors that are stored for a long period of time. The validation of pH sensors requires temperature measurements in order to perform correction of the recorded data. Sensors may be set up in continuous reading mode with a SBE sensor in a Tris buffer #26 solution of a known pH. It is important to let the sensor warm-up and stabilize for some time before taking the measurements. Typical values of a SeaFET pH sensor are shown in the table below.
Table 2.4.3.1a: SeaFET pH sensor calibration tests
Recommendations:
- Validation of the sensor should be done with special care of the probe. Special equipment may be needed to keep the probe wet and protected when the protective cap is not on during calibration.
- Only dedicated staff using specialized equipment should perform the calibration. However, validation of the sensor should be done in the lab with a known solution.
- The accuracy of the reference measurements depends strongly on the operator. Special attention should be taken with measuring the temperature and applying the correction factors.
- Sensors should be visually inspected prior to calibration. For instance, chemical sensors need to be checked for water flow across the sensor.
- The calibration should be done to avoid extended length of storage time prior to deployment.
- The sensor should be stored between calibrations according to the manufacturer's recommendation and FixO3 documents. For instance, in order to avoid gel depletion, the probe needs to be stored in a wet cap containing seawater.
- Instrumentation should be handled and stored according to manufacturer recommendations. Detailed logbooks of calibration experiments should be archived for traceability.
- Calibration results should be accompanied by a declaration of uncertainty and information on reference material.
The nitrate sensors such as ISUS or SUNA calibration uses one-point calibration method. This involves nitrate calibration standards at various concentrations. Common values are 5, 10, 20 and 40 μM concentrations prepared using nitrate standard stock of 1000uM and ultra- pure deionised water (Milli-Q DIW). The values obtained during these measurements should agree within a few percents with the manufacturer’s calibration. Figure 2.4.3.1 a shows an example of a ISUS sensor calibrated with the Milli-Q DIW sample for both in-lab and on- board bench calibration. It shows values in the range of the Satlantic’s specifications (red dots) of 0±2 μM for Milli-Q DIW.
Recommendations:
- The accuracy of the reference measurements depends strongly on the operator. Inter calibration experiments should be performed to eliminate that factor.
- Sensors should be visually inspected prior to calibration. The optical windows should be clear and clean.
- The sensor should be stored between calibrations according to the manufacturer's recommendation.
- Instrumentation should be handled and stored according to manufacturer recommendations. Detailed logbooks of calibration experiments should be archived for traceability.
- Calibration results should be accompanied by a declaration of uncertainty and information on reference material.
Figure 2.4.3.1 a: An ISUS sensor calibrated with the Milli-Q DIW sample for both in-lab and on-board bench
calibration.
To assess seawater carbonate chemistry is challenging because of the complex and not yet fully known interrelation between different forms of the carbonate system. The four main measurable parameters of the carbonate system are Alkalinity, Dissolved Inorganic Carbon (DIC), pH and pCO2 . If two of these parameters are provided along with temperature, pressure, total phosphate concentration and total silicate concentration the other two can be empirically calculated based on the work done by Lewis and Wallace (1998). To facilitate these calculations openly available Excel based software called CO2sys is available on the internet and/or as a cell phone application called CO2calc. It should be noted that in seawater only a small portion, some %, of the dissolved inorganic carbon is present as CO2 gas = pCO2 .
For aquatic in situ measurements of pCO2 , four main detection principles have been used:
-
Infrared (NDIR): based on the equilibration of CO2 gas dissolved in water through a gas permeable membrane to an inner air-filled, pumped gas-circuit of the analyzer, where the CO2 concentration is measured optically using non-dispersive infrared absorption spectrometry. These sensors often have scrubber that occasionally can remove CO2 . This gives the possibility of obtaining an internal 0 CO2 reading to compensate for detector drift. It also leads to that these sensors consume CO2 which implies that good water circulation should be maintained in front of the membrane. Manufacturers of these sensors include: HydroCTM/CO2 , CONTROS Systems & Solutions GmbH, www.contros.eu and CO2-ProTM, PSI, www.pro- oceanus.com.
-
Colorimetric: based on optical detection of the pH induced color change of the indicator solution, which is equilibrated with ambient seawater pCO 2 through a gas- permeable membrane. These sensors have to be equipped with a reagent that has to be renewed in-between deployments. SAMI pCO2 from www.sunburstsensors.com manufactures these sensor.
-
Foil based optode: CO2 gas diffuses from the surrounding water into the hydrophobic (only gas can pass) pH sensing indicator, where as a consequence the pH is modified. The magnitude of pH change is correlated to the pCO2 level outside the membrane. The embedded DLR (Dual Lifetime Referencing) material exhibits a pH dependent fluorescence change, which is detected as a phase shift value of returning modulated red light. These sensor foils cannot be used in sulfidic waters (presence of H2S) and should always be kept wet during transport and storage. Aanderaa Data Instruments (www.aanderaa.no) manufacturers and sells these type of sensors.
-
ISFET based sensor: The principle of pCO2 measurement using ISFET-pH technology is as follows: Both the ISFET-pH electrode and the Cl-ISE reference electrode of the pH sensor are sealed in a unit with a gas permeable membrane, CO2 diffuses through, whose inside is filled with an inner electrolyte solution that contains a NaCl solution. The pH sensor can measure changes in pCO2 from changes in the pH of the inner solution, which is caused by penetration of CO2 through the membrane. These sensors are not yet commercially available.
-
Solid-state electrolyte cell: Measures partial pressure of CO2 gas in a gas mixture, which is equilibrated with the water outside the gas permeable membrane. Detection is based on solid-state electrolyte cell. This is a relatively new technology that is in development by e.g. http://www.franatech.com/index.html.
There is no absolute reference method for pCO2 which makes calibrations of such sensors difficult.
One method is to use constant bubbling with gas mixtures with known concentrations of pCO2 . All sensors are affected by temperature changes therefore for sensors to be accurate the calibrations should be done at multiple temperatures. The gas bubbling method is relatively robust but since it takes long time for the bubbled seawater to reach equilibrium these calibrations can take many days. If reference sensors could be included in the calibrations system, just like it is done in some of the O2 calibration facilities, the calibration procedures could be speeded up considerably. There are on-going trials to use Cavity Ring- Down Spectroscopy instruments normally used in atmospheric measurements as a reference during calibrations.
Another calibration method is based on changing the pH. A small pH change will lead to a significant change in pCO2 which can be calculated with CO2sys and used as a reference point. Difficulties with this method is again the time for equilibration, the risk of contamination from the atmosphere and that the pH determination will have to be very accurate.
The sensors should be calibrated against the ship CTD when possible. The salinity, oxygen and chlorophyll fluorescence could be calibrated against Niskin bottles. The samples could be analysed during the cruise or in the lab but it is important to take and label them correctly. Ideally, this calibration must be done before and after the deployment. It is critical to find the correction of the sensor after deployment - especially for long deployments - because of the changes of the sensor during the deployment including sensitivity drift and biofouling effects. In order to account for the status of the sensor during the deployment, the recovered sensors must be calibrated before cleaning the fouling. They must be installed at the same height in the CTD frame with a good calibrated sensor to be compared with (e.g. SBE37 ODO sensors are usually adjusted through a dedicated CTD-rosette cast at 1000m depth during 30min).
Nitrate calibration It is recommended to calibrate the sensor previous to a deployment using a CTD cast. The nitrate sensor can be set in continuous mode at the frame and record values internally for the trial duration. Measurements should be calibrated against Total Oxidised Nitrogen measurements (TON = NO3- + NO2-) from the Niskin bottles sampled at various discrete depths, according to the deployment depths. For PAP1 observatory, typical calibration depths are 5, 10, 20, 30, 40, 50, 60, 80, 100, 150 and 200 m. Configuration of the sensor involves the sampling mode and rate and logging options. The values should consider the duration of the deployment and the life of the battery that is used to power the sensor.
CO2 calibration and configuration, FixO3 pCO2 sensor inter-comparisons One goal in FixO3 is to perform a longer deep-water pCO2 sensor inter-comparison of the two technologies, pCO2 optodes and ISFET based sensors, that can handle high pressure without special modifications. This work was started in June 2016 by deploying a mooring, at 2500 m at the IFREMER Antares site. This mooring carried an Aanderaa SeaGuard multi- sensor platform with a pCO2 optode included, and two ISFET based pCO2 sensors from Tokyo University.
Before this deep-water deployment a pre-evaluation was carried out in the form of a two- month inter-comparative deployment at shallow water at the cabled Koljoefjord observatory, operated since 2011 by the University of Gothenburg, on the West Coast of Sweden (http://koljofjord.cmb.gu.se). FixO3 deals with underwater platforms hence we did not include any technologies that are limited to deployments from surface platforms (buoys and land based stations).
Overall the deployment included 14 different pH and pCO2 sensors. The table below lists and compares the pCO2 sensors that took part in this test. In the figure below the mooring frame that carried the sensors is described and further below some examples of results are given. A detailed report comparing the different sensors with reference data is available as a separate FixO3 report (contact Anders Tengberg, [email protected]).
To summarize the test for the pCO2 sensors, there was important fouling of the frame at the end of the deployment, which affected the sensors that were not equipped with antifouling protection. Due to power cuts there was gaps in the data from some of the more power hungry sensors that were dependent on power from land. Before fouling affected the sensors they all displayed similar relative variations in this dynamic environment with natural pCO2 oscillations from about 200-500 μatm (see data examples below).
None of the sensors was consistently agreeing with the reference data that was obtained from frequent water samples during the deployment (see data examples below). Therefore, it is not possible to judge which sensor(s) were the most accurate in absolute terms. The NDIR based sensors had better initial calibrations but when the Optodes and ISFET sensors where adjusted with the first reference value they gave similar dynamic changes and noise level (precision).
Sensors with pumps, two NDIR sensors from PSI, can normally not be sampled with the same frequency as the other sensors and the readings from the pumped sensors seems reflect a larger water volume since water is drawn from the surroundings. This was more visible at the end of the deployment when the un-pumped sensors were more affected by local fouling on sensors and the mooring frame.
The response time of the tested sensors is relatively slow, t 63 around 5-6 minutes which makes them challenging to use in applications where a fast response is required (e.g. for profiling).
Power consumption can be a serious impediment to longer deployments if reliable land power is not available. The pCO2 and ISFET sensors consume about 100 times less power than the other technologies.
Table 2.4.3.2a pCO 2 Instruments used during the Koljo Fjord inter-comparison study
*Cu tape normally used in other deployments for antifouling protection
Fig.2.4.3.2a. The FixO node with the sensors after assembling.
Fig.2.4.3.2b. Overview of pCO2 data recorded with CONTROS HydroCTM.
Fig.2.4.3.2c. Overview of pCO2 data recorded with Aanderaa pCO2 optodes. The insert is a blow-up of data from
the first part of the measurement campaign.
The primary objective of any ocean observing infrastructure is the acquisition of sensor data which in turn is used to provide information about relevant ocean variables (e.g. salinity, alkalinity, particulate organic matter, Chlorophyll-a concentration) for ocean system understanding as well as for monitoring. The data that is delivered by a sensor requires certain procedures for its conversion into an ocean variable. The procedures must consider two basic expectations:
- Quality control according to defined standards (or alternatively proper documentation of new standards)
- Quantification of data quality (e.g. standard error, error distribution)
Only if both expectations are addressed the derived variables can be used to quantify, describe and assess the oceanic environment. Moreover, any use of data within a network approach, that is merging information from one infrastructure with observations of other networks, requires the two data quality procedures to be applied. For example, a chlorophyll- a observation from a glider that surveys the waters around a FixO3 mooring can only be merged with the single depth mooring data if the parameter “Chlorophyll-a” derived from sensor data at the mooring is similar to the “Chlorophyll-a” derived from a maybe very different sensor data that comes from the glider. In order to protocol the quality control we follow the OceanSITES recommendations outlined in the “OceanSITES User’s Manual” available from their website(www.oceansites.org).
Similar processes of quality control are used by other disciplines of ocean science active in FixO3. The principles presented in this chapter are similar and under implementation by projects such as EMODNET and SeaDataNet. EMSO is supporting these processes.
For quality control flagging the concept as being worked out by the Data management team of the international OceanSITES initiative is applied (table below).
Table 5.1.1a OceanSites quality flags signification
Based on the OceanSITES User’s Manual the quality control and other processing procedures applied to all the measurement of a variable must also be set and protocolled as a value in the data structure. string values (in the ‘Meaning’ column) are used as an overall indicator (i.e. one summarizing all measurements) in the attributes of each variable in the processing level attribute.
Table 5.1.1a Processing level codes
The objective here is to address the most common data qualification methods used for physical and biogeochemical data acquired from autonomous sensors installed on fixed stations. These qualifications should be applied in post-deployment procedures (second level correction) in order to correct drifts and to qualify dataset. For physical data (pressure, temperature, conductivity and currents), a first review has been presented in 2005 with all Q/C procedures by Kartensen et al. 2005 (ANIMATE report). These procedures are still relevant and should be used as best practices for mooring sites.
In this paragraph, we focus essentially on correction methods that should be applied for biogeochemical essential climate variables recently implemented in observing system as FixO3 (oxygen, nitrate, pH, pCO2). The advantage of fixed stations versus Argo floats and gliders is the opportunity to perform robust comparison with in situ samples collected near the moorings when recurrent ship visits are possible. If regular ship visits are not possible, the use of climatology dataset near the fixed stations is also possible under steady state conditions (e.g. deep waters, no mixing zone, and low concentration level).
Table 5.2a Sensors capacity in the FixO3 stations network
Today, thanks to the Argo community efforts, oxygen qualification from optical sensors (optode) has improved especially due to the new calibration procedure performed by the manufacturers (multi-points calibration and Stern-Volmer equation). However, it is still admitted by the scientific community that the optode deployment always presents an offset around 20 μmol/kg. To correct this offset, we recommend using the last method from Takeshita et al. (2013). This linear method allows us to determine the slope and offset coefficients estimated from in situ sampling (Winkler titration) or climatology dataset (World Ocean Atlas). In sensitive oceanic regions, where seasonal variability is significant, we recommend to use in situ sampling as reference measurements (before mooring deployment and collection).
An example of calculations for the optode 3830 has been detailed here:
DOXY [μmol/kg] = C0 + C1DPHASE + C2DPHASE2 + C3DPHASE3 + C4DPHASE4 with Ci = Ci0 + Ci1T + Ci2T2 + Ci3T3 and DPHASE = BPHASE (blue light) considering RPHASE = 0 (red light)
This equation needs 20 sensor-dependant coefficients and calibration at this time was performed with two points calibrations (0 and 100% saturation). The more recent optode 4330 (faster time response) offers more accurate oxygen concentrations but with more complex equations:
Δp = C0 x Tm0 x CalPhasen0 + C1 x Tm1 x CalPhasen1 + ... + C27 x Tm27 x CalPhasen27
Air Saturation(%) = Δp x 100 / [(Nom Air Press – pvapour(T)) x Nom Air Mix]
DOXY [μmol/L] = [C* x 44.614 x Air Saturation] / 100
with CalPhase = f(Tphase) and TPhase = C1Phase – C2Phase (blue-red light)
This equation used 27 calibration coefficients and was reduced to 7 coefficients from Stern- Volmer equation taking into account the pressure and salinity compensation (Uchida et al. 2008, Bittig et al. 2012, Asaro& McNeil, 2013):
DOXY [μmol/L] = [(C4+C5 x T)/(C6+C7 x TPhase) - 1] / [C1+C2xT+C3xT2]
Since 2012, Aanderaa (XYLEM) performs a multipoint calibration method for the new optode in order to improve accuracy of Ci and DOXY precision (40 calibration points). However, after some tests, an offset of oxygen is still observed.
Figure 5.2.1 a: Oxygen measurements through optodes 4330 mounted on Argo floats. The line colors represent the different floats deployed over 3 days in pool for testing. The dots represent the Winkler in situ sampling. The mean difference between optodes on floats and Winkler values ranged around 5-12 μmol/L.
Once the optode measurements have been done, it is recommended to proceed to some corrections. Two methods are usually proposed:
- adjusting the calibration coefficients by using a polynomial fitting model (in situ Winkler or WOA climatology or pO2 in air are used as oxygen reference value)
- simple linear method to estimate the offset and gain (slope) of oxygen data (Takeshita et al., 2013).
DOXY_adj [μmol/L] = offset + gain x DOXY [μmol/L]
The second method proposed here is easier and any modifications of calibration coefficients are necessary. However, a large amount of in situ or climatology data is required to guarantee the robustness of the optode correction.
Figure 5.2.1 b: Example of polynomial fitting model used to adjust the calibration coefficient of optode (left panel). To perform this adjustment in situ profile at the deployment is needed (right panel)
Figure 5.2.1 c: Oxygen saturation measured from optode (floats) compared to oxygen climatology dataset from Takeshita et al. 2013 (WOA climatology comparison is not working in high latitudes areas).
The nitrate concentrations (NO3 ) in seawater can be obtained from optical sensors by using UV absorption technology (e.g. ISUS, SUNA, Johnson and Coletti, 2002). Currently, several methods are available for the deconvolution of the NO3 from the observed spectrum. The most widely used is the Temperature Compensated Salinity Subtracted (TCSS) algorithm which take into account the observed salinity and subtracted to the total absorbance and a temperature-dependant correction of seawater extinction coefficient (Sakamoto et al. 2009). Recently, a Temperature Compensated Salinity Subtracted (TCSS) algorithm has been developed for the deconvolution of nitrate concentration from the observed spectrum, taking into account the observed salinity and subtracted to the total absorbance and a temperature- dependant correction of seawater extinction coefficient (Sakamoto et al. 2009).
A(λ) − S x ESWtis(λ) = NO3 x ENO3 (λ) + α 1 + α 2 x λ
S x ESWtis(λ) = seawater absorption with T correction
NO3 x ENO3 (λ) = nitrate absorption
However, this method is sensitive and problematic in the Mediterranean Sea where low NO3 (0-9μM) and high salinity content (37-38.6) are observed (D’ortenzio et al., 2012; Pasqueron de Fommervault et al. 2015). Recently, an improved algorithm was developed and substantially improved the NO3 estimation in the Mediterranean Sea (Pasqueron de Fommervault et al., 2015). Major changes in the algorithm are:
- Treating the wavelength offset (wl, see Sakamoto et al., 2009 for details) as a tuneable parameter (Johnson, 2014): correction of bias in NO3 above about 20°C
- Considering the vertical lag between the CTD and the NO3 sensor (only in the case of the SUNA) by interpolating T and S at the depth of the SUNA (~ 1.5m): spikes removal in the thermocline
- Application of a pressure-dependent correction to the bromide spectrum: better estimation of NO3 at depth Regarding item (3) it was, indeed, observed that NO 3 at 1000m could be underestimated up to 60% in the Eastern Mediterranean. An empirical correction of 2% per 1000 dbar on the seawater absorption coefficient was decided:
ESW(λ,Tis,P) = ESW(λ,Tis,P)x [1−0.02 x P/1000]
The SUNA sensor still undergo offset and temporal drift which can be corrected by comparing with climatological values at depth (http://www.seasiderendezvous.eu)
Figure 5.2.2 a.: Example of nitrate measurements from Argo float in the NW Mediterranean Sea (lovbio068d) at cycle 73 (15 March 2015). The black line represents the raw data transmitted in real-time (RT). The blue line, the nitrate adjusted in RT (offset correction) and finally the red line the nitrate adjusted in delayed mode (pressure, offset and drift correction).
D'Ortenzio F, Antoine D, Martinez R, d'Alcala MR (2012) Phenological changes of oceanic phytoplankton in the 1980s and 2000s as revealed by remotely sensed ocean-color observations. Global Biogeochem Cy 26 | doi: 10.1029/2011gb004269.
Pasqueron de Fommervault O, et aL. (2015) Seasonal variability of nutrient concentrations in the Mediterranean Sea: Contribution of Bio-Argo floats. J Geophys Res-Oceans 120 | doi: 10.1002/2015JC011103.
Sakamoto, C. M. K. S. Johnson, and L. J. Coletti, (2009) An improved algorithm for the computation of nitrate concentrations in seawater using an in situ ultraviolet spectrophotometer. Limnol. Ocean- ogr. Methods, 7, 132– 143.
pCO2 optodes from Aanderaa have been used in a wide range of applications from shallow water lake, coastal and aquaculture measurements to deep water deployments on fixed or moving (gliders) platforms. The main advantages of these sensors include good long term stability once they have been deployed, compact size and low power consumption, 6000 m pressure rated and low pressure hysteresis. The main disadvantages are the difficulty to get them well calibrated, relatively slow response time and unknown stability over longer deployment period than 12 months. Before start using these sensors there are limitations that the end user need to be aware of including:
- They cannot be used in environments where there is H2S which is normally found in anoxic (no O2 ) environments. H2S will irreversibly contaminate the sensor foils.
- The sensor cannot be allowed to dry out. A cap with water in it will have to be placed on the sensor whenever it is in air.
- The sensor has slow response (5 min) and has to be used with caution in profiling applications.
To obtain high accuracy data the end user will have to take high quality water samples for field adjustment sometime during the deployment. Preferably DIC and Alkalinity to calculate pCO2. It is recommended to always measure O2 (with optode) in parallel, this give quality control and a better understanding of the ongoing processes. As can be seen from the calibration figure below (example from Atamanchuk et al., 2014) pCO2 optodes, just like 68other optodes, does not have a linear response with respect to solute concentration and temperature. For field correction, if a multipoint calibration has been performed, the whole calibration plane can however be moved up or down using just one reference point.
Figure 5.2.3a: Typical multipoint calibration curve for pCO2 optode from Atamanchuk et al. (2014). For field adjustments the whole plane can be moved up or down using just one/some reference points.
One simple and approximate field method for single point referencing of a pCO2 sensor before and after deployment is to place the sensors in constantly air-bubbled water, e.g. with aquarium pump, with approximately the same salinity as at the deployment site. This should be done in an environment that is open to the atmosphere. Doing this inside a laboratory should be avoided since there could be large changes in CO2 depending on the number of people and activities in the room. As a reference for the concentration in the bubbled water the atmospheric concentration (normally around 400 μatm), which is specific for the region and season, can be used.
The recommended steps of sensor handling to obtain higher quality long-term field measurements are the following:
- After calibration put the sensor in water that has similar salinity (±5 psu) as the water in which it will be deployed.
- Keep the sensor wet at all times until it is submerged. For longer storage the black cap + sponge delivered with the sensor can be used. Tape around the lower edges of the cap to prevent water from escaping. To keep the sensor wet also just before deployment place toilet paper soaked in seawater of the right salinity on the foil. This paper will dissolve and be washed away when the sensor enters the water.
- Let the sensor acclimatise (possible osmotic effects) for at least three days before taking reference water samples for adjustment. If possible, deploy O2 optode and Salinity sensor close to the pCO2 optode. The normal reverse correlation between O2 and pCO2 will give additional quality control and the possibilities to study on-going processes.
- If possible, take reference water samples in the proximity of the sensors during at least two occasions during the deployment to be analysed for DIC and Alkalinity and calculate pCO2 (with CO2 sys).
- Use CO2 optode specific Excel sheet, provided with the sensor, that contains the equations to convert sensor raw data (Cal Phase) to pCO2 values taking into account the sensors specific calibration coefficients. Please verify that the coefficients in the sensor and in sheet are the same. The coefficients are available by connecting the sensor to a PC (se manual).
- Use adjustment coefficient in the Excel sheet to tune first set of reference data to sensor readings.
- Verify with reference readings at other occasions that the sensor has been stable during the entire deployment.
- If it is of interest to study and better understand the on-going processes in the studied area pCO2 should be combined with another parameter in the carbonate system (preferably Alkalinity) to obtain DIC (calculated using CO2 sys).
- Changes in molar ratios between DIC and Oxygen can be compared and used to quantify and understand the on-going processes. For more information about how this can be done se Atamanchuk et al. (2015) Continuous long-term observations of the carbonate system dynamics in the water column of a temperate fjord. Journal of Marine Systems 148, 272–284.
The step-by-step procedure described above is targeted for pCO2 optodes. Several of these steps should also be applicable to other pCO2 technologies including: Take reference water samples at least two times during a deployment; Measure O2 and Salinity in parallel with pCO2 ; Establish relation between Salinity and Alkalinity to be able to use measured Salinity as a proxy for Alkalinity. Then use Alkalinity + pCO2 to calculate DIC; Use changes in molar ratios of DIC and Oxygen to quantify and understand the on-going processes.
The table below shows all correction methods usually applied in oceanography to adjust physical and biogeochemical data obtained from autonomous sensors. These methods can be used as Best Practices methods to correct bias observed in FixO3 data moorings.
Table 5.2.4 Summary of correction methods used nowadays for physical and biogeochemical parameters
Finally, pH and pCO2 data from recent sensors (CONTROS, Aanderaa, Seafet, ...) are evoked but the quality control procedures are not ready to be proposed as best practices yet. So far taking water samples and analysing those for DIC and Alkalinity, using certified reference material, and calculating pH and pCO2 appears to be the most robust. A calibration coefficients fitting seems to be the best way to correct the data for accuracy offset and drift.