
A Guardian report on dual waterspouts in San Francisco Bay, dated November 17, 2026, described a pair of waterspouts recorded in the San Francisco Bay area while storms moved through the region. The article, by Uwa Ede-Osifo, describes the sighting as uncommon for the area and says that "inclement weather is expected to continue for few days." The summary does not address drinking water, source quality, or any utility operation, so this is not a water quality story in the narrow sense. It is useful to operators for a different reason. Storm periods are where monitoring records tend to become hard to interpret, and a dated account of stormy conditions gives a reference point for reviewing what the instruments recorded.
What a storm changes at the intake
Rain and wind move material. Runoff carries sediment, organic matter, and debris into surface water and onto open reservoirs, and wind can stir up material in shallow areas. A turbidity or particle sensor at an intake reads whatever passes it. A rising trace during a storm may therefore reflect a real change in source water. It may also reflect air bubbles, debris caught on the optical window, or a probe that has shifted from its normal position. The number on the screen looks the same in each case. The cause does not.
Flag the window, do not erase the value
When storm readings look wrong, staff often face a choice between removing outliers and keeping everything unannotated. Deleting values keeps a monthly average tidy but removes evidence that the source changed. Keeping every value with no note makes later analysis harder, because a spike from a fouled probe can resemble a spike from a real sediment pulse. A more defensible approach keeps the raw value, attaches a storm flag to a defined time window, and logs each manual cleaning, verification check, or calibration as its own event. The flag does not decide whether a reading is valid. It records that the reading was taken under conditions where validity needs a second look.
Schedule the check before the front
Sensor checks are easier to interpret when they happen on a known schedule. Before a forecast storm, a team can confirm that the probe is clean, the reference standard is current, and the logger time matches a reference clock. After the storm, a second check shows whether readings drifted during the event or whether the probe was simply covered. A record that contains both checks is more useful than a record with a single post-storm cleaning. Neither check proves the storm reading was correct. Together they narrow the uncertainty and document how far the instrument can be trusted.
Gaps and delayed batches
Storms also interrupt power and communications. A pump station running on backup power, a radio link that drops, or a logger that buffers data internally can produce gaps or late batches. When delayed data arrive together, their timestamps should be checked against the logger clock and the telemetry log before they are merged into the main record. Gaps should be kept as gaps and labeled as such. Filling them by interpolation may be acceptable for some analyses, but only with a note that the values were estimated, not measured.
Staff time is part of the data
Storm response also competes for staff time. A crew that cleans probes during a front is not checking pump controls, and a crew checking telemetry is not documenting the flag. Small utilities often assign these tasks to one operator. A written list that names the storm tasks, their order, and who signs each entry keeps the record complete when the day is busy. The list does not reduce the uncertainty in the data, but it prevents the most common gap, which is a reading that no one can explain later.
What a storm record can support
A record from a stormy period can answer narrow questions, such as whether turbidity rose at the intake during a defined window and whether the rise coincided with a documented rain event. It cannot by itself establish cause, rule out sensor fault, or characterize the source for the rest of the year. Those conclusions need the flagged data, the calibration log, the telemetry record, and, where available, samples collected under the same conditions. The storm is the context. The records decide what can be concluded from it.