Minnesota ice-out dates, 1950 to 2026: what 65 long-record lakes show

The short answer

Across 65 Minnesota lakes with at least 50 years of reports between 1950 and 2026, the day the ice went out moved earlier by about 1.37 days per decade. The 95% range around that is 0.4 to 2.35 days earlier per decade. Over 77 years that adds up to roughly 10.4 days earlier, with a range of 3.1 to 17.8 days.

That is a description of these lakes and these volunteer reports. It is not a measure of every Minnesota lake, and it cannot say why the date moved.

Three findings

1. The pooled trend is earlier, and the range is wide. The straight-line trend is −1.4 days per decade (95% range −2.4 to −0.4). Leaving out 1950, a cold spring that sits at the very start of the record, gives −1.2 (−2.2 to −0.2). Starting in 1970 gives −1.5, but the range then runs from 3.13 days earlier to 0.2 days later, so a flat line from 1970 on is not ruled out.
2. Every lake leans the same way, but not every lake is clear on its own. All 65 of the 65 lake-by-lake trends point earlier (median −1.3 days per decade, from −2.9 to −0.1). For 36 of the 65 the 95% range stays below zero. For the other 29 a single lake's record is too noisy to tell on its own.
3. The north is later, and the data cannot separate the two halves' trends. Typical ice-out is April 24 for the 24 northern lakes and April 10 for the 41 southern lakes, 14 days apart. The northern trend is −1.0 days per decade (−1.9 to −0.1) and the southern trend is −1.6 (−2.7 to −0.5). Those ranges overlap, so this data does not show that one half is changing faster.

Decade by decade

The decade table is easier to read than a single line. Ice-out ran later than each lake's own average in the 1950s, 1960s and 1970s, and earlier in every decade since. The shift looks more like a step around the 1980s than a smooth slope. That is what the table shows, not something we tested.

The 2020 to 2026 row is the one to handle most carefully. It covers seven years and 46 lakes, so a few early years weigh heavily in its southern figure (−8.3 days): 28 of the 65 lakes set their earliest ice-out on record in 2024.

Earliest and latest

The earliest ice-out in the sample is Mar 1, 2024 on Shields Lake. The latest is May 25, 1989 on Lower Hay Lake. Across lakes, the earliest date on record fell in 58 of 65 lakes in 2000 or later, and 26 of 65 set their latest date before 1980. Those counts are lumpy: 28 lakes share 2024 as their earliest year and 23 share 2013 as their latest, because one warm or cold spring hits many lakes at once.

Planning around the fishing opener? Our opener study joins these ice-out reports with the opener date for each year since 1999.

How we did this

Source

The Minnesota DNR ice-out page, which publishes dates reported by volunteer observers. The DNR says its own medians use 1950 to the present.

Collected

Nov 6, 2026. The lake list and each lake's file were downloaded that day.

Lakes

The DNR index lists 654 lakes. We downloaded the 105 with at least 50 entries in the index, then kept the 65 with at least 50 different years of reports from 1950 to 2026.

Why 50 years

A long record lets a lake's own average mean something and keeps the lake set steady from decade to decade.

One value per lake-year

Several agencies report some lakes, so 3,310 of 10,155 rows repeat another row. Where sources gave different dates for the same lake and year, we used the middle one (or the earlier of two). The sources disagreed in 579 lake-years, by up to 38 days.

Pooling

For each lake we found its own average date over 1950 to 2026, then measured each year against that average. A year's value is the mean of those gaps across lakes reporting that year, counted only when at least 10 lakes reported.

Trend

An ordinary least-squares line through the yearly values, with a 95% range from the standard error. We also checked whether neighboring years move together, and widened the range if they did (they did not). Each lake also has its own line.

North and south

North means latitude 46.0 degrees or higher. That is a convenient cut, not an official region.

Checks on the result

Limits

• Volunteer reports. Observers decide what counts as ice-out, and the DNR does not publish the definition. Different observers, or a change of observer on one lake, could shift dates by days without the lake changing.
• Not a random sample. Lakes with long records tend to be those near people and cabins. They are 65 of 654 lakes in the DNR index, and say little about remote or very small lakes.
• Gaps. 84.1% of possible lake-years have a report (4,209 of 5,005). Coverage was lowest in the 1950s (63.1%) and highest in the 1990s, and is 71.6% for 2020 to 2026.
• A straight line is a summary. It fits a pattern that looks more like a step than a slope. One extreme year at the start (1950) pulls on it, which is why we show the 1951 and 1970 versions.
• No cause is tested here. The DNR's climate pages report that Minnesota lakes have lost an average of 10 to 14 days of ice cover over the past 50 years. Ice cover (freeze-up to ice-out) is a different measure from ice-out alone, and nothing in this dataset can show the cause of either.

What we could not collect

• Lakes with fewer than 50 years of reports. 549 lakes in the index have shorter listings and were not downloaded. 39 downloaded lakes had too few years in the window and were left out.
• Ice-in (freeze-up) dates, ice thickness, and any weather or water-temperature data to set beside the dates.
• Any record of how each observer defined ice-out.
• Lakes outside the DNR index, including border lakes reported by other agencies.

Every lake in the study

We publish the statistics above, not the DNR's individual reports. Its data license allows derived work but not redistributing the data in its entirety.