Methods and Evaluation of Data Quality

Hydrography methods and evaluation of data quality

At each station, an electronic CTD (for conductivity - temperature - depth) probe is lowered to the bottom or a depth of 150 m, whichever is less, using a hand-cranked reel. These probes measure water temperature, water salinity (using the electrical conductivity of the water which arises from the presence of dissolved sea salt), and pressure, from which depth can be determined, continuously through the whole vertical profile. RBR-Concerto CTD probes were used from 2015-2020. In 2021 these were replaced with AML-6 CTD probes.

CTD data is transferred through a wifi link to a tablet equipped with a purpose-written software app, at which point meta-data (e.g. station names and locations) is added.  This data is then transmitted for archiving at Ocean Networks Canada.

Although data is transmitted electronically to the Ocean Networks Canada archive, the resulting dataset still contains a number of problems. These can include casts without positions, profiles which appear to be erroneous when compared with nearby data because of sensor malfunction or failure to remove protective "caps", spurious “spikes” in salinity that may occur if the sensor hits the bottom, and so on.  Many of these issues are identified and corrected as ONC personnel carry out a cast-by-cast processing of the input data stream, generating "profiles" with values at standard depths.

However, a small number of errors still remain, and are most easily identified by examining an entire season's data as a whole. Thus an additional manual edit of the dataset was carried out by an experienced oceanographer at the University of British Columbia before this atlas was generated. Typically, these final corrections are made to less than 5% of the 700–1500 profiles observed in any year, but they can make noticeable differences in the atlas figures.

RBR CTD
A new AML CTD used for hydrographic measurements
Deploying a CTD using the downrigger reel

Temperature and Salinity

In this dataset, temperatures are in °C on the ITS-90 scale. Temperatures are provided using vendor calibration only, and have a nominal accuracy of about ±0.01°C.

Salinities are computed from temperature and conductivity. Conductivity measurements are also provided using vendor calibration only. Salinities are shown as mass fractions (grams of sea salt per kilogram of seawater) on the TEOS-10 Reference Composition Salinity Scale with a nominal accuracy of ±0.01 g/kg. Salinities on the older Practical Salinity Scale 1978 (PSS78), which are sometimes indicated with the label "PSU", are numerically smaller than Reference Salinities by a factor of about 0.9953; the difference is usually indistinguishable in figures.

Chlorophyll

The CTD profilers are also equipped with an extra sensor (a Turner Cyclops 7F Fluorometer) to measure a fluorescence signal arising from concentrations of chlorophyll in the water column.  Chlorophyll concentrations, in units of micrograms of chlorophyll per liter of seawater, are nominal and based on factory calibrations only, which are carried out using a particular chlorophyll standard. However, the response of live phytoplankton, when measured in-situ, often differs from this factory standard.  A common heuristic is that such measurements are accurate within a factor or two, and this appears to be true for measurements in the Strait of Georgia.

As part of the sampling plan, each patrol also obtains particulate matter from duplicate water samples obtained from a depth of 5 m at one station in each survey. A fixed volume of sample water is passed through a glass-fibre filter with a nominal pore size of 1.2 µm, so that all larger particles, which include phytoplankton containing chlorophyll, are retained. This sample is then analyzed in the laboratory for the actual amount of Chlorophyll at that depth.

Comparing chlorophyll concentrations. Laboratory analysis of water samples plotted against CTD fluorometer measurements at the same depth.

Scatter plots comparing these two independent measurements of chlorophyll concentrations at a depth of 5m are then made for each year. CTD fluorometer readings are plotted on the x-axis, and laboratory analysis of water samples at the same depth and location on the y-axis. The line shown is a least squares fit to a one-parameter line through the origin, with data weighted using an assumed error that is a constant percentage of the measured value. This best-fit suggests that multiplying the measured fluorescence by a factor of 0.6 is enough to match the laboratory measurements; this scaling factor is then applied to all chlorophyll values in this atlas until 2020, with the new scaling factor to be 1 for 2021 and 2022 due to a change in vendor calibration procedures. However, the scaling factor returned to 0.6 for 2023 for unknown reasons.

 

The scatter in these intercomparisons are relatively large for several reasons. First, it is simply difficult to compare measurements at the same nominal depth from profiles not taken simultaneously. The ocean changes! Second, the sampling depths may not even be exactly the same. Third, it is known that daytime in-situ chlorophyll fluorescence readings very near the surface can be lowered to as little as 50% relative to nighttime values in the Strait of Georgia by "non-photochemical fluorescence quenching", a physiological response of chloroplasts to large quantities of light.  Presumably the decrease would be less at depth where light levels are lower, and on cloudy or winter days, and perhaps even on bright days when vertical mixing is vigorous.

Dissolved Oxygen

The CTD profilers are  also equipped with another sensor (either an Aanderaa optode or an Alec Rinko-III) to measure dissolved oxygen using a "fluorescence quenching" technique which has become widespread in recent years. Such dissolved oxygen measurements are measured as a factor relative to atmospheric saturation. By using a standard equation that provides the actual concentration, when in equilibrium with the atmosphere at different temperatures, salinities, and pressures, these measurements can be converted into molar units of O2, which we use here. Although no in-situ calibrations are carried out, these measurements are cross-compared with calibrated oxygen concentrations from surveys of the Strait carried out by Fisheries and Oceans Canada. The estimated static accuracy is around ±10 µM.

However, the oxygen sensor response time is generally longer than for temperature and salinity, so that profiles show a smoothed version of the actual changes in oxygen concentrations.

Secchi Depth

Secchi depth measurements of water clarity are also made at each station, essentially by lowering a white “Secchi disk” into the water until it cannot be seen by an observer on the surface; the depth at which this occurs is recorded.

The Secchi depth is a measure of the penetration into the water column of light that can drive photosynthesis, and is affected by the amount of particulate matter in the water column. Particulate matter can either be living (i.e., phytoplankton), or nonliving, mostly clays and fine sediments that enter the strait in the freshwater outflow of Fraser River. Roughly speaking, we expect phytoplankton in coastal areas to be growing in depths shallower than about 2 Secchi depths.

Because of its reliance on human perception, Secchi disk datasets are often very ”noisy”, but Secchi depths are a cheap and relatively simple measurement with a long history of use.

Lowering the Secchi disk into the water to measure water clarity