Reading path: Coastal water observation →
Understand what the line represents
A moving average replaces each point with an average calculated over a chosen neighborhood of observations. This can reduce short-term fluctuation and make a gradual seasonal pattern easier to see. The smoothed values are calculated summaries, not additional measurements, so label them accordingly and keep the raw series available.
Imagine a temperature record with a brief sharp drop followed by a quick recovery. A wide averaging window includes observations from before and after the event, softening the drop and spreading its influence across nearby dates. The resulting line may look calmer, but the event has not disappeared from the underlying record.
State the window and alignment
Report the window length in a meaningful unit, such as seven days or a specified number of hourly observations. The same number of points covers different amounts of time if sampling intervals vary. Also say whether the average is centered on each date or uses only current and previous observations. These choices affect how a feature appears and where it falls in time.
At the beginning and end of a record, a centered window may not have enough observations on both sides. Some workflows shorten the window, leave endpoints blank, or use a different convention. Make the choice explicit, because endpoint handling can create misleading edge behavior.
Compare smoothed and raw views
Show the raw data in a lighter line or points alongside the smoother, or place the views in aligned panels. This lets readers see both short-lived variability and the broader shape. If the raw record contains missing intervals, explain how the smoothing calculation handled them; a window spanning a long gap should not quietly imply continuous data.
Try more than one reasonable window during exploration to understand which features depend on the choice. If a trend appears only under a very wide window, describe it as a broad smoothed pattern rather than implying the raw observations themselves trace a smooth trajectory.
Match smoothing to the question
Smoothing is useful when the question concerns a broad time scale, such as whether a seasonal rise is visible beyond day-to-day fluctuation. It is less suitable when brief events, extremes, or timing of transitions are central. For those questions, raw values and event annotations should remain prominent.
A good caption identifies the raw and smoothed series, the window and alignment, and any special handling of missing observations. Readers can then judge what the moving average clarifies and what it suppresses, instead of mistaking a display choice for a natural feature of the coast.
Have a correction or a source to suggest? Send a note →