Easterlies (2 Viewers)

[johnraven]

last years late September Sibes were the result of a similar "dual high" btw. Basically it all depends on what the high over us does..and the forecasts on this change by the hour....but there there is indeed every chance it will drift more north and east then we could be in. Even in clear conditions...choice birds will land..ok so maybe not major rain related falls..but enough birds could be noted on the east coast, indeed the clear weather also means the likes of Portland and Lundy etc could get their share.

I'm with you. The forecast for the South East and East Anglia is essentially three days (Thurs-Sat) of moderate (c.15mph) NE winds, which is fantastic.

http://www.metoffice.gov.uk/weather/uk/ee/lowestoft_forecast_weather.html

I'm a bit concerned, however, that the wind in the north east of Europe is not as north easterly... (e.g. Copenhagen)
http://www.metoffice.gov.uk/weather/europe/copenhagenforecast.html

Still, as you say, past September easterlies due to a high pressure sat to the north have produced significant increases in scarce+ birds.

 

[tittletattler]

I'm with you. The forecast for the South East and East Anglia is essentially three days (Thurs-Sat) of moderate (c.15mph) NE winds, which is fantastic.

You seem to be changing your mind as much as I have!

Worth getting out into the field, especially if you live on the east coast.

 

[ColinD]

You seem to be changing your mind as much as I have!

Worth getting out into the field, especially if you live on the east coast.

or you are staying in Bridlington for the weekend :t:

 

[username]

Perhaps some birders on the east coast might like to indulge in a bit of 'dove from above' encouragement to entice a few rares out the sky...[vic n bob 'shooting stars' stylee]........of course....it might not 'produce' owt 'good'...plus...you run the risk of getting hauled away by the blokes in white coats...worth a 'punt' tho?!

 

[johnraven]

You seem to be changing your mind as much as I have!

Indeed. I fear I may have changed my mind again. All the past influxes of scarces involved easterly winds in NL/Denmark etc as well - but that is not the forecast now. Perhaps the wind direction here is not as important as the wind direction there - where our erstwhile rarities set off...

Worth getting out into the field, especially if you live on the east coast.

Ooh, I would love to. Not here, in this blasted office, but out there, rares or not.

John

 

[johnraven]

Using wunderground.com data on historic daily weather conditions, I've had a go at quantifying some of this.

Basically I've regressed the daily number of 'first records' of 'scarce and rarer' species in East Anglia, for each September 2004-2008 (Birdguides data) on weather conditions. I've excluded Honey Buzzard (too many) and seabirds (need different winds).

The results are pleasing from a statistical point of view, and mostly match expectations. Interesting results include:

(1) The wind direction in Copenhagen is a more powerful predictor of scarce birds in East Anglia than the wind direction in East Anglia itself.

(2) The best wind conditions are Easterly in Copenhagen and North Easterly in East Anglia. Conditions build up over three days - so three days of NE in a row is better than two days.

(3) Low pressure and rain are both good.

(4) Weekends see a big increase in scarce+ finds, particularly if the winds are from the east (presumably that attracts more birders).

Nothing that surprising - except perhaps the Copenhagen thing.

I've used this statistical model to predict scarce finds over the next few days:
Monday = 4 (Actual)
Tuesday = 5 (Actual)
Wednesday = 3 (Prediction)
Thursday = 4 (Prediction)
Friday = 6 (Prediction)
Saturday = 12 (Prediction)

So the model says no real increase until Saturday, which will be moderately good (a real monster day would see 20+ first records).

John

 

[ColinD]

........

Nothing that surprising - except perhaps the Copenhagen thing.

I've used this statistical model to predict scarce finds over the next few days:
Monday = 4 (Actual)
Tuesday = 5 (Actual)
Wednesday = 3 (Prediction)
Thursday = 4 (Prediction)
Friday = 6 (Prediction)
Saturday = 12 (Prediction)

So the model says no real increase until Saturday, which will be moderately good (a real monster day would see 20+ first records).

John

Great information John. It will be interesting to see how accurate you are! The only improvement I would like to see to your data is I wish that you could have predicted Sunday as well! You would think it would be at least as good as Saturday, since the north easterlies will have been with us a day longer.....

 

[johnraven]

I've extended it. The model predicts 13 for Sunday, so slightly better.

This just for East Anglia, btw. There have been two Scarce+ finds so far today.
The model explains 60% of the variation in the number of finds, which isnt bad, although clearly there is a lot of unexplained variation. One should expect that, of course.

This chart shows how well the model performs on past data, showing the predicted number of finds versus the actual number of finds, for the 5 years of data I've used. This is an "in sample" prediction.

 

[dbradnum]

This is brilliant... deserves a thread of its own. If you have the time to keep running the model, John, I'd be really interested to see how this performs over the course of the autumn.

What variables are you using? Just wondering if there are other readily quantifiable ones out there that might strengthen the model further....

 

[ColinD]

I've extended it. The model predicts 13 for Sunday, so slightly better.

This just for East Anglia, btw. There have been two Scarce+ finds so far today.
The model explains 60% of the variation in the number of finds, which isnt bad, although clearly there is a lot of unexplained variation. One should expect that, of course.

This chart shows how well the model performs on past data, showing the predicted number of finds versus the actual number of finds, for the 5 years of data I've used. This is an "in sample" prediction.

Thanks John. That's great info. Almost better than birding!

 

[dbradnum]

Oh, and one other thing - which date is shown by the big spike at observation c115?

 

[johnraven]

Oh, and one other thing - which date is shown by the big spike at observation c115?

The giant peak was Saturday 29th September 2007. This was a period of brisk N to NE winds, with NE to E winds in Copenhagen, with falling pressure and rain.

The variables of the model are as follows:

(1) Day of month - the number of records tends to increase over the month. I tried a non-linear function (polynomial), but only the simple upward trend passed the 95% confidence test. The number of records increases by 0.127 each day.

(2) Wind direction today - only N, NE and E wind directions matter. I thought SE might matter, but it doesn't. N are NE are best, both boosting records by 3.

(3) Wind direction yesterday - again N, NE and E. NE is the best, boosting records by 4.3. East is next best, by 2.

(4) Wind direction day before yesterday - only NE proved to be statistically significant.

(Overall NE wind is therefore best)

(5) Weekend - Sat and Sun boost records by 1.6, unless there is an easterly wind (NE or E or SE) then records are boosted by 3.5. I suspect easterlies attract more birders.

(6) Pressure - the pressure in isobars (e.g. 1020).

(7) Rain - actually, this drops out of the equation when I include Pressure, but is significant if I do not. The two are correlated. Rain is good, btw.

(8) Copenhagen wind yesterday - East wind is best, then NE wind. Surprisingly, SE is not very good, although I note a lack of SE wind in my data.

I also tried wind speed, but the relationship is too complex, for my 150 observations (5 years of 30 days) of data. A strong SW is bad, but a weak NE is not as good as a moderate NE. I tried a few possibilities, but nothing has worked. I need to think about this one.

The functional form of my equation is not right. I use simple OLS, which means the prediction can be negative (see the chart). Obviously the actual number of finds is limited to zero and discrete. But I just cant be arsed to deal with the statistical complexity of a non-linear discrete function - I would rather test the OLS model.

I'm very keen to try any suggestions you might have though. I will need daily data back to 2004.

John

 

[johnraven]

Oh, one other thing I need to test, is the year. Perhaps the number of records has changed over the years? I'm not sure the number of birders in East Anglia has changed, but my method is based on Birdguides data, and their reporting may have changed.

 

[dbradnum]

The giant peak was Saturday 29th September 2007. This was a period of brisk N to NE winds, with NE to E winds in Copenhagen, with falling pressure and rain.

Ah yes, a good weekend... Paddyfield, Greenish, Blyth's Reed and Subalpine Warblers (though two of these were in Kent), plus a good seawatch the day before.

I'll have a think about other potential variables and chat to a few friends. Have you tried wind direction from other locations significantly north and south of Copenhagen? Also, how does the model stand up for October? I guess day of month may be less significant, but I assume most factors remain similarly weighted?

 

[johnraven]

I'll have a think about other potential variables and chat to a few friends. Have you tried wind direction from other locations significantly north and south of Copenhagen? Also, how does the model stand up for October? I guess day of month may be less significant, but I assume most factors remain similarly weighted?

Good ideas. Both require some data capture (particularly expanding to October), so it depends how busy I am, but I'm keen to do so. Any suggestions on other locations? I think Amsterdam is too close.
Oslo? (for the north Scandinavian vibe)
Leningrad/Moscow? (to capture deeper Asian air flows?)

It depends on what wunderground have to offer in terms of historic data.

Any advice on wind speed most welcome. I know it must matter, but its complexity is clear and any 'intuition' on how it works would be useful.

I'm pleased with the initial results - 60% explanatory power and all of the variables passing the 95% confidence interval. The dataset is a bit unclean, as it contains dodgy sightings and it would be too much work to clean it up, but I'm not sure that matters. Dodgy sightings should be "white noise" in statistical terms, which just increase the 'error' of my forecast model rather than systematically bias the results.

But the real test is 'out-of-sample' forecasting. Will the prediction above bear any resemblence to what actually happens over the next few days? I notice that the number of first records has been quite high recently - 5 today already!

 

[Rob Smallwood]

Fascinating stuff....

 

[dbradnum]

Out of interest, what are the five new rarities you're referring to today?

What proportion of the records in general are flypast seabirds? I wonder if running two separate regressions for seabirds and passerines (or just non-seabirds) would give interesting results - might expect the arrival / detection pattern to be different?

I agree that Amsterdam is a bit close as well - probably too similar to wind direction over the East Anglian coast. I'd choose Bergen over Oslo for a more northerly dimension with respect to East Anglia (would expect Copenhagen and Oslo to be pretty similar). For the 'further east' dimension, maybe try one of St Petersburg / Tallinn / Helsinki - not too far away, but still the potential source of plenty of good birds! Hard to know whether a central Europe wind vector would be significant, but perhaps choices like Frankfurt or Munich could be options to try.

As regards wind speed in East Anglia, I wonder whether there is a time-lagged relationship. For instance, if it's a howling north-easterly, then many good birds (typically small passerines) are harder to find (and identify) even though they may be arriving in droves. However, if you had those conditions on a Thursday or Friday, and then lessening wind speed over the next two days, the prospects would be very good for the weekend.

I've been meaning to buy >this Poyser book< on birds and weather for some time - no doubt it would shed some light on the topic and suggest some other variables. Might have to get the credit card out...

 

[johnraven]

Out of interest, what are the five new rarities you're referring to today?

My analysis is based on doing a search of the Birdguides database, with the following criteria:
- East Anglia
- Scarce or rarer
- relevant dates
- First records only

I do no cleaning of the data whatsoever, except for identifying a few species I have chosen to exclude (Honey Buzzard and a list of seabird). I simply count the number of records.

If you do that search for today you get the following results:
- Common Crane: Leiston
- Isabelline Wheatear: Landguard
- Grey Phalarope: Holme (EXCLUDED - SEABIRD)
- Common Crane: Martham
- Montagu's Harrier: Willingham
- Eleonora's Falcon: Ness Point

So that gives 5 records once we exclude the seabirds.

You will note a few issues here. Firstly, one of the records was later corrected (Isabelline Wheatear) and one is only a possible (Eleonora's Falcon). Secondly, Common Crane could refer to resident birds. But I simply cant/wont clean so much data by hand, and would need to create an objective rule list in order to do so.

Does this matter? Perhaps not, if these sightings can be considered to be "white noise". It's a matter of debate.

(Today was a particularly bad day for less meaningful records for this purpose, btw).

John

 

[johnraven]

This is the same list for 29 Sep 2007:

Common Rosefinch
Richard's Pipit
Ross's Goose
Yellow-browed Warbler
Richard's Pipit
Yellow-browed Warbler
Great Grey Shrike
Red-necked Phalarope
Wryneck
Great Grey Shrike
Wryneck
Yellow-browed Warbler
Temminck's Stint
Great Grey Shrike
Great Grey Shrike
Great Grey Shrike
Red-breasted Flycatcher
Yellow-browed Warbler
Great Grey Shrike
Great Grey Shrike
Barred Warbler
Yellow-browed Warbler
Great Grey Shrike
Red-flanked Bluetail
Red-backed Shrike
Serin
Great Grey Shrike

You can see that in a non-good day (like today), the crap seems to dominate, but in good days (like 29/9/7), the scarce+ method does produce a good indicator of interesting birds.

 

[StuartReeves]

Fascinating analysis John, could make an interesting basis for a BF 'rareometer' now that the Punks aren't running theirs anymore. I'm heading east myself tomorrow (for a weekend at Falsterbo) and I'm wondering if my absence may be a more qualitative predictor of good birding in East Anglia.