Claim: The PM sea conditions near IOS are

[leong]

After a recent paddling trip out to the Isles of Shoals (IOS) in the morning, we found much “rougher†conditions on the return trip in the afternoon (incidentally, it didn’t happen that way for the majority of my six trips to IOS). Several individuals claimed that what happened was because afternoon conditions near IOS are usually rougher than the morning conditions there. Although, by nature I’m skeptical, especially when conclusions are based on anecdotal evidence, I’m an agnostic on this claim. I guess the claim is plausible, but is there any statistical (or weather theoretic) evidence to support the claim. I don’t know.

Is the claim based on a “sample†with a built-in bias? To illustrate the kind of built-in bias that might lead to an erroneous conclusion, consider the following very simple example:

Suppose the sea conditions are in one of two possible states; calm or rough. Also, suppose the probability of a calm morning (AM) is 1/3 and the probability of a calm afternoon (PM) is also 1/3. Furthermore, assume that the PM sea conditions are independent of the AM sea conditions. Of course, being a risk adverse sea kayaker, you would only paddle out on a calm morning. So each time that you paddle out in the morning you find that 2/3 of the time the paddle back is in rough water. Thus, based on this built-in bias, you might erroneously claim that the PM sea conditions in the IOS region are usually rougher than the AM conditions. But that’s a contradiction in this example since the probability of a rough AM is exactly equal to the probability of a rough PM.

I’m not saying that anyone succumbed to this built-in bias or is lying with statistics. All I’m asking is this: is there valid data to support the claim in the subject line?

[kate]

It's been my experience that this is generally the case - afternoon winds on open coastal waters are greater than in the morning. (This doesn't include a situation where a front blows through at 9 am or some such.) I'm surprised that you have not noticed this too! Something about the anabatic/katabatic winds stirred up by differentials between land mass thermal gain and sea temp. I've always tried to do the bulk of long paddling days in the morning before the wind gets going, or wait until near dusk when it calms again.

Just my 2 cents... offered timidly and under cover after the last IOS thread!

kate

[brambor]

If the sun gets to work it's magic on the water (sunny or partially sunny) all day then I believe the afternoon winds will be stronger than morning's. And with the winds come the waves.

It's been my experience that this is generally the case - afternoon winds on open coastal waters are greater than in the morning. (This doesn't include a situation where a front blows through at 9 am or some such.) I'm surprised that you have not noticed this too! Something about the anabatic/katabatic winds stirred up by differentials between land mass thermal gain and sea temp. I've always tried to do the bulk of long paddling days in the morning before the wind gets going, or wait until near dusk when it calms again.

Just my 2 cents... offered timidly and under cover after the last IOS thread!

kate

[tyson]

Suppose the sea conditions are in one of two possible states; calm or rough. Also, suppose the probability of a calm morning (AM) is 1/3 and the probability of a calm afternoon (PM) is also 1/3. Furthermore, assume that the PM sea conditions are independent of the AM sea conditions. Of course, being a risk adverse sea kayaker, you would only paddle out on a calm morning. So each time that you paddle out in the morning you find that 2/3 of the time the paddle back is in rough water.

I think that your stats are off a bit here. I think you'll find that:

- The paddle out is always "flat".

- 1/3 of the time the paddle back is "rough".

An additional factor is that people will more often remember the bumpy rides than the flat ones. This means that the 1/3 rough return trips may seem more like 2/3.

I have little direct data about the conditions in the AM vs. PM. I have done that trip twice. On the first it was small seas and moderate winds for most of the paddle out. Conditions stepped up a fair bit windier and larger as we neared IOS. Our first return trip was a bit epic.

On the second trip it was eerily calm, flat and glassy on the ride out and and light breezes with tiny ripples on the trip back. Certainly not rough, but still rougher than the paddle out.

There are some micro-meteorology patterns that may bias things toward being windier in the afternoon. During the night it is common to get a "temperature inversion". Normally, the temperature drops as you go up in altitude. This creates some amount of instability and mixing of layers of air. In the case of a temperature inversion, the lower air is cooler. This prevents mixing. The result can be considerable calm on the ground with the winds kept up high. I have experienced taking off in a small aircraft (such as a 2 seat, 1946 Piper J3) where it is dead calm on the ground in the early morning. At about 100-200 feet, there will be a bumpy layer. Above the bumps, you will find yourself flying a bit sideways in a considerable wind. I mostly know about temperature inversions in valleys at night. As the sun warms the ground and the air near it, vertical currents develop which mix the upper and lower layers of air and the wind reaches the ground. I don't know if this may apply to air over the ocean.

An other effect, which seems more likely to be at play is land breezes vs. sea breezes. During the night the land cools more than the sea. The result is air rising over the sea and sinking over the land. At low altitudes you get a breeze that flows from the land to the sea. (and the reverse up high) During the day, the sun heats the land much more than the water and the air currents reverse. As the day progresses into the afternoon, the heating of the land continues and these winds will grow. It might be notable that both during my trip last year and the infamous recent trip, the winds had a "sea breeze" component (from the sea, onto land) which is consistent with this.

It may be that either or both of the above effects (in addition to any bias in selection of conditions or memories of conditions) are at play. ...or I could be flat wrong.

In the recent case, there was also an approaching hurricane which may have had much more to do with the conditions of the afternoon than any common pattern. ...and nothing to do with the original question.

Cheers!

Ty

[leong]

I think that your stats are off a bit here. I think you'll find that:

- The paddle out is always "flat".

- 1/3 of the time the paddle back is "rough".

Cheers!

Ty

If you're referring to my example, they were not stats ... just an example that I made up to illustrate one type of built-in bias.

You said

- The paddle out is always "flat".

I say the paddle out is always "flat" only if you don’t paddle out when it’s not flat. Surely there are rough mornings on the sea near Rye, NH. I’ve been there twice when it was pretty rough in the morning. Keven B. once canceled the trip to IOS when it was very rough in the AM when he arrived.

You said

- 1/3 of the time the paddle back is "rough".

I say, I have no idea if this is true. My question related to what data or scientific evidence implies that a statement such as this is true. You're just stating it. How do you support the statement?

I can’t agree or disagree with anything else you stated. Don’t know, but some of it may be irrelevant with respect to my question.

[leong]

It's been my experience that this is generally the case - afternoon winds on open coastal waters are greater than in the morning. (This doesn't include a situation where a front blows through at 9 am or some such.) I'm surprised that you have not noticed this too! Something about the anabatic/katabatic winds stirred up by differentials between land mass thermal gain and sea temp. I've always tried to do the bulk of long paddling days in the morning before the wind gets going, or wait until near dusk when it calms again.

Just my 2 cents... offered timidly and under cover after the last IOS thread!

kate

Worth a lot more than 2 cents in my opinion. However, I used to kayak fish Ipswich Bay, almost everyday between Lanes Cove and Halibut Point. Not sure, but it seems that it usually calmed down as the sun began going down. Perhaps this is selective memory or bias. The main point of my question was to see if anyone had any real data and scientific theory to explain it. You've given some decent scientific theory, but not enough data to show that it holds (mostly anecdotal evidence).

Respectively, CAMable Leon.

[tyson]

If you're referring to my example, they were not stats ... just an example that I made up to illustrate one type of built-in bias.

You said

- The paddle out is always "flat".

I say the paddle out is always "flat" only if you don’t paddle out when it’s not flat. Surely there are rough mornings on the sea near Rye, NH. I’ve been there twice when it was pretty rough in the morning. Keven B. once canceled the trip to IOS when it was very rough in the AM when he arrived.

You said

- 1/3 of the time the paddle back is "rough".

I say, I have no idea if this is true. My question related to what data or scientific evidence implies that a statement such as this is true. You're just stating it. How do you support the statement?

I can’t agree or disagree with anything else you stated. Don’t know, but some of it may be irrelevant with respect to my question.

My comments have been mis-understood. I was working with your example and almost completely agreeing with the conclusions reached based on it.

I was being explicit that if your self-select for days that are nice in the morning, then all paddle to the IOS are flat. Obviously this is overstated, but the bias is strong and my intent was to be explicit about a point that had been somewhat implicit.

I was also pointing out that where you stated that the return trip would be rough 2/3's of the time, the example/model you presented would actually result in rough conditions for the return trip 1/3 of the time. This does not result in rough conditions most (or more than half the time), but in contrast to the selected flat mornings, it is much more often rough for the return trip.

So, I'm suggesting a minor correction (1/3 vs. 2/3), but mostly agreeing and emphasizing the AM vs. PM contrast.

Ty

[brambor]

Not sure, but it seems that it usually calmed down as the sun began going down.

Of course, makes sense. The sun affects the wind and wind affects the waves. Wind driven waves grow over time. If the sun goes down, the wind slows down and the wave height decreases.

Thus it makes sense that the conditions are calmer in the morning. In the afternoon the wind driven waves build up - so the conditions are worse than in the morning.

Here is an excerpt from this link:

Wave energy, in one sense, is just another form of solar energy. This might sound odd, but just consider that waves start from wind, which forms as a result of the sun's heating of the Earth.

The sun doesn't ever heat the Earth evenly. Depending on the Earth's natural formations as well as its orientation to the sun, some spots get heated more than others. As some air gets heated, it becomes less dense, and thus lighter, and naturally floats upward. This leaves an open space for denser, colder air to rush in and take its place. This air rush is the refreshing cool breeze you feel on a sunny day.

Wind is also responsible for our very powerful waves. As wind rushes up along the water, the friction causes ripples. Wind continues to push against these ripples in a snowball effect that eventually creates a large wave. Essentially, this action is a transfer of energy from the sun to the wind to the waves.

[kate]

The main point of my question was to see if anyone had any real data and scientific theory to explain it. You've given some decent scientific theory, but not enough data to show that it holds (mostly anecdotal evidence).

Respectively, CAMable Leon.

from http://www.solarnavigator.net/wind.htm :

Differential heating is the motive force behind land breezes and sea breezes (or, in the case of larger lakes, lake breezes), also known as on- or off-shore winds. Land absorbs and radiates heat faster than water, but water releases heat over a longer period of time. The result is that, in locations where sea and land meet, heat absorbed over the day will be radiated more quickly by the land at night, cooling the air. Over the sea, heat is still being released into the air at night, which rises. This convective motion draws the cool land air in to replace the rising air, resulting in a land breeze in the late night and early morning. During the day, the roles are reversed. Warm air over the land rises, pulling cool air in from the sea to replace it, giving a sea breeze during the afternoon and evening.

That's about as scientific as I'm gonna get, but maybe John Huth will weigh in....

K

[subaruguru]

Hi folks.

Re IoS last time:

After we collectively witnessed those tourists in their rowboats getting rescued while lunching on IoS a couple of weeks ago I DO remember our Smuttynose'd caretaker volunteering the generalization that "the winds pick up in the afternoon". Although she's a part-time caretaker, lacking other evidence, it would seem reasonable to give her statement strong credence, maybe even more-so than your anecdotal evidence of 4, perhaps even 5 of 6 trip returns being flat. Oh well.

I now find the initial statistical suppositions incomprehensible, but that may be because Ty confused the issue in response. But now I wonder why it matters?

Barring reasonably high confidence in a previously-unknown bias statement like: higher winds in afternoon return seas, a decision to repeat a similar trip when initial conditions are flat and have stayed that way 2/3+ of the time seems like a pretty good bet. This is of course only in the absence of the probablility of rising seas' weather predictions, the timing of which is not always time-correlated strongly enough to not become a nuisance anyway, as we found out.

Hope I'm not bending your thread, Leon.

[alec]

Here is data from the IOS weather station (from noaa). I averaged the hourly wind speed data for all of 2009.

Note the scale. It would appear to support modestly higher wind speeds in the afternoon.ios2009.pdf

[leong]

Here is data from the IOS weather station (from noaa). I averaged the hourly wind speed data for all of 2009.

Note the scale. It would appear to support modestly higher wind speeds in the afternoon.ios2009.pdf

Bingo Alec. Thank you very much. That's the kind of data I was looking for and was too busy (perhaps too lazy also) to find it by myself. So the data shows about a 1-knot average wind speed increase in the afternoon out of 14 (about a 7% increase). Based on this average increase, I don't think I'm going to worry about returning from IOS in the PM (unless the 7% increase affects the wave height in a very nonlinear way; i.e. much more than 7%). We would need a weather expert to discuss this, or preferable get data for average wave height throughout the day.

Given scientific theories versus actual data (especially for something as complex as weather) to quantify past historical conditions, I'll always believe the data before accepting what the scientific theory says what the data should have been. Theory is great when it helps you to understand why the data are what they are. But when they're in conflict then I believe the data.

However, in the case of predicting what the weather will be in the afternoon for a particular day, I'll go with the weather prediction versus the average historical data. So, in the case of planning my next trip, I'm not going to worry that PM winds are (on average) slightly higher than AM winds ... I'll just look at what the meteorologist say for that particular day.

Leon

[leong]

My comments have been mis-understood. I was working with your example and almost completely agreeing with the conclusions reached based on it.

Ok, maybe I partly misunderstood.

I was also pointing out that where you stated that the return trip would be rough 2/3's of the time, the example/model you presented would actually result in rough conditions for the return trip 1/3 of the time.

So, I'm suggesting a minor correction (1/3 vs. 2/3), but mostly agreeing and emphasizing the AM vs. PM contrast.

I think I'm right. In the example it's calm 1/3 of the time in the PM and the PM conditions are independent of the AM conditions. So, if it's calm 1/3 of the time in the PM it must be rough the other 2/3 of the time (since there are only two states in the example the two probabilities must sum to 1.0 (1/3 + 2/3 =1).

Ty

[subaruguru]

Too funny. I had originally assumed that there were THREE states in the original example: 1/3 calm, 1/3 rough, and 1/3 "average" seas. Hence MY confusion.

[tyson]

Too funny. I had originally assumed that there were THREE states in the original example: 1/3 calm, 1/3 rough, and 1/3 "average" seas. Hence MY confusion.

...and my mistake was that I thought Leon had said 1/3 rough when he actually said 1/3 calm. Therefore, he was correct.

Cheers!

Ty

[Phil Allen]

Bingo Alec. Thank you very much. That's the kind of data I was looking for and was too busy (perhaps too lazy also) to find it by myself. So the data shows about a 1-knot average wind speed increase in the afternoon out of 14 (about a 7% increase). Based on this average increase, I don't think I'm going to worry about returning from IOS in the PM (unless the 7% increase affects the wave height in a very nonlinear way; i.e. much more than 7%). We would need a weather expert to discuss this, or preferable get data for average wave height throughout the day.

Given scientific theories versus actual data (especially for something as complex as weather) to quantify past historical conditions, I'll always believe the data before accepting what the scientific theory says what the data should have been. Theory is great when it helps you to understand why the data are what they are. But when they're in conflict then I believe the data.

However, in the case of predicting what the weather will be in the afternoon for a particular day, I'll go with the weather prediction versus the average historical data. So, in the case of planning my next trip, I'm not going to worry that PM winds are (on average) slightly higher than AM winds ... I'll just look at what the meteorologist say for that particular day.

Leon

Leon-

While I won't argue with your decision to use the daily forecast, which is substantially better than an average, don't be misled by Alec's plot. Its an average across many many data points and weather conditions over a year, which is likely to smear out differences. I'm trying to work through the data, but I think a more relevant approach would be how often the winds in the afternoon exceeded the winds in the morning by some value, like 50%. Still slogging through too many numbers and Xcel at the moment to answer the question though.

Phil

[Phil Allen]

(caution, pointy headed geek at work)

Ok, so I felt like suffering through some spreadsheet data crunching. I took the last 45 days of wind speeds from the IoS report, and did a day by day ratio of the 6 wind measurements between 9am-10am, and between 3pm and 4pm. So ratio's greater than 1 means the wind increased on that day, ratio's less than one means it decreased. Note, nothing below is taking in to account initial wind speeds, so the analysis give equal output to 3knots going to 4.5 as 10 going to 15.

Take home is that over 20 percent of the days, windspeeds increase more than 1.5 fold between 9am and 3pm. Now this data is taken from August and Septemeber, where afternoon sea breezes are likely to be common. I should, but probably dont have the skills to, go back and cross-correlate the days with big increases to the initial wind speed, and throw out the data where one wouldn't have paddled, and throw out the data where a 2 fold increase was from 2 to 4 knots. But there you go, a pretty strong indicator that, at least for the last 45 days, winds tend to increase in the afternoons at IoS.

Phil

IoS_winds_ratio_last_45_days.pdf

[leong]

Leon-

While I won't argue with your decision to use the daily forecast, which is substantially better than an average, don't be misled by Alec's plot. Its an average across many many data points and weather conditions over a year, which is likely to smear out differences. I'm trying to work through the data, but I think a more relevant approach would be how often the winds in the afternoon exceeded the winds in the morning by some value, like 50%. Still slogging through too many numbers and Xcel at the moment to answer the question though.

Phil

Super Phil. Exactly what should be done (told many students something like what you said when I taught probability and statistics as a part-timer adjunct when I retired from industry). Not sure, but perhaps the data comparisons between AM and PM should be done with some non-parametric statistical tests to determine the significance of the differences. Boy, I love this bulletin board -- I just pose a problem and everyone smarter than me works on it.

[Lallen]

And this is why I rely on my super geek of a husband's weather forecasting for the days we choose to paddle.

[Phil Allen]

And this is why I rely on my super geek of a husband's weather forecasting for the days we choose to paddle.

Usually I just listen to the clouds and watch the radio..

[leong]

(caution, pointy headed geek at work)

Ok, so I felt like suffering through some spreadsheet data crunching. I took the last 45 days of wind speeds from the IoS report, and did a day by day ratio of the 6 wind measurements between 9am-10am, and between 3pm and 4pm. So ratio's greater than 1 means the wind increased on that day, ratio's less than one means it decreased. Note, nothing below is taking in to account initial wind speeds, so the analysis give equal output to 3knots going to 4.5 as 10 going to 15.

Take home is that over 20 percent of the days, windspeeds increase more than 1.5 fold between 9am and 3pm. Now this data is taken from August and Septemeber, where afternoon sea breezes are likely to be common. I should, but probably dont have the skills to, go back and cross-correlate the days with big increases to the initial wind speed, and throw out the data where one wouldn't have paddled, and throw out the data where a 2 fold increase was from 2 to 4 knots. But there you go, a pretty strong indicator that, at least for the last 45 days, winds tend to increase in the afternoons at IoS.

Phil

Phil, that's a great way to present the data and it shows something else that you didn't mention. If I'm reading the graph right, it shows that about half of the time the wind speed decreased in the afternoon. That's surprising, given that there is a scientific basis for afternoon winds on open coastal waters being greater than in the morning.

If you send me the spreadsheet or links to the raw data (preferably both), perhaps I can write a Mathcad program (or something) to filter out the days when:

1. the wind speed in the morning precluded safe paddling that day.

2. the magnitude of the afternoon winds, although a large multiple of the morning winds, were not large enough to be unsafe for paddling.

3. other conditions that I might think of after seeing the data.

I don’t know, but perhaps the tail of the frequency of PM to AM ratios graph may not be so ominous after the data is filtered as above. Who knows?

[subaruguru]

Phil,

Interesting distribution in that it's more centered around 1.0 than I expected, given the conventional wisdom, eh? Yet it's still skewed sufficiently to the right to bear further investigation of these highest values raw components. As Leon suggests, if many of them indicate high AM winds then for our purposes of whether to paddle or not they may be considered outliers and the distribution becomes even more normalized!

I'd also like to see an overlay of the two AM and PM curves, or even a sequential one where one could look at possible trends lasting more than a 24 hr data window.

One reason is that I wonder if AM and PM longer-term trends are independent of each other.

Do you have the data as 45 days' simple AM and PM duos? That's a big enough sample pool for Leon to play with to study correlation and possibly variance analysis too.

[Phil Allen]

Phil,

Interesting distribution in that it's more centered around 1.0 than I expected, given the conventional wisdom, eh? Yet it's still skewed sufficiently to the right to bear further investigation of these highest values raw components. As Leon suggests, if many of them indicate high AM winds then for our purposes of whether to paddle or not they may be considered outliers and the distribution becomes even more normalized!

I'd also like to see an overlay of the two AM and PM curves, or even a sequential one where one could look at possible trends lasting more than a 24 hr data window.

One reason is that I wonder if AM and PM longer-term trends are independent of each other.

Do you have the data as 45 days' simple AM and PM duos? That's a big enough sample pool for Leon to play with to study correlation and possibly variance analysis too.

The data I have was from the webpage of the IoS buoy, but it's not simple. Give me an email address and I can forward what I've done on. Also, I neglected to include in the plot any ratios greater than 4, but there were 15 or 20 data points (not percentage points) with extreme 3pm/9am ratios heading up towards 20.

Phil

[jdkilroy]

keep in mind that the anemometer on IOS is about 80 ft above sea level. The wind readings are always higher.

Jon

[cgr]

Just pointing out an additional tool. Iwindsurf.com has historical graphing for many NOAA sites.

Isle of Shoals

http://www.iwindsurf.com/windandwhere.iws?...=Wind+Yesterday

They also have some additional readings from their own network(costs $ but you can still view historical data free).

Data for Kalmus beach in Hyannis:

http://www.iwindsurf.com/windandwhere.iws?...=Wind+Yesterday