Question: do the relationships hold in different water?
An understanding that works at one place is not yet an understanding. This dive occupies a station in different water — a second site, or the same site under distinctly different conditions — and asks whether the relationships observed at the first station travel.
Follow the Station Occupation Procedure in full.
This is the dive on which the Station Report is completed. It is required for certification.
Specific to this dive
Site selection is the point. The useful contrast is optical: a site with visibly different water clarity, a different bottom type, or a different exposure to runoff or sediment. A second site optically identical to the first tests nothing.
Check the array against the new site. A shallower station needs fewer loggers and a deeper one needs more — one per 3 m / 10 ft of water column, never fewer than three. The endpoint pair always supplies the reported coefficient.
If this is a new site, establish the station position and its relationship to whatever survey exists for it, and record explicitly how the conditions differ from the first station.
Name the relationship under test before the dive. The optical product is the primary one. The team should say beforehand what each outcome would mean: no resolved shift in the product across different water would be evidence that the relationship extends beyond the first station; a resolved shift would indicate the two waters differ in the balance between scattering and absorption, and therefore in what is suspended or dissolved in them.
Additional Analysis
After the standard analysis steps, compare the second station against the first. This is two questions, and they must be asked in order.
First: was a shift in Π resolved within the first-station set? You have three values — Dives 2, 4 and 5. They are not replicates: Dive 5 was deliberately conducted under a different condition, and the course is explicit that unresolved is not the same as unchanged. What can be asked is narrower and honest: did any pair of them differ by more than its threshold?
Test all three pairs, not two. Two adjacent differences can each sit below their thresholds while the difference between the endpoints exceeds its own.
| Comparison | Threshold |
|---|---|
| Dive 2 against Dive 4 | C from the two days' propagated uncertainties |
| Dive 4 against Dive 5 | C from their propagated uncertainties |
| Dive 2 against Dive 5 | C from their propagated uncertainties |
- If no pair resolved, compute a descriptive first-station reference value — the mean of the three — and report the observed spread alongside it. Call it what it is: a reference value and an observed spread. It is not an estimate of repeatability and the three observations are not replicates.
- If any pair resolved, Π moved within the first station. Transferability is untestable, and that is the finding. Record which pair differed and by how much. A team that discovers Π shifting within one day at one station has learned something more interesting than a transferability result.
Every Π entering the comparison must be valid. Confirm all four before proceeding, and treat any failure as making transferability untestable:
- computed from Kd(photopic);
- from a column that passed the optical linearity test — Supplement A.7 states that Π is not computed where it fails;
- from an extinction distance that was not censored;
- carrying a propagated uncertainty.
Supplement B.9 works this whole comparison with numbers — open it beside the Station Report.
Second, only if no pair resolved: does it travel? Compare Dive 6's Π against the first station's reference value. The threshold carries the same coverage factor as every other comparison in the course:
C₆ = 2√(s² + u²₆)
where s is the first station's observed spread, used as a conservative reference uncertainty, and u₆ is Dive 6's own propagated uncertainty. Without the factor of 2 this dive would use a decision standard roughly half as strict as everything before it.
The verdict is three-way, and the wording matters:
- No resolved shift at the resolution of this course. The difference does not exceed the threshold. This is not proof that the two values are equal, and not evidence that the relationship holds generally — it is a statement about what this equipment, on these two days, could and could not distinguish. The course teaches throughout that an unresolved difference is not equality, and this verdict says so rather than quietly claiming more.
- Resolved shift. The difference exceeds the threshold. The two waters differ in the balance between scattering and absorption, and therefore in what is suspended or dissolved in them.
- Untestable. A shift resolved within the first-station set, a validity condition failed, or something else prevented the comparison — a censored extinction distance at either station, or a failed logger.
Treat Kd and the extinction distance as context, not as parallel tests. Π is the transferability question, because it is the dimensionless relationship the course is asking about. Kd and y will differ between two different waters — that is what makes them different waters — and reporting those differences describes the sites rather than testing anything. Record them; do not give them a transferability verdict.
Then complete the Station Report.
Completing the Station Report
The report is the deliverable for the course and certification depends on it. Its full format is given in the Station Report Recording Forms. In outline it contains:
- Metadata — who, where, when, in what time zone, with which instruments by serial number, referenced to the site survey. Recorded once per occupation, so four times.
- Calibration — pre-dive and post-dive offsets, drift, and the measured single-value uncertainties for each parameter on each day.
- Measurements — sensed quantities only, each with its measured uncertainty or method limit, each reported to the number of figures that quantity supports, and with censored observations recorded as such.
- Derived quantities — attenuation coefficients labelled Kd(photopic), the thermal gradient, the optical product, and the propagated uncertainty of each.
- Horizontal Survey — the Dive 3 survey table: position, depth, time, observed value against matched reference, and whether each difference resolved. Filled in once for the course.
- Interpretation — both written predictions as they were recorded, the outcomes, the comparison thresholds Dive 4 produced for each parameter, which differences were resolved against their applicable threshold and which were not, which were attributed and which were not, and the transferability result.
- What went wrong — flooded instruments, abandoned measurements, lost depth, weather that interrupted a record.