NOAA Makes It Official

[meteorologist]

http://www.scientificamerican.com/article.cfm?id=noaa-makes-2011-most-extreme-weather-year

[Ellinwood]

The second half of April brought a swarm of tornadoes that leveled parts of the Midwest, including the twister that killed 151 people in Joplin, Mo. So far, 2011 has seen the sixth-highest number of tornado deaths on record, prompting many people to wonder whether climate change has played a role. So far, scientists say there's no good evidence for or against a climate change influence on tornado behavior.

Article FAIL.

[Ed Lizard]

Article FAIL.

May, April, what is a month among friends.

[JoMo]

Did they refer to it as the "Joplin tornadoes"? So sick of people saying that, we had 1 tornado, it was an EF-5.

[BillB]

Of course, measuring "extreme" weather is relatively recent in the climate record. NCDC Storm Events database goes back around 60 years. The longer periods of weather records go back approximately 140 years. Long by human standards, but just a few seconds by terrestrial standards. So using a superlative is questionable, and using one without defining the period of record is, well, less than impressive.

[Sunny and Warm]

Of course, measuring "extreme" weather is relatively recent in the climate record. NCDC Storm Events database goes back around 60 years. The longer periods of weather records go back approximately 140 years. Long by human standards, but just a few seconds by terrestrial standards. So using a superlative is questionable, and using one without defining the period of record is, well, less than impressive.

By my back of the envelop calc, our 140 year record is akin to 1/1000th of a second if we assume the earth is 4.5B years old and a "few" means three. That's a mighty thin record indeed in the scheme of things.

[Mallow]

By my back of the envelop calc, our 140 year record is akin to 1/1000th of a second if we assume the earth is 4.5B years old and a "few" means three. That's a mighty thin record indeed in the scheme of things.

Depends on what you're comparing the 4.5b yrs to. A minute? An hour? A day? A year?

[Sunny and Warm]

Depends on what you're comparing the 4.5b yrs to. A minute? An hour? A day? A year?

All the numbers were in my post, and additional to Billb's comments. I thought it was pretty obvious. 140 years is to 4.5B years as X is to 140 years. 140 years is much less than a blink of an eye.

[k***]

Who cares about 4.5 billion years ago? It's not necessary to go back and take into account when the earth was molten. 140 years is a pretty decent (although imperfect) timeframe to make some climate assumptions.

[Sunny and Warm]

Who cares about 4.5 billion years ago? It's not necessary to go back and take into account when the earth was molten. 140 years is a pretty decent (although imperfect) timeframe to make some climate assumptions.

140 years is barely 1% of this particular interglacial period and you think that's good enough to make climate assumptions? Hardly. My belief is that we should have a thousand year time period to make climate assumptions so that we get at least a couple warm and cold periods into the dataset. The last 140 years has been completely post LIA so one would expect a skewed result.

[k***]

140 years is barely 1% of this particular interglacial period and you think that's good enough to make climate assumptions? Hardly. My belief is that we should have a thousand year time period to make climate assumptions so that we get at least a couple warm and cold periods into the dataset. The last 140 years has been completely post LIA so one would expect a skewed result.

Why should we include data from the little ice age when talking about climatology re: current wx? We aren't in one.

Of course, when talking about a long term trend or debating AGW this might be more of a concern, but my point is simply that having a bigger data set isn't always better.

[Sunny and Warm]

Why should we include data from the little ice age when talking about climatology re: current wx? We aren't in one.

Of course, when talking about a long term trend or debating AGW this might be more of a concern, but my point is simply that having a bigger data set isn't always better.

Your answer is nonsensical. A bigger dataset is always better when discussing things like weather extreme statistics, which is the basis for this thread. Which is better for statistical analysis, one year of weather data or a hundred years? A hundred years or a thousand??

You say that 140 years is enough. Ask a solar scientist if 140 years of data is enough for them. They have 400 years of data, and they'll tell you that it isn't nearly enough to understand the parameters by which the sun operates within. If so, why is 140 years good enough for earth weather?

[k***]

Why include climate periods that are nothing like the present? It's irrelevant for discussing modern extremes.

[Sunny and Warm]

Why include climate periods that are nothing like the present? It's irrelevant for discussing modern extremes.

<facepalm>

[Mallow]

All the numbers were in my post, and additional to Billb's comments. I thought it was pretty obvious. 140 years is to 4.5B years as X is to 140 years. 140 years is much less than a blink of an eye.

Why X to 140 years?

140/4500000000 = X/Y

The choice of Y is completely arbitrary. There's no logical reason you should choose the numerator on the left hand side as the denominator on the right hand side. Indeed, since we're trying to compare it to something we might be able to "grasp", setting Y = 1 year would make a lot more sense to me.

And your math was wrong, anyway.

140 years / 4.5B years = X years / 140 years

Solving for X gives ~4.36x10^-6 years, or ~137 seconds.

If you use the arguably more logical right-hand denominator of 1 year, you get ~1 second (0.98s). In other words, if the whole history of Earth was compressed into 1 year, the past 140 years of history would equate to approximately 1 second.

[Ed Lizard]

In my business, the Plio/Pleistocene are very, very recent...

[BillB]

Look folks, my comment wasn't intended as a lead for an argument about "how many seconds can fit on the head of a pin". I used the phrase "few seconds" in a qualitative sense. How can a scientist accurately judge the significance of a low-frequency "extreme weather" event given the relative short period of data that we have? What is the return period in New England for a storm like the Hurricane of '38? What is the return period in New England for a tornado like the Worcester tornado? Until we know the return period, how can we judge changes in the return period?

As for necessary time span, the weather of the Little Ice Age and Medieval Warm Period and other periods are relevant. If we were to exclude them, we might as well exclude January and July...because the weather patterns are so much different during those parts of the year, right? (Please note that this is offered as an analogy.)

[A-L-E-X]

Who cares about 4.5 billion years ago? It's not necessary to go back and take into account when the earth was molten. 140 years is a pretty decent (although imperfect) timeframe to make some climate assumptions.

Agreed, what many people dont realize when bringing this up is the earth was radically different for a large part of its history and not even inhabitable. Hell, the composition of the atmosphere was even different.

[A-L-E-X]

Your answer is nonsensical. A bigger dataset is always better when discussing things like weather extreme statistics, which is the basis for this thread. Which is better for statistical analysis, one year of weather data or a hundred years? A hundred years or a thousand??

You say that 140 years is enough. Ask a solar scientist if 140 years of data is enough for them. They have 400 years of data, and they'll tell you that it isn't nearly enough to understand the parameters by which the sun operates within. If so, why is 140 years good enough for earth weather?

It depends what you're actually using the data to prove. It's perfectly valid to confine oneself to a given climate..... the earth was basically a different planet 4.5 billion years ago.... something which (we hope) is not ever going to happen again. Yes 140 years is a short period even in our recent climate (what I think Bill was saying) but comparing it to 4.5 billion years is WAY too extreme. You need to understand that for the vast part of its history, earth didn't even have multicellular life on it-- for good reason-- it wouldn't be able to support that kind of life back then.

[ORH_wxman]

Why include climate periods that are nothing like the present? It's irrelevant for discussing modern extremes.

This is a loaded question. I agree with your premise though.

Why should we ever include outliers? DCA has seen their snowfall decline throughout its history until 2009-2010 when it spiked up and helped the 10 year mean spike up to above average levels? So should we count that for something? Or should we ignore it because its quite obviously an "outlier"?

Should we just exclude '02-03 and '09-'10 from DCA snowfall stats because they were "outliers" or should we include them because they count toward a "more extreme environment"? I'm not sure. I think the answer is to include all cases because the climate is always undergoing stress and changing. With trying to risk not becoming a thread for the CC forum, every single year we have huge anomalies, that's just the way weather works. We have a better way of identifying it and measuring it now than we did before so large event are more likely to break records than previously. That is just my little opinion though.

[Sunny and Warm]

Look folks, my comment wasn't intended as a lead for an argument about "how many seconds can fit on the head of a pin". I used the phrase "few seconds" in a qualitative sense. How can a scientist accurately judge the significance of a low-frequency "extreme weather" event given the relative short period of data that we have? What is the return period in New England for a storm like the Hurricane of '38? What is the return period in New England for a tornado like the Worcester tornado? Until we know the return period, how can we judge changes in the return period?

As for necessary time span, the weather of the Little Ice Age and Medieval Warm Period and other periods are relevant. If we were to exclude them, we might as well exclude January and July...because the weather patterns are so much different during those parts of the year, right? (Please note that this is offered as an analogy.)

Bill,

I took your comment as intended. I was just trying to show how infinitesimal 140 really is in the scheme of things. I failed. And yes for those questioning, my math really was on the back of the napkin, so it may be off by a magnitude.

The gist if all this as ORH stated is that you cannot exclude time periods that you no longer believe relevant in judging what is extreme any more than you can exclude January or a large heat wave or the 2005 hurricane season. They were all weather at one time and are now part of climate history.

[A-L-E-X]

This is a loaded question. I agree with your premise though.

Why should we ever include outliers? DCA has seen their snowfall decline throughout its history until 2009-2010 when it spiked up and helped the 10 year mean spike up to above average levels? So should we count that for something? Or should we ignore it because its quite obviously an "outlier"?

Should we just exclude '02-03 and '09-'10 from DCA snowfall stats because they were "outliers" or should we include them because they count toward a "more extreme environment"? I'm not sure. I think the answer is to include all cases because the climate is always undergoing stress and changing. With trying to risk not becoming a thread for the CC forum, every single year we have huge anomalies, that's just the way weather works. We have a better way of identifying it and measuring it now than we did before so large event are more likely to break records than previously. That is just my little opinion though.

Will, I think his point was that bringing in billions of years into the equation isn't a good idea because the earth was like a different planet back then-- the chemical composition of the planet wasn't anywhere close to what it's like right now. And it definitely wasn't anywhere near hospitable to life for the first billion years (and this means any kind of life-- including microbial.) And atmospheric composition has changed radically from era to era. You might as well use Martian climate to compare it to Terran weather.

Yes, 140 years is a small sample size but it should be taken in the framework of the current geological period not the entire history of the planet.

[tacoman25]

People keep on referencing 140 years of records, but as someone pointed out earlier in the thread, detailed severe weather records only go back about 60 years. And even then, the technology and number of people in the country has increased quite a bit since then, so the numbers are going to be skewed a bit higher just because of that.

Not to say this hasn't been a severe/extreme year. Just some things to keep in mind for a proper context.

[tamarack]

Such statements about extremes ought always to have context ("records back thru 1950"), IMO, and perhaps the disclaimer that today's observations are considerably more comprehensive than those of earlier decades.

It also bothers me to read that 2011 is the deadliest tornado year ever, with 500-and-something. Sure, the official records go back only thru 1950, but there's pretty good data on the March 1925 twister that puts the death toll at 695. We don't have an exact count of tornado fatalities in 1925, but don't we know enough to not (yet, anyway, and I hope never) list 2011 as deadliest?

[A-L-E-X]

Such statements about extremes ought always to have context ("records back thru 1950"), IMO, and perhaps the disclaimer that today's observations are considerably more comprehensive than those of earlier decades.

It also bothers me to read that 2011 is the deadliest tornado year ever, with 500-and-something. Sure, the official records go back only thru 1950, but there's pretty good data on the March 1925 twister that puts the death toll at 695. We don't have an exact count of tornado fatalities in 1925, but don't we know enough to not (yet, anyway, and I hope never) list 2011 as deadliest?

Completely agree and it's akin to saying that 2005 was the worst year for the frequency of tropical systems when we don't even know if it was the worst year in the past 100 since we don't have any satellite data for 1933 and surely wouldn't have caught any systems recurving in the middle of the Atlantic or farther to the east. Also you can say the same about the 1995-96 snow season because, as Don S. has pointed out, we've probably had more snow than that here in the late 1800s just before regular records started being kept at NYC. It supports my contention that 100 inches of annual snowfall is indeed possible here given the ideal winter set up but that's another debate for another day lol.

[jpeters3]

This is a loaded question. I agree with your premise though.

Why should we ever include outliers? DCA has seen their snowfall decline throughout its history until 2009-2010 when it spiked up and helped the 10 year mean spike up to above average levels? So should we count that for something? Or should we ignore it because its quite obviously an "outlier"?

Should we just exclude '02-03 and '09-'10 from DCA snowfall stats because they were "outliers" or should we include them because they count toward a "more extreme environment"? I'm not sure. I think the answer is to include all cases because the climate is always undergoing stress and changing. With trying to risk not becoming a thread for the CC forum, every single year we have huge anomalies, that's just the way weather works. We have a better way of identifying it and measuring it now than we did before so large event are more likely to break records than previously. That is just my little opinion though.

For purposes of statistical analysis, more data is almost always a good thing, because it allows for more rigorous testing of statistical robustness of results.

As for subjective exclusion of particular climate periods or "outlier events", there are a plethora of objective ways of accounting for these low-frequency and high-frequency variations without arbitrarily "slicing up" your data. Low frequency and trends (i.e. long term climate variations) and high frequency trends (i.e. interanual variablity) can be disentangled relatively easily, and "noise" due to outlier events can be dampened via low-pass filtering. One would then seek to establish connections between low-frequency trends and, say, severe weather episodes (i.e. warmer climates correlates to more or less severe weather events, or there is no correlation).

My point is that more data, be it from the same type of climate regime or a different one, will almost always be a better thing than less data.

With that being said, changes in the understanding/documentation/observation frequency of these events (and these differences are difficult to quantify) render much of the older data we do have close to useless, IMO, though it is really all we have to work with.

[Sunny and Warm]

For purposes of statistical analysis, more data is almost always a good thing, because it allows for more rigorous testing of statistical robustness of results.

As for subjective exclusion of particular climate periods or "outlier events", there are a plethora of objective ways of accounting for these low-frequency and high-frequency variations without arbitrarily "slicing up" your data. Low frequency and trends (i.e. long term climate variations) and high frequency trends (i.e. interanual variablity) can be disentangled relatively easily, and "noise" due to outlier events can be dampened via low-pass filtering. One would then seek to establish connections between low-frequency trends and, say, severe weather episodes (i.e. warmer climates correlates to more or less severe weather events, or there is no correlation).

My point is that more data, be it from the same type of climate regime or a different one, will almost always be a better thing than less data.

With that being said, changes in the understanding/documentation/observation frequency of these events (and these differences are difficult to quantify) render much of the older data we do have close to useless, IMO, though it is really all we have to work with.

Nice way of explaining the issue!