Spring Pattern Hinting At 2011 East Coast Hurricane Landfall Potential

[Bob Chill]

I understand what the mets are getting at with the analogs and the increasing the timeframe but I think that is over-analyzing what the op was trying to present.

The data and Apr-May analogs presented were only trying to point out the spring pattern is conducive for landfalling hurricanes on the EC. There has been a string of years lately that have had no landfalling hurricanes. The point I think BW was making is that the odds of a US landfall are higher because of the spring 500mb height anomolies.

In a simplistic way, it would be interesting to see the same anomoly pattern for years when there are no US landfalling tropical cyclones.

[Mallow]

I should have been clearer in my argument... while it seems you have found a correlation with late-spring anomalies and EC landfalls, your conclusion was quite flawed:

All you did was prove that just ahead of a landfalling hurricane there was a ridge present in the NE. I looked at the 7 EC landfalls in your OP and looked at the 1 month height anomalies prior to the landfall threat... only 2 of them showed the NE ridge in the long-term pattern. Others were neutral, others had a NE trough. This would indicate that the ridge is more transient and short-term in nature ahead of an EC landfalling 'cane (or at least the ones that you listed). There's (almost?) always a strong NE ridge at some point during the hurricane season... it's just a matter of lining that up properly with a TC.

Ya, this is pretty much exactly what I was trying to say... you just said it better.

It's no surprise that there's a ridge over the NW Atlantic during east-coast landfalls... that's pretty much necessary to steer the storm into the east coast. You could pick any random subset of "analog years" with east-coast landfalls and see the exact same thing--as long as a hurricane is making landfall on the east coast, no matter what the spring pattern was, you're liable to have anomalous ridging off the east coast.

[Mallow]

The data and Apr-May analogs presented were only trying to point out the spring pattern is conducive for landfalling hurricanes on the EC. There has been a string of years lately that have had no landfalling hurricanes. The point I think BW was making is that the odds of a US landfall are higher because of the spring 500mb height anomolies.

But his composite anomaly map for hurricane landfalls doesn't show that. That's the point. All it shows is the composite anomaly map for hurricane landfalls (which may be interesting, but is not predictive), not that we're more or less likely to get such a composite. As Ellinwood said, there's going to be ridging off the east coast at some point during just about every summer--whether the spring pattern was similar or not. It's a matter of getting a storm to time itself properly with that.

[Bob Chill]

But his composite anomaly map for hurricane landfalls doesn't show that. That's the point.

I totally understand the point you and the other mets are making. I'm not defending the op on that. All I'm saying is what I took away from the op is the Apr-May 2011 maps may translate into higher odds of landfalling tc's. That is the part that I found interesting. The analog maps for the same months were not identical but pretty close to Apr-May 2011.

I would be very interested to see Apr-May analog composites for years with no landfalling tc's. I know this is a very simplistic way at looking at the potential connection but it is interesting nonetheless.

[VAwxman]

The ONLY things in common with those three examples are the blocking ridge in the NE and a cool NW Canada.

I believe that is exactly his point.

[VAwxman]

I should have been clearer in my argument... while it seems you have found a correlation with late-spring anomalies and EC landfalls, your conclusion was quite flawed:

All you did was prove that just ahead of a landfalling hurricane there was a ridge present in the NE. I looked at the 7 EC landfalls in your OP and looked at the 1 month height anomalies prior to the landfall threat... only 2 of them showed the NE ridge in the long-term pattern. Others were neutral, others had a NE trough. This would indicate that the ridge is more transient and short-term in nature ahead of an EC landfalling 'cane (or at least the ones that you listed). There's (almost?) always a strong NE ridge at some point during the hurricane season... it's just a matter of lining that up properly with a TC.

Ahh yes, this post defines your argument much more clearly. His initial post seemed fine to me, in that the spring pattern supported more anomalous ridging in the areas needed, but I have not dug into each case like you apparently did.

[bluewave]

I should have been clearer in my argument... while it seems you have found a correlation with late-spring anomalies and EC landfalls, your conclusion was quite flawed:

All you did was prove that just ahead of a landfalling hurricane there was a ridge present in the NE. I looked at the 7 EC landfalls in your OP and looked at the 1 month height anomalies prior to the landfall threat... only 2 of them showed the NE ridge in the long-term pattern. Others were neutral, others had a NE trough. This would indicate that the ridge is more transient and short-term in nature ahead of an EC landfalling 'cane (or at least the ones that you listed). There's (almost?) always a strong NE ridge at some point during the hurricane season... it's just a matter of lining that up properly with a TC.

This year like the analog years lined up properly with the TC so my analogs worked out.

Hazel and Donna are two more examples.

Hazel

Donna

[Mallow]

I totally understand the point you and the other mets are making. I'm not defending the op on that. All I'm saying is what I took away from the op is the Apr-May 2011 maps may translate into higher odds of landfalling tc's. That is the part that I found interesting. The analog maps for the same months were not identical but pretty close to Apr-May 2011.

I would be very interested to see Apr-May analog composites for years with no landfalling tc's. I know this is a very simplistic way at looking at the potential connection but it is interesting nonetheless.

But where did he show that that was the case?

Ahh yes, this post defines your argument much more clearly. His initial post seemed fine to me, in that the spring pattern supported more anomalous ridging in the areas needed, but I have not dug into each case like you apparently did.

But he never showed that. Notice that his hurricane-landfall composite only has individual days (days that the hurricanes were near landfall)... they don't illustrate a general summer pattern, but rather the pattern at the time of hurricane landfall on the eastern seaboard. Anyone could tell you that that would feature a ridge just off the east coast and deep troughing in the SE (the hurricane itself).

[bluewave]

But where did he show that that was the case?

But he never showed that. Notice that his hurricane-landfall composite only has individual days (days that the hurricanes were near landfall)... they don't illustrate a general summer pattern, but rather the pattern at the time of hurricane landfall on the eastern seaboard. Anyone could tell you that that would feature a ridge just off the east coast and deep troughing in the SE (the hurricane itself).

You seem to be working hard to make tangential arguments that don't address the main points made in my original post.

Maybe you didn't feature the east coast for your forecast this season. Irene verifies my original thoughts on an east

coast threat this season.The exact duration of the ridging was not part of my forecast, only that the ridging would be present

at the time of the storm blocking a recurve. You can see that illustrated very clearly with the analogs provided.You said a few days

was not enough so I gave you a whole week.

[Mallow]

You seem to be working hard to make tangential arguments that don't address the main points made in my original post.

Maybe you didn't feature the east coast for your forecast this season. Irene verifies my original thoughts on an east

coast threat this season.The exact duration of the ridging was not part of my forecast, only that the ridging would be present

at the time of the storm blocking a recurve. You can see that illustrated very clearly with the analogs provided.You said a few days

was not enough so I gave you a whole week.

Apparently I'm not doing a good job communicating the issue here.

The problem is that, as far as I see, you have not shown what you claimed to have shown--that is, that the analogs support the notion that the pattern this summer would favor east coast landfalls. My issue is that using only the days near landfall of previous "analog" cases does not constitute proper use of analogs, but rather is a circular argument.

Let's say, for the sake of argument, that I matched the July-August pattern with some set of analog years... say 1996, 1999, 2007, and 2009. Now let's say I found all the heavy rain events in the Pac NW for the following winters and looked at the composite pattern for only time periods around those heavy rain days, and found anomalous southwesterly flow into the northwest with ridging in the intermountain west. Have I shown that this winter is likely to favor flooding rain events in the Pac NW with ridging in the intermountain west? No... all I have shown is that heavy rain events in the Pac NW show up in composites as having anomalous southwesterly flow and ridging in the intermountain west.

Does my analogy make sense? It's really quite difficult to explain the point I'm trying to make (though the concept to me seems so obvious).

[Juliancolton]

Apparently I'm not doing a good job communicating the issue here.

The problem is that, as far as I see, you have not shown what you claimed to have shown--that is, that the analogs support the notion that the pattern this summer would favor east coast landfalls. My issue is that using only the days near landfall of previous "analog" cases does not constitute proper use of analogs, but rather is a circular argument.

Let's say, for the sake of argument, that I matched the July-August pattern with some set of analog years... say 1996, 1999, 2007, and 2009. Now let's say I found all the heavy rain events in the Pac NW for the following winters and looked at the composite pattern for only time periods around those heavy rain days, and found anomalous southwesterly flow into the northwest with ridging in the intermountain west. Have I shown that this winter is likely to favor flooding rain events in the Pac NW with ridging in the intermountain west? No... all I have shown is that heavy rain events in the Pac NW show up in composites as having anomalous southwesterly flow and ridging in the intermountain west.

Does my analogy make sense? It's really quite difficult to explain the point I'm trying to make (though the concept to me seems so obvious).

Basically, your argument can be boiled down to pointing out the post hoc ergo propter hoc fallacy?

[bluewave]

Apparently I'm not doing a good job communicating the issue here.

The problem is that, as far as I see, you have not shown what you claimed to have shown--that is, that the analogs support the notion that the pattern this summer would favor east coast landfalls. My issue is that using only the days near landfall of previous "analog" cases does not constitute proper use of analogs, but rather is a circular argument.

Let's say, for the sake of argument, that I matched the July-August pattern with some set of analog years... say 1996, 1999, 2007, and 2009. Now let's say I found all the heavy rain events in the Pac NW for the following winters and looked at the composite pattern for only time periods around those heavy rain days, and found anomalous southwesterly flow into the northwest with ridging in the intermountain west. Have I shown that this winter is likely to favor flooding rain events in the Pac NW with ridging in the intermountain west? No... all I have shown is that heavy rain events in the Pac NW show up in composites as having anomalous southwesterly flow and ridging in the intermountain west.

Does my analogy make sense? It's really quite difficult to explain the point I'm trying to make (though the concept to me seems so obvious).

Your use of the term proper is a subjective one and in no way refutes my original thesis that the spring pattern was a hint at Irene's recent landfall pattern.

Maybe you just have a very narrow view of what constitutes a legitimate analog and what does not.The point is that the pattern was present at the same

time as Irene like I said in my original post.

[Mallow]

Your use of the term proper is a subjective one and in no way refutes my original thesis that the spring pattern was a hint at Irene's recent landfall pattern.

Maybe you just have a very narrow view of what constitutes a legitimate analog and what does not.The point is that the pattern was present at the same

time as Irene like I said in my original post.

My use of "proper" doesn't refute your hypothesis.... it wasn't intended to. I explained why you didn't support your hypothesis several times already, and it has little to do with the "proper" use of analogs. I don't know how I can make it any more clear.

Of course the pattern present at the time Irene made landfall was similar to the pattern present when other storms made landfall. None of that says anything about the spring pattern or analogs. All it shows is that there's a preferred pattern (on a short timescale) for east coast landfalls. That's what you have shown in your composite. There is, unfortunately, no real predictive power in this, as the "preferred pattern" for east coast landfalls is already well-known and well-understood.

[VAwxman]

But he never showed that. Notice that his hurricane-landfall composite only has individual days (days that the hurricanes were near landfall)... they don't illustrate a general summer pattern, but rather the pattern at the time of hurricane landfall on the eastern seaboard.

Yeah good point. Those Spring analogs should have been rolled forward into the heart of cane season to see if they supported the anomalous ridge.

[Hugo]

So far this April and May we have seen the 500 mb positive height anomalies near the East Coast shift from south to north.

April 2011

May 2011 so far

All the analog years that featured this April to May northward progression of 500 mb height anomalies had landfalling

East Coast hurricanes with the exception of 1981 which saw a tropical storm.

April composite

May composite

This spring pattern seems to signal that a strong ridge will be present to the north later on during the hurricane season

blocking the recurvature of the tropical cyclones to the northeast.I put together a 500 mb anomaly composite that was

present during the landfalls.

Hurricane Isabel 2003

Hurricane Hugo 1989

Tropical Storm Dennis 1981

Hurricane Donna 1960

Hurricane Carol 1954

Hurricane Edna 1954

Hurricane Hazel 1954

Landfall date composite

[Mallow]

Yeah good point. Those Spring analogs should have been rolled forward into the heart of cane season to see if they supported the anomalous ridge.

Exactly. And according to Ellinwood's earlier post, only two of the analog years did support the general pattern of east coast landfalls.

[bluewave]

Exactly. And according to Ellinwood's earlier post, only two of the analog years did support the general pattern of east coast landfalls.

Wrong.I showed that wasn't the case with my composites that i posted.

[bluewave]

My use of "proper" doesn't refute your hypothesis.... it wasn't intended to. I explained why you didn't support your hypothesis several times already, and it has little to do with the "proper" use of analogs. I don't know how I can make it any more clear.

Of course the pattern present at the time Irene made landfall was similar to the pattern present when other storms made landfall. None of that says anything about the spring pattern or analogs. All it shows is that there's a preferred pattern (on a short timescale) for east coast landfalls. That's what you have shown in your composite. There is, unfortunately, no real predictive power in this, as the "preferred pattern" for east coast landfalls is already well-known and well-understood.

It's been shown over and over that patterns in one season can be a precursor to patterns in another season.

I can remember how people were saying that we can't have strong blocking with the strong La Nina last winter

when clearly the signs were there well before that.

[Ellinwood]

It's been shown over and over that patterns in one season can be a precursor to patterns in another season.

Exactly, only you didn't show us a seasonal pattern that was favorable for EC landfalls... you only showed a brief period of time that had the favorable pattern for an EC landfall. That's our argument. Your hypothesis is correct, but your conclusion about having the NE ridge is flawed since the ridge is only temporarily and not part of the long-term, seasonal pattern (in most of your analogs).

[bluewave]

Exactly, only you didn't show us a seasonal pattern that was favorable for EC landfalls... you only showed a brief period of time that had the favorable pattern for an EC landfall. That's our argument. Your hypothesis is correct, but your conclusion about having the NE ridge is flawed since the ridge is only temporarily and not part of the long-term, seasonal pattern (in most of your analogs).

Two week patterns around the hurricane landfalls from my analog years including the second half of this August:

2011

2003

1989

1960

1954

[Mallow]

Two week patterns around the hurricane landfalls from my analog years including the second half of this August:

Because there are few similarities between your analogs in that period (do a composite mean to show what I mean), two weeks is still probably too short a time period to say anything conclusive (though if there is a strong signal, it's getting arguably more useful).

Pretty much every summer is going to have a pattern which "favors" east coast landfalls at some point. The key is to show that it's a general monthly or seasonal pattern, in which case the chances of hurricanes being timed just right to landfall along the east coast is increased for the month/season. This is because hurricanes, by nature, are chaotic, small-scale features. They aren't going to time themselves consistently with short-time-scale mid-latitude troughs and ridges. You're problem is therefore more of a statistical question--are ridges during peak hurricane season (Aug-Sep) more likely to exist in the right places? Transient ridges that just so happened to time themselves perfectly with east coast landfalls in the past don't help you answer that statistical question, because neither the transient ridges nor the hurricanes themselves can be said to be "the same ones" that occur in another year.

[VAwxman]

Here is what I get when rolling the analog years into Aug-Sep, to cover the heart of cane season.

[Mallow]

Here is what I get when rolling the analog years into Aug-Sep, to cover the heart of cane season.

Thanks, Brian!

Ya, doesn't look like there's a real strong signal one way or the other, unfortunately.

However, it does look like there's lower than normal heights over the tropics, which, IIRC, is a good thing for tropical development in general.

[bluewave]

Here is what I get when rolling the analog years into Aug-Sep, to cover the heart of cane season.

You are missing the critical two week periods form my earlier post which lined up with the hurricanes.

You need to fine tune your composites to the time period around the hurricane landfalls like I did

in earlier posts.

[Mallow]

You are missing the critical two week periods form my earlier post which lined up with the hurricanes.

You need to fine tune your composites to the relevant time period.

But they're not even the same two-week periods. You just chose the two weeks closest to the actual hurricane landfalls. How does that tell you anything? "If there was a landfall, this is what the two-week pattern might look like during that time"?

[bluewave]

Because there are few similarities between your analogs in that period (do a composite mean to show what I mean), two weeks is still probably too short a time period to say anything conclusive (though if there is a strong signal, it's getting arguably more useful).

Pretty much every summer is going to have a pattern which "favors" east coast landfalls at some point. The key is to show that it's a general monthly or seasonal pattern, in which case the chances of hurricanes being timed just right to landfall along the east coast is increased for the month/season. This is because hurricanes, by nature, are chaotic, small-scale features. They aren't going to time themselves consistently with short-time-scale mid-latitude troughs and ridges. You're problem is therefore more of a statistical question--are ridges during peak hurricane season (Aug-Sep) more likely to exist in the right places? Transient ridges that just so happened to time themselves perfectly with east coast landfalls in the past don't help you answer that statistical question, because neither the transient ridges nor the hurricanes themselves can be said to be "the same ones" that occur in another year.

You are missing my point .I said the pattern would be present at the same time that the hurricanes were.

You are trying to twist around my original outlook. You can see that blocking intervals over two weeks

were adequate for me to verify. Here is a more specific verification from the 1960 analog:

Statement as of 9:44 am EDT on August 28, 2011

At 8:42 am EDT this morning the tide level at The Battery reached

9.5 feet MLLW. This is the sixth highest level ever recorded at The

Battery. At 9:06 am a maximum surge of 4.5 feet occurred with this

high tide in combination with the astronomical tide.

Other notable levels reached include the top two of 11.2 feet MLLW

and 10.9 feet MLLW which occurred during the hurricane of September

1821 and hurricane Donna in 1960 respectively... and 9.6 feet MLLW

during the noreaster of December 1992.

[bluewave]

But they're not even the same two-week periods. You just chose the two weeks closest to the actual hurricane landfalls. How does that tell you anything? "If there was a landfall, this is what the two-week pattern might look like during that time"?

The hurricanes all occurred around the two week periods.

[Mallow]

You are missing my point .I said the pattern would be present at the same time that the hurricanes were.

You are trying to twist around my original outlook. You can see that blocking intervals over two weeks

were adequate for me to verify. Here is a more specific verification from the 1960 analog:

But you never showed that. You just said it. That's been my main point all along.

[VAwxman]

Thanks, Brian!

Ya, doesn't look like there's a real strong signal one way or the other, unfortunately.

However, it does look like there's lower than normal heights over the tropics, which, IIRC, is a good thing for tropical development in general.

One point in his favor (though he suggests that my way of showing this is wrong) is that, even though the height anoms in the north-central Atlantic on that map I posted are not actually above normal, you do have anomalous easterly flow due to below normal heights to the south couple with "normal" heights to the north. The anomalous wind flow does indeed suggest a better chance for an E.C. hit even though it's not as easy to pick out by the height anoms themselves, since we're tuned to look for above normal heights rather than picturing the anomalous flow. This map shows it better.

Basically, the flow in those years for systems north of the Caribbean was geared more toward the East Coast than a typical year.

[bluewave]

But you never showed that. You just said it. That's been my main point all along.

I am satisfied that I showed it.I guess that we'll just have to agree to disagree on this one.