Duluth Water Temperature by Season: What the 2021–2025 Station Record Shows

A short, screened record from one named station

This analysis uses the 2021–2025 NOAA water-temperature observations retained for Duluth station 9099064, with the site's seasonal summaries checked on October 23, 2026. NOAA's data API identifies water temperature as a station measurement. The location and identity matter: calling these figures simply 'Lake Superior averages' would extend the observations beyond what was measured.

The calculation groups source-quality-screened observations into UTC calendar hours. A month in a particular year qualifies only when at least 70% of its possible hours are represented. A calendar month needs at least three qualifying years before publication. The displayed mean pools the qualifying hourly values, so it is an observation-weighted mean, not an equal-weight average of annual monthly means.

The accompanying 10th–90th percentile band summarizes the distribution of those retained hourly values. It is a description of historical variability across the qualifying observations. It is not a confidence interval around the mean, a prediction interval for an upcoming visit or a measure of instrument accuracy.

Source: NOAA CO-OPS water-temperature observations, summarized by Water Temperature Live. Values rounded to one decimal; conversions use the unrounded Celsius means. This fixed research snapshot may differ from later backfilled summaries.

The local record rises into August and remains warm in September

The difference between the May and August means is 8.2°C, or 14.8°F. That calculation describes the seasonal contrast between two pooled monthly summaries. It does not mean the water warmed continuously: individual hours and individual years can depart substantially from the sequence of monthly means.

August is the highest qualified mean in this snapshot. September averages 17.6°C, only 0.8°C below August after rounding the difference. Its mean is also very close to July's. These relationships support a narrow conclusion: the local baseline retains much of its summer warmth into September. They cannot establish the warmest day, the peak in every year or the peak across the open lake.

The percentile bands also overlap. July and August share a 19.7°C upper percentile, while September's 16.3–18.5°C band overlaps both. Thus 'August is warmer on average' does not make every August observation warmer than every September observation. Retaining the distribution beside the mean makes that distinction visible.

Missing winter rows cannot diagnose ice conditions

December through April do not meet the publication rule in this retained baseline. The appropriate table entry is 'Insufficient data.' Filling those cells with zero would invent measurements; connecting November directly to May would hide a substantial gap. The site's collection and coverage rules do not deliberately exclude winter months.

There is a second gap worth preserving: May and November have four qualifying years, with 2022 absent from each. The table does not treat that missing year as an average year. Differences in coverage can affect the pooled distributions, which is why the year list is part of the result rather than an optional footnote.

A missing temperature summary alone cannot establish whether the sensor was surrounded by ice, whether the open lake froze, or why the source record was incomplete. Those are different questions requiring operational records or an independent ice product. The absence of winter means also prevents calculation of a defensible full-year mean from this table.

How the pattern fits broader Lake Superior research

Lake Superior's size and depth make a local surface or nearshore record only one part of its thermal story. EPA lists its average depth as 147 meters. That geographic fact is useful context, but it supplies neither the sensor depth nor a correction that converts the Duluth series into a lakewide mean. EPA's lake dimensions describe the larger setting.

Historical shipboard research found late-summer maxima in Superior's upper-water temperatures and deeper mixing during autumn. It also distinguished nearshore and midlake thermal regimes. Those findings make seasonal heat retention physically plausible, but the surveys covered other years, locations and depths; they do not validate each Duluth figure. NOAA's summary of Assel's 1986 thermal-structure study preserves those distinctions.

For present conditions, use the separately updated Duluth observation page and read its timestamp. For a lakewide comparison, use a lakewide surface product with its own reference period. This article's fixed five-year station summary is most useful as local seasonal context, with its missing months kept in view.