2024-2025 La Nina

[michsnowfreak]

21 minutes ago, LibertyBell said:

wow I'm surprised all I heard coming out of the Midwest is how historic the arctic outbreak in February was and that was what brought our winter back after the January thaw,  Over here in NYC we were below normal all three winter months (and March too)  I was surprised to see it was only a little below normal with temperatures.

It seems that here in the NE our snowfall is more dependent on storm track and less on temperatures. Even here in the southernmost part of the NE. And that's what most winter lovers care about.

 

1995-96 was a colder than avg winter in the midwest, but nothing noteworthy (at least here) outside of the early February cold snap, which still paled in comparisons to multiple cold snaps since then. That cold snap was noteworthy in chicago and Detroit in that it came with bare ground. One of my least favorite winters here in terms of missing all the action. Cold and dry ruled, and Detroit has easily never come as close to getting screwed in the 29 years since.

1995-96 ranks as detroits 44th coldest (of 151 years) winter on record. Since then, the following winters have been colder:

2000-01
2002-03
2008-09
2010-11
2013-14
2014-15

[michsnowfreak]

3 hours ago, LibertyBell said:

I'm shocked 1993-94, 1995-96, 2002-03, 2003-04, 2004-05, 2010-11, 2013-14, 2014-15 aren't in the top 20 coldest list.

A poster in the lakes forum has awesome monthly graphs ranking midwest/lakes climate back to 1807. 

In the Great Lakes/Midwest:

the top 5 coldest winters

1.) 1903-04

2.) 1978-79

3.) 1884-85

4.) 2013-14

5.) 1874-75

 

Top 5 warmest winters

1.) 1877-78

2.) 2023-24

3.) 1931-32

4.) 2001-02

5.) 1997-98

[LibertyBell]

10 minutes ago, michsnowfreak said:

1995-96 was a colder than avg winter in the midwest, but nothing noteworthy (at least here) outside of the early February cold snap, which still paled in comparisons to multiple cold snaps since then. That cold snap was noteworthy in chicago and Detroit in that it came with bare ground. One of my least favorite winters here in terms of missing all the action. Cold and dry ruled, and Detroit has easily never come as close to getting screwed in the 29 years since.

1995-96 ranks as detroits 44th coldest (of 151 years) winter on record. Since then, the following winters have been colder:

2000-01
2002-03
2008-09
2010-11
2013-14
2014-15

Many of those other winters had a lot of snow here, so I'm surprised you got screwed in 1995-96.

And it was a la nina too!

 

[LibertyBell]

Just now, michsnowfreak said:

A poster in the lakes forum has awesome monthly graphs ranking midwest/lakes winters back to 1807. 

In the Great Lakes/Midwest:

the top 5 coldest winters

1.) 1903-04

2.) 1978-79

3.) 1884-85

4.) 2013-14

5.) 1874-75

 

Top 5 warmest winters

1.) 1877-78

2.) 2023-24

3.) 1931-32

4.) 2001-02

5.) 1997-98

Those do match for here.... how did you do in 1917-18, that was our coldest, the rest are similar.

Warmest is almost a perfect match, surprised that number 1 is a winter from the 1800s though lol. How much snow did you get that winter?

 

[bluewave]

19 minutes ago, michsnowfreak said:

But what I'm saying is noting how cold or warm the conus is is totally irrelevant to how someone perceived their winter any individual location. It goes without saying that how cold anyone felt their winter was at their location is best gauged by how it compares to their climate record and their recent winters, not the CONUS. A cold winter by new york city 1951-80 standards would still be a mild winter in Detroit by 1991-20 standards. To say nothing of the recent cold winters in already cold upper midwest/plains and the warmth in already warm snowless places like Florida and Arizona. 

Local perceptions of what is considered cold are irrelevant to local climate records since they mostly form their perception based on recent years which have been much warmer. So recent perceptions will naturally lack the historical perspective of what cold winters used to be like. Fewer and fewer local areas are having historically cold winters any more just like the CONUS wider averages reflect. 

[GaWx]

5 minutes ago, LibertyBell said:

Those do match for here.... how did you do in 1917-18, that was our coldest, the rest are similar.

Warmest is almost a perfect match, surprised that number 1 is a winter from the 1800s though lol. How much snow did you get that winter?

 

1877-8 was one of the, if not the, strongest El Niño winters on record.

[LibertyBell]

6 minutes ago, GaWx said:

1877-8 was one of the, if not the, strongest El Niño winters on record.

wow, how did they know it was an el nino back then and how did they measure it? That's absolutely fascinating.

[GaWx]

14 minutes ago, LibertyBell said:

wow, how did they know it was an el nino back then and how did they measure it? That's absolutely fascinating.

There is SST data of the tropical Pacific back to the 1800s. So, based on that data, scientists have been able to retrospectively classify each year’s ENSO at least roughly.

This is a link to Eric Webb’s Nino 3.4 table, which goes way back to 1850:

https://www.webberweather.com/ensemble-oceanic-nino-index.html

 

 Here’s a table from the JMA back to 1868. The JMA region isn’t the same as Nino 3.4. It uses a variation of Nino 3 (4N to 4S rather than 5N to 5S) and averaged over 5 months instead of 3 months:

https://data.coaps.fsu.edu/pub/JMA_SST_Index/jmasst1868-today.filter-5

 

[chubbs]

3 hours ago, michsnowfreak said:

Most climate site deals with moves over the course of the climate period. Just one of those things in the climate record. Roads were dirt in the 1870s. DTW airport had much better radiational cooling in the 1960s-80s before the airport expanded to its current status. You just have to deal with the changes as part of the climate record, rather than pick and choose the ones we like and don't like. 

That's why I like NOAA (and other groups), who correct for inconsistency due to: station moves, measurement method changes, heat island, etc. Detroit winter warming looks very similar to other US cities.

[PhiEaglesfan712]

2 hours ago, LibertyBell said:

Those do match for here.... how did you do in 1917-18, that was our coldest, the rest are similar.

Warmest is almost a perfect match, surprised that number 1 is a winter from the 1800s though lol. How much snow did you get that winter?

 

1917-18 was the last of a triple la nina, and it piggy backed off the very strong la nina of 1916-17. Explains why global temperature bottomed out in 1917, and why many locations have a coldest winter in 1916-17 or 1917-18. (Global temperatures in 2024 was over 2C warmer than 1917. It's no surprise many of those same locations have a warmest winter of 2023-24.)

[LibertyBell]

1 hour ago, PhiEaglesfan712 said:

1917-18 was the last of a triple la nina, and it piggy backed off the very strong la nina of 1916-17. Explains why global temperature bottomed out in 1917, and why many locations have a coldest winter in 1916-17 or 1917-18. (Global temperatures in 2024 was over 2C warmer than 1917. It's no surprise many of those same locations have a warmest winter of 2023-24.)

La Ninas used to be much colder and snowier back then.  As a matter of fact when 2010-11 happened the analog people came up with was that one.

[michsnowfreak]

6 hours ago, LibertyBell said:

Many of those other winters had a lot of snow here, so I'm surprised you got screwed in 1995-96.

And it was a la nina too!

 

The Chicago to Detroit corridor was very screwed in 1995-96. Would've been even worse if not for a march 20th snowstorm. Of those winters mentioned, the other ones were much better winters. 

1995-96: 27.6"

2000-01: 39.0"
2002-03: 60.9"
2008-09: 65.7"
2010-11: 69.1"
2013-14: 94.9"
2014-15: 47.5"

[michsnowfreak]

6 hours ago, LibertyBell said:

Those do match for here.... how did you do in 1917-18, that was our coldest, the rest are similar.

Warmest is almost a perfect match, surprised that number 1 is a winter from the 1800s though lol. How much snow did you get that winter?

 

1917-18 ranks as the 6th coldest, and 1918-19 ranks 13th warmest. Quite a rubber band effect.

1877-78 is widespread known as "year without a winter" in the upper midwest. We actually did ok with avg snowfall here (43.4") but as I've said before temps and snowfall here don't always go hand in hand. Locally the winter of 1881-82 was the worst for warmth + lack of snow. That remains Detroits warmest winter on record.

The period of 1875-1882 was absolutely wild here with every other year alternating between bitterly cold winters and very warm winters. Not just talking either side of normal. Talking all winters that still currently reside in the top 20 coldest/warmest list.

[michsnowfreak]

4 hours ago, chubbs said:

That's why I like NOAA (and other groups), who correct for inconsistency due to: station moves, measurement method changes, heat island, etc. Detroit winter warming looks very similar to other US cities.

I've never heard of these. Is there a link to look at this data? Not sure about loess curves. It implies starting 1896 with a mean of what, 23°? The mean temp from 1874-1896 was 26.9°, when there was literally only one accurate wayne county station.

Regardless I don't dispute that winters have warmed. They've also gotten snowier and wetter. This is why I like looking at things like the "winter severity index" which takes all things into account. I have studied local winters extensively and certainly concentrate most heavily on snow and snowcover. I can absolutely tell you that 1970s-80s and 2000s-10s winters I'll take in a heartbeat. Hard pass on 1930s-50s winters. In the 1870s-1920s era I'd take majority of them but there are a handful of UGLY ones thrown in I'd need to ditch lol.

[GaWx]

It’s unanimous for a 2nd and 3rd strong dip (both would set new daily records on and near those dates):

[Maestrobjwa]

1 hour ago, GaWx said:

It’s unanimous for a 2nd and 3rd strong dip (both would set new daily records on and near those dates):

Wow you're like...really committed to zonal winds, lol

[GaWx]

38 minutes ago, Maestrobjwa said:

Wow you're like...really committed to zonal winds, lol

 Indeed, I am! I find wx data that approaches or sets new records, like for these zonal winds, extra interesting to say the least. Other examples: record obliterating DJF +PNA for non-Nino ‘24-‘25, -5 AO in Feb ‘25, 1/21-2/2025 heaviest sleetstorm on record in my area, etc. My professional and educational background is statistics related in general, which I’ve enjoyed following since being a kid. Since I’m a big wx enthusiast in addition to being a big statistics enthusiast, wx stats are a natural for me to enjoy following.

[chubbs]

8 hours ago, michsnowfreak said:

I've never heard of these. Is there a link to look at this data? Not sure about loess curves. It implies starting 1896 with a mean of what, 23°? The mean temp from 1874-1896 was 26.9°, when there was literally only one accurate wayne county station.

Regardless I don't dispute that winters have warmed. They've also gotten snowier and wetter. This is why I like looking at things like the "winter severity index" which takes all things into account. I have studied local winters extensively and certainly concentrate most heavily on snow and snowcover. I can absolutely tell you that 1970s-80s and 2000s-10s winters I'll take in a heartbeat. Hard pass on 1930s-50s winters. In the 1870s-1920s era I'd take majority of them but there are a handful of UGLY ones thrown in I'd need to ditch lol.

Here's the NOAA link. I choose the lowess fit because, as you noted, there are warm and cool periods before 1970 that aren't fit well by a straight line. The fit only uses the data points plotted in the NOAA series, i.e. 1895 and later. If there was any change before 1895 wouldn't be reflected in the Lowess fit.  

In a quick search, found 2 local stations with long-term data without the station moves in the Detroit series: UM at Ann Arbor and Pontiac. NOAA used these stations and many more for the Wayne County series. Weather data is correlated for hundreds of miles so stations outside of Wayne County provide useful information. I spent 3 years in NW Ohio 1978-81 so can attest to how cold midwest winters were 45 years ago.

https://www.ncei.noaa.gov/access/monitoring/climate-at-a-glance/national/time-series

[LibertyBell]

9 hours ago, michsnowfreak said:

1917-18 ranks as the 6th coldest, and 1918-19 ranks 13th warmest. Quite a rubber band effect.

1877-78 is widespread known as "year without a winter" in the upper midwest. We actually did ok with avg snowfall here (43.4") but as I've said before temps and snowfall here don't always go hand in hand. Locally the winter of 1881-82 was the worst for warmth + lack of snow. That remains Detroits warmest winter on record.

The period of 1875-1882 was absolutely wild here with every other year alternating between bitterly cold winters and very warm winters. Not just talking either side of normal. Talking all winters that still currently reside in the top 20 coldest/warmest list.

We had some amazing winters here back then too, culminating in the great Blizzard of 1888 (that season had many historic blizzards nationwide.) 

The 1800s were pretty wild overall, in 1896 we had our only month with over 30 inches of snow (in March) followed by a historic 10 day super heatwave that killed over 1,500 people in August.  That heatwave propelled Teddy Roosevelt into the presidency as he was the police commissioner here and allowed people to sleep in parks and turn on fire hydrants to cool off for the first time.  And then of course we had the amazing winter of 1898-1899 that culminated in one of the greatest arctic outbreaks and blizzards of all time that dumped snow from Tampa to NJ and 3 feet of snow in the southernmost point of New Jersey, Cape May.

[bluewave]

9 hours ago, michsnowfreak said:

Regardless I don't dispute that winters have warmed. They've also gotten snowier and wetter.

While we have had several epic snowfall seasons from the mid 1990s into the 2010s, earlier eras measured snowfall differently. So the seasons prior to this period undermeasured snow compared to the way we do now. So correcting these undercounts will show heavier snows from the late 1800s into 1980s. This would lead to a longer term decline snowfall when the earlier totals are raised. So places that don’t show any long term trend will show a decline when the measurements are standardized. And some areas which already have long term declines will show even steeper declines. Plus areas which show longer term increases may shift to not much trend when the earlier era snowfall is increased. 
 

https://news.ucar.edu/14009/snowfall-measurement-flaky-history

As a hydrometeorological instructor in UCAR’s COMET program and a weather observer for the National Weather Service, I am keenly interested in weather trends. In this case, climate change is an important factor to explore, since we know that the heaviest precipitation events have intensified in many parts of the world (see related story: Torrents and droughts and twisters - oh my!).

But when we turn to snowstorms in the Northeast, or elsewhere in the U.S., there is an additional factor at work when comparing modern numbers with historical ones. Quite simply, our measuring techniques have changed, and we are not necessarily comparing apples to apples. In fact, the apparent trend toward bigger snowfalls is at least partially the result of new—and more accurate—ways of measuring snowfall totals. Climate studies carefully select a subset of stations with consistent snow records, or avoid the snowfall variable altogether.

Official measurement of snowfall these days uses a flat, usually white, surface called a snowboard (which pre-dates the popular winter sport equipment of the same name). The snowboard depth measurement is done ideally every 6 hours, but not more frequently, and the snow is cleared after each measurement. At the end of the snowfall, all of the measurements are added up for the storm total. 

NOAA’s cooperative climate observers and thousands of volunteers with the Community Collaborative Rain, Hail and Snow (CoCoRaHS), a nationwide observer network, are trained in this method. This practice first became standard at airports starting in the 1950s, but later at other official climate reporting sites, such as Manhattan’s Central Park, where 6-hourly measurements did not become routine until the 1990s.

Earlier in our weather history, the standard practice was to record snowfall amounts less frequently, such as every 12 or 24 hours, or even to take just one measurement of depth on the ground at the end of the storm.

You might think that one or two measurements per day should add up to pretty much the same as measurements taken every 6 hours during the storm. It’s a logical assumption, but you would be mistaken. Snow on the ground gets compacted as additional snow falls. Therefore, multiple measurements during a storm typically result in a higher total than if snowfall is derived from just one or two measurements per day.

That can make quite a significant difference. It turns out that it’s not uncommon for the snow on the ground at the end of a storm to be 15 to 20 percent less than the total that would be derived from multiple snowboard measurements.  As the cooperative climate observer for Boulder, Colorado, I examined the 15 biggest snowfalls of the last two decades, all measured at the NOAA campus in Boulder. The sum of the snowboard measurements averaged 17 percent greater than the maximum depth on the ground at the end of the storm. For a 20-inch snowfall, that would be a boost of 3.4 inches—enough to dethrone many close rivals on the top-10 snowstorm list that were not necessarily lesser storms!

Another common practice at the cooperative observing stations prior to 1950 did not involve measuring snow at all, but instead took the liquid derived from the snow and applied a 10:1 ratio (every inch of liquid equals ten inches of snow). This is no longer the official practice and has become increasingly less common since 1950. But it too introduces a potential low bias in historic snowfalls because in most parts of the country (and in the recent blizzard in the Northeast) one inch of liquid produces more than 10 inches of snow.

This means that many of the storms from the 1980s or earlier would probably appear in the record as bigger storms if the observers had used the currently accepted methodology. Now, for those of you northeasterners with aching backs from shoveling, I am not saying that your recent storm wasn’t big in places like Boston, Portland, or Long Island. But I am saying that some of the past greats—the February Blizzard of 1978, the Knickerbocker storm of January 1922, and the great Blizzard of March 1888—are probably underestimated.

So keep in mind when viewing those lists of snowy greats: the older ones are not directly comparable with those in recent decades. It’s not as bad as comparing apples to oranges, but it may be like comparing apples to crabapples.

Going forward, we can look for increasingly accurate snow totals. Researchers at NCAR and other organizations are studying new approaches for measuring snow more accurately (see related story: Snowfall, inch by inch).  

But we can’t apply those techniques to the past. For now, all we can say is that snowfall measurements taken more than about 20 or 30 years ago may be unsuitable for detecting trends – and perhaps snowfall records from the past should not be melting away quite as quickly as it appears

 

[LibertyBell]

13 minutes ago, bluewave said:

While we have had several epic snowfall seasons from the mid 1990s into the 2010s, earlier eras measured snowfall differently. So the seasons prior to this period undermeasured snow compared to the way we do now. So correcting these undercounts will show heavier snows from the late 1800s into 1980s. This would lead to a longer term decline snowfall when the earlier totals are raised. So places that don’t show any long term trend will show a decline when the measurements are standardized. And some areas which already have long term declines will show even steeper declines. Plus areas which show longer term increases may shift to not much trend when the earlier era snowfall is increased. 
 

https://news.ucar.edu/14009/snowfall-measurement-flaky-history

As a hydrometeorological instructor in UCAR’s COMET program and a weather observer for the National Weather Service, I am keenly interested in weather trends. In this case, climate change is an important factor to explore, since we know that the heaviest precipitation events have intensified in many parts of the world (see related story: Torrents and droughts and twisters - oh my!).

But when we turn to snowstorms in the Northeast, or elsewhere in the U.S., there is an additional factor at work when comparing modern numbers with historical ones. Quite simply, our measuring techniques have changed, and we are not necessarily comparing apples to apples. In fact, the apparent trend toward bigger snowfalls is at least partially the result of new—and more accurate—ways of measuring snowfall totals. Climate studies carefully select a subset of stations with consistent snow records, or avoid the snowfall variable altogether.

Official measurement of snowfall these days uses a flat, usually white, surface called a snowboard (which pre-dates the popular winter sport equipment of the same name). The snowboard depth measurement is done ideally every 6 hours, but not more frequently, and the snow is cleared after each measurement. At the end of the snowfall, all of the measurements are added up for the storm total. 

NOAA’s cooperative climate observers and thousands of volunteers with the Community Collaborative Rain, Hail and Snow (CoCoRaHS), a nationwide observer network, are trained in this method. This practice first became standard at airports starting in the 1950s, but later at other official climate reporting sites, such as Manhattan’s Central Park, where 6-hourly measurements did not become routine until the 1990s.

Earlier in our weather history, the standard practice was to record snowfall amounts less frequently, such as every 12 or 24 hours, or even to take just one measurement of depth on the ground at the end of the storm.

You might think that one or two measurements per day should add up to pretty much the same as measurements taken every 6 hours during the storm. It’s a logical assumption, but you would be mistaken. Snow on the ground gets compacted as additional snow falls. Therefore, multiple measurements during a storm typically result in a higher total than if snowfall is derived from just one or two measurements per day.

That can make quite a significant difference. It turns out that it’s not uncommon for the snow on the ground at the end of a storm to be 15 to 20 percent less than the total that would be derived from multiple snowboard measurements.  As the cooperative climate observer for Boulder, Colorado, I examined the 15 biggest snowfalls of the last two decades, all measured at the NOAA campus in Boulder. The sum of the snowboard measurements averaged 17 percent greater than the maximum depth on the ground at the end of the storm. For a 20-inch snowfall, that would be a boost of 3.4 inches—enough to dethrone many close rivals on the top-10 snowstorm list that were not necessarily lesser storms!

Another common practice at the cooperative observing stations prior to 1950 did not involve measuring snow at all, but instead took the liquid derived from the snow and applied a 10:1 ratio (every inch of liquid equals ten inches of snow). This is no longer the official practice and has become increasingly less common since 1950. But it too introduces a potential low bias in historic snowfalls because in most parts of the country (and in the recent blizzard in the Northeast) one inch of liquid produces more than 10 inches of snow.

This means that many of the storms from the 1980s or earlier would probably appear in the record as bigger storms if the observers had used the currently accepted methodology. Now, for those of you northeasterners with aching backs from shoveling, I am not saying that your recent storm wasn’t big in places like Boston, Portland, or Long Island. But I am saying that some of the past greats—the February Blizzard of 1978, the Knickerbocker storm of January 1922, and the great Blizzard of March 1888—are probably underestimated.

So keep in mind when viewing those lists of snowy greats: the older ones are not directly comparable with those in recent decades. It’s not as bad as comparing apples to oranges, but it may be like comparing apples to crabapples.

Going forward, we can look for increasingly accurate snow totals. Researchers at NCAR and other organizations are studying new approaches for measuring snow more accurately (see related story: Snowfall, inch by inch).  

But we can’t apply those techniques to the past. For now, all we can say is that snowfall measurements taken more than about 20 or 30 years ago may be unsuitable for detecting trends – and perhaps snowfall records from the past should not be melting away quite as quickly as it appears

 

I compiled a list of snowstorms that were probably underestimated. Beginning with January 1996 and working our way backwards to include February 1978, December 1947, February 1920, February 1899 and March 1888.

Also keep in mind many of our borderline events that switched back and forth between precipitation types were also likely not measured properly because of this behavior of only measuring when an event ended, with lots of melting occurring before that happened.

 

[michsnowfreak]

8 hours ago, GaWx said:

 Indeed, I am! I find wx data that approaches or sets new records, like for these zonal winds, extra interesting to say the least. Other examples: record obliterating DJF +PNA for non-Nino ‘24-‘25, -5 AO in Feb ‘25, 1/21-2/2025 heaviest sleetstorm on record in my area, etc. My professional and educational background is statistics related in general, which I’ve enjoyed following since being a kid. Since I’m a big wx enthusiast in addition to being a big statistics enthusiast, wx stats are a natural for me to enjoy following.

I too am a huge wx stat fan. Its easily my #1 favorite aspect of weather. I have so many different charts/graphs/papers in folders detailing Detroits weather history from the beginning of record to present, much of which never comes up to be posted here. I also browse other areas stats as well, whenever something comes up that makes me wonder "I wonder how this compares to past events there". My #2 aspect would be actually observering the weather. Been a wx observer since 2004 and measured snow since 1995. A distant #3 is the favorite of many, the actual tracking with this often headache-inducing model data 

[michsnowfreak]

2 hours ago, chubbs said:

Here's the NOAA link. I choose the lowess fit because, as you noted, there are warm and cool periods before 1970 that aren't fit well by a straight line. The fit only uses the data points plotted in the NOAA series, i.e. 1895 and later. If there was any change before 1895 wouldn't be reflected in the Lowess fit.  

In a quick search, found 2 local stations with long-term data without the station moves in the Detroit series: UM at Ann Arbor and Pontiac. NOAA used these stations and many more for the Wayne County series. Weather data is correlated for hundreds of miles so stations outside of Wayne County provide useful information. I spent 3 years in NW Ohio 1978-81 so can attest to how cold midwest winters were 45 years ago.

https://www.ncei.noaa.gov/access/monitoring/climate-at-a-glance/national/time-series

Thanks. Ill bookmark it. I like looking at the Ann Arbor data because they have a consistent record to 1880 and are only about 20 miles due west of DTW, with their weather observing site not changing since 1880. The only problem with that, and im sure its a minor one, is they are a coop site, not first order, so there are bound to be a few errors here and there.

You really did pick a doozy 3-year time period to live in the region lol. That cold was brutal and not normal. At Toledo, the winters of '76-77, '77-78, '78-79 rank as 1st, 2nd, and 8th coldest (since 1874). At Detroit they rank 3rd, 6th, & 12th coldest (since 1874), and at Ann Arbor 4th, 10th, & 13th coldest (since 1881). Depending on the airmass, there are many situations where NW OH can actually be colder (esp at night) than SE MI, with more urban sprawl in SE MI and more rural areas in NW OH. Id suspect there were multiple instances of that in the late 1970s. These colder nights offset the days when warm fronts dont get as far north as Detroit, and you have Toledo & Detroit winter temp averages nearly identical (DTW just a TAD colder), despite the fact that Detroit gets noticebly more snowfall and snowcover.

[michsnowfreak]

1 hour ago, bluewave said:

While we have had several epic snowfall seasons from the mid 1990s into the 2010s, earlier eras measured snowfall differently. So the seasons prior to this period undermeasured snow compared to the way we do now. So correcting these undercounts will show heavier snows from the late 1800s into 1980s. This would lead to a longer term decline snowfall when the earlier totals are raised. So places that don’t show any long term trend will show a decline when the measurements are standardized. And some areas which already have long term declines will show even steeper declines. Plus areas which show longer term increases may shift to not much trend when the earlier era snowfall is increased. 
 

https://news.ucar.edu/14009/snowfall-measurement-flaky-history

As a hydrometeorological instructor in UCAR’s COMET program and a weather observer for the National Weather Service, I am keenly interested in weather trends. In this case, climate change is an important factor to explore, since we know that the heaviest precipitation events have intensified in many parts of the world (see related story: Torrents and droughts and twisters - oh my!).

But when we turn to snowstorms in the Northeast, or elsewhere in the U.S., there is an additional factor at work when comparing modern numbers with historical ones. Quite simply, our measuring techniques have changed, and we are not necessarily comparing apples to apples. In fact, the apparent trend toward bigger snowfalls is at least partially the result of new—and more accurate—ways of measuring snowfall totals. Climate studies carefully select a subset of stations with consistent snow records, or avoid the snowfall variable altogether.

Official measurement of snowfall these days uses a flat, usually white, surface called a snowboard (which pre-dates the popular winter sport equipment of the same name). The snowboard depth measurement is done ideally every 6 hours, but not more frequently, and the snow is cleared after each measurement. At the end of the snowfall, all of the measurements are added up for the storm total. 

NOAA’s cooperative climate observers and thousands of volunteers with the Community Collaborative Rain, Hail and Snow (CoCoRaHS), a nationwide observer network, are trained in this method. This practice first became standard at airports starting in the 1950s, but later at other official climate reporting sites, such as Manhattan’s Central Park, where 6-hourly measurements did not become routine until the 1990s.

Earlier in our weather history, the standard practice was to record snowfall amounts less frequently, such as every 12 or 24 hours, or even to take just one measurement of depth on the ground at the end of the storm.

You might think that one or two measurements per day should add up to pretty much the same as measurements taken every 6 hours during the storm. It’s a logical assumption, but you would be mistaken. Snow on the ground gets compacted as additional snow falls. Therefore, multiple measurements during a storm typically result in a higher total than if snowfall is derived from just one or two measurements per day.

That can make quite a significant difference. It turns out that it’s not uncommon for the snow on the ground at the end of a storm to be 15 to 20 percent less than the total that would be derived from multiple snowboard measurements.  As the cooperative climate observer for Boulder, Colorado, I examined the 15 biggest snowfalls of the last two decades, all measured at the NOAA campus in Boulder. The sum of the snowboard measurements averaged 17 percent greater than the maximum depth on the ground at the end of the storm. For a 20-inch snowfall, that would be a boost of 3.4 inches—enough to dethrone many close rivals on the top-10 snowstorm list that were not necessarily lesser storms!

Another common practice at the cooperative observing stations prior to 1950 did not involve measuring snow at all, but instead took the liquid derived from the snow and applied a 10:1 ratio (every inch of liquid equals ten inches of snow). This is no longer the official practice and has become increasingly less common since 1950. But it too introduces a potential low bias in historic snowfalls because in most parts of the country (and in the recent blizzard in the Northeast) one inch of liquid produces more than 10 inches of snow.

This means that many of the storms from the 1980s or earlier would probably appear in the record as bigger storms if the observers had used the currently accepted methodology. Now, for those of you northeasterners with aching backs from shoveling, I am not saying that your recent storm wasn’t big in places like Boston, Portland, or Long Island. But I am saying that some of the past greats—the February Blizzard of 1978, the Knickerbocker storm of January 1922, and the great Blizzard of March 1888—are probably underestimated.

So keep in mind when viewing those lists of snowy greats: the older ones are not directly comparable with those in recent decades. It’s not as bad as comparing apples to oranges, but it may be like comparing apples to crabapples.

Going forward, we can look for increasingly accurate snow totals. Researchers at NCAR and other organizations are studying new approaches for measuring snow more accurately (see related story: Snowfall, inch by inch).  

But we can’t apply those techniques to the past. For now, all we can say is that snowfall measurements taken more than about 20 or 30 years ago may be unsuitable for detecting trends – and perhaps snowfall records from the past should not be melting away quite as quickly as it appears

 

Unless I am a reincarnated weather observer who worked at the Weather Bureau in the early 20th century, I cant say anything for certain (none of us can) about any individual snowfall of the past. But what I CAN say is that I know the practice of measuring every 6 hours has been in place at Detroit since about 1950, and I assume the same for many other first-order stations. I highly doubt that 6-hour intervals didnt start in NYC til the 1980s. Airports began this in 1950. Also, one thing that sets apart snowfall around here vs the east coast is that we are a climate of smaller, more frequent snowfalls. We do not get big 18-24" storms that can inflate (or deflate) our totals in a 2-day period. Lots of T-1, 1-2, 2-4, 3-6" type snowfalls, then the occasional 7-12" type storms. Also, the old days of the weather bureau had to manually monitor weather hourly, so while most NWS offices have corrected the lazy measurement practices that took many places when the FAA took over in the 1990s, Ive heard countless complaints about observers calling a snowfall a T or something because it melted before obs time. That would NEVER have happened in the old days. 

Another thing I look at when analyzing old weather data for here is the liquid equivalent and snow depth numbers. Is the depth making sense (ie: is there some settling, or the following days snow depth always just the snowfall added onto the previous depth). For the liquid equivalent, are they just applying a 10:1 ratio or are they actually measuring liquid and snowfall separately? In the old days, 1870s-1890s, they usually (but not always) applied a 10:1 ratio to every snowfall. So, does this mean they measured snow and just applied 10:1, or even worse that the melted liquid but didnt measure the snow? No one will ever really know.

In the end, my analysis of snowfall data for Detroit, which dates to 1874 but the NWS begins in 1880 due to some M data in the 1870s...my assessment is that data is very accurate from around 1900 - present. Accurate does not mean there likely werent flaws, but looking at all factors (temps, precip, snowfall, snow depth, following days snow depth, etc) the data makes very good sense. Snowfall measurement practices from 1950-present have remained unchanged. Im not happy with some of the data from the mid-1990s to early 2000s (when the FAA took over and before snowpaid observers were hired), although none of it affected the record books in a huge way. And of course with many different obsevers, wind, etc, snow is always subject to error. Always has been, always will be. Snowfall from 1900-1949 seems quite accurate as well. Its is only the snowfall data from 1874-1899 that I look at with a grain of salt. You can still create a great mental picture of when the big storms were, when the snowy/cold winters were and when the warm, bare, "open winters" were, but the actual measurments likely had much more error than today.

[40/70 Benchmark]

2 hours ago, bluewave said:

While we have had several epic snowfall seasons from the mid 1990s into the 2010s, earlier eras measured snowfall differently. So the seasons prior to this period undermeasured snow compared to the way we do now. So correcting these undercounts will show heavier snows from the late 1800s into 1980s. This would lead to a longer term decline snowfall when the earlier totals are raised. So places that don’t show any long term trend will show a decline when the measurements are standardized. And some areas which already have long term declines will show even steeper declines. Plus areas which show longer term increases may shift to not much trend when the earlier era snowfall is increased. 
 

https://news.ucar.edu/14009/snowfall-measurement-flaky-history

As a hydrometeorological instructor in UCAR’s COMET program and a weather observer for the National Weather Service, I am keenly interested in weather trends. In this case, climate change is an important factor to explore, since we know that the heaviest precipitation events have intensified in many parts of the world (see related story: Torrents and droughts and twisters - oh my!).

But when we turn to snowstorms in the Northeast, or elsewhere in the U.S., there is an additional factor at work when comparing modern numbers with historical ones. Quite simply, our measuring techniques have changed, and we are not necessarily comparing apples to apples. In fact, the apparent trend toward bigger snowfalls is at least partially the result of new—and more accurate—ways of measuring snowfall totals. Climate studies carefully select a subset of stations with consistent snow records, or avoid the snowfall variable altogether.

Official measurement of snowfall these days uses a flat, usually white, surface called a snowboard (which pre-dates the popular winter sport equipment of the same name). The snowboard depth measurement is done ideally every 6 hours, but not more frequently, and the snow is cleared after each measurement. At the end of the snowfall, all of the measurements are added up for the storm total. 

NOAA’s cooperative climate observers and thousands of volunteers with the Community Collaborative Rain, Hail and Snow (CoCoRaHS), a nationwide observer network, are trained in this method. This practice first became standard at airports starting in the 1950s, but later at other official climate reporting sites, such as Manhattan’s Central Park, where 6-hourly measurements did not become routine until the 1990s.

Earlier in our weather history, the standard practice was to record snowfall amounts less frequently, such as every 12 or 24 hours, or even to take just one measurement of depth on the ground at the end of the storm.

You might think that one or two measurements per day should add up to pretty much the same as measurements taken every 6 hours during the storm. It’s a logical assumption, but you would be mistaken. Snow on the ground gets compacted as additional snow falls. Therefore, multiple measurements during a storm typically result in a higher total than if snowfall is derived from just one or two measurements per day.

That can make quite a significant difference. It turns out that it’s not uncommon for the snow on the ground at the end of a storm to be 15 to 20 percent less than the total that would be derived from multiple snowboard measurements.  As the cooperative climate observer for Boulder, Colorado, I examined the 15 biggest snowfalls of the last two decades, all measured at the NOAA campus in Boulder. The sum of the snowboard measurements averaged 17 percent greater than the maximum depth on the ground at the end of the storm. For a 20-inch snowfall, that would be a boost of 3.4 inches—enough to dethrone many close rivals on the top-10 snowstorm list that were not necessarily lesser storms!

Another common practice at the cooperative observing stations prior to 1950 did not involve measuring snow at all, but instead took the liquid derived from the snow and applied a 10:1 ratio (every inch of liquid equals ten inches of snow). This is no longer the official practice and has become increasingly less common since 1950. But it too introduces a potential low bias in historic snowfalls because in most parts of the country (and in the recent blizzard in the Northeast) one inch of liquid produces more than 10 inches of snow.

This means that many of the storms from the 1980s or earlier would probably appear in the record as bigger storms if the observers had used the currently accepted methodology. Now, for those of you northeasterners with aching backs from shoveling, I am not saying that your recent storm wasn’t big in places like Boston, Portland, or Long Island. But I am saying that some of the past greats—the February Blizzard of 1978, the Knickerbocker storm of January 1922, and the great Blizzard of March 1888—are probably underestimated.

So keep in mind when viewing those lists of snowy greats: the older ones are not directly comparable with those in recent decades. It’s not as bad as comparing apples to oranges, but it may be like comparing apples to crabapples.

Going forward, we can look for increasingly accurate snow totals. Researchers at NCAR and other organizations are studying new approaches for measuring snow more accurately (see related story: Snowfall, inch by inch).  

But we can’t apply those techniques to the past. For now, all we can say is that snowfall measurements taken more than about 20 or 30 years ago may be unsuitable for detecting trends – and perhaps snowfall records from the past should not be melting away quite as quickly as it appears

 

I think the fact that major airports have been utilizing the 6 hourly swipe method since the 1950's mitigates this issue to some degree......I know that currently, most spotters are still doing end of storm depth measurememnts. I know for a fact because I had my 31.5" measurement for a March 2018 event ultimately rejected by BOX because it was obtained via the 6 hour method.

I think the larger issue is the lack of consistency...its not as standardized as you are implying.

[40/70 Benchmark]

They way I see it.....snowfall is measured in 6 hour increments, and snow depth is obtained the old-fashioned way.

[LibertyBell]

36 minutes ago, michsnowfreak said:

Unless I am a reincarnated weather observer who worked at the Weather Bureau in the early 20th century, I cant say anything for certain (none of us can) about any individual snowfall of the past. But what I CAN say is that I know the practice of measuring every 6 hours has been in place at Detroit since about 1950, and I assume the same for many other first-order stations. I highly doubt that 6-hour intervals didnt start in NYC til the 1980s. Airports began this in 1950. Also, one thing that sets apart snowfall around here vs the east coast is that we are a climate of smaller, more frequent snowfalls. We do not get big 18-24" storms that can inflate (or deflate) our totals in a 2-day period. Lots of T-1, 1-2, 2-4, 3-6" type snowfalls, then the occasional 7-12" type storms. Also, the old days of the weather bureau had to manually monitor weather hourly, so while most NWS offices have corrected the lazy measurement practices that took many places when the FAA took over in the 1990s, Ive heard countless complaints about observers calling a snowfall a T or something because it melted before obs time. That would NEVER have happened in the old days. 

Another thing I look at when analyzing old weather data for here is the liquid equivalent and snow depth numbers. Is the depth making sense (ie: is there some settling, or the following days snow depth always just the snowfall added onto the previous depth). For the liquid equivalent, are they just applying a 10:1 ratio or are they actually measuring liquid and snowfall separately? In the old days, 1870s-1890s, they usually (but not always) applied a 10:1 ratio to every snowfall. So, does this mean they measured snow and just applied 10:1, or even worse that the melted liquid but didnt measure the snow? No one will ever really know.

In the end, my analysis of snowfall data for Detroit, which dates to 1874 but the NWS begins in 1880 due to some M data in the 1870s...my assessment is that data is very accurate from around 1900 - present. Accurate does not mean there likely werent flaws, but looking at all factors (temps, precip, snowfall, snow depth, following days snow depth, etc) the data makes very good sense. Snowfall measurement practices from 1950-present have remained unchanged. Im not happy with some of the data from the mid-1990s to early 2000s (when the FAA took over and before snowpaid observers were hired), although none of it affected the record books in a huge way. And of course with many different obsevers, wind, etc, snow is always subject to error. Always has been, always will be. Snowfall from 1900-1949 seems quite accurate as well. Its is only the snowfall data from 1874-1899 that I look at with a grain of salt. You can still create a great mental picture of when the big storms were, when the snowy/cold winters were and when the warm, bare, "open winters" were, but the actual measurments likely had much more error than today.

in NYC snowfalls have been undermeasured in recent times too, January 1996 for example, a major outstanding case.

[LibertyBell]

3 minutes ago, 40/70 Benchmark said:

I think the fact that major airports have been utilizing the 6 hourly swipe method since the 1950's mitigates this issue to some degree......I know that currently, most spotters are still doing end of storm depth measurememnts. I know for a fact because I had my 31.5" measurement for a March 2018 event ultimately rejected by BOX because it was obtained via the 6 hour method.

I think the larger issue is the lack of consistency...its not as standardized as you are implying.

Central Park still has undermeasuring issues compared to the airports

[40/70 Benchmark]

2 minutes ago, LibertyBell said:

Central Park still has undermeasuring issues compared to the airports

Right...because the airports are still using the wipe and clear method, while most spotters are not.

I use the wipe and clear method because as I said, I am measuring the amount of snow that falls.....not a random depth.

Its only a big deal in the larger events.

[LibertyBell]

1 hour ago, 40/70 Benchmark said:

Right...because the airports are still using the wipe and clear method, while most spotters are not.

I use the wipe and clear method because as I said, I am measuring the amount of snow that falls.....not a random depth.

Its only a big deal in the larger events.

Yes it's particularly notable in an event like January 1996.

Unfortunately the spotter at Central Park is somewhat lazy and they also mess up in borderline events where it melts during the event and some measurements are missed. For example in one event this winter they recorded 0.6 while everyone else (including the airports) all recorded between 2-2.5 inches.