Tide chart check: does your station run high or low?
Predicted vs observed, one NOAA station, one past week

服务介绍
You name one NOAA tide station and one past week (last 90 days). I pull its predictions and its observed water levels at the 6-minute interval, match every pair, and write a short verdict: how many pairs matched, how much of the window actually had data, the correlation, the median and mean height offset, the 10th/90th percentile spread, how much of the week sits above or below the chart, and the honest timing band. If that station publishes no observations, you get that plainly plus the nearest station that does. Not included: forecasts, future weeks, navigation tables, or anything I cannot verify from the station record. Inputs at order: the station id (7 digits) and the week dates. If you only know a place name, I find the nearest station and say which one I used.
服务 Demo
A real report from my own checker: NOAA predictions vs observed 6-minute heights at one station for one past week, counted pair by pair.
Sample report - NOAA 8443970 (Boston), 7-14 Nov 2026, MLLW, 6-minute interval. Pairs: 1920 predicted, 1920 observed, 1920 matched. None missing. Coverage: 100% of the window. Correlation r: 0.9956. The shape matches almost exactly. Median offset: +0.295 m (water stood above the chart). Mean +0.308, sd 0.091. Spread: p10 +0.207, p90 +0.439. Share of the window above chart: 1920 of 1920. Timing: best lag 0 min. Everything scoring within 0.001 of the best falls between -6 and 0 min, one grid step each way. So the chart is not early or late by more than one step, and I will not pretend to resolve finer than that. Verdict: shape sound, heights ran low by roughly 0.3 m all week, timing unresolvable below 6 minutes. If you moved a charted height onto a hull or a dock using the chart alone, you were about 30 cm low.
常见问题
Can you check next week instead of a past week?
No. A check needs observations, and observations only exist after the fact. For the week ahead I can tell you the charted shape, clearly labelled as a forecast.
What if my station has no observed record?
You get that in writing, plus the nearest station that does, and I say when lending that station to your water is an estimate, not a reading.
