What the Latest, Gust, Average and PeakGust wind speeds mean

[brassing]

I have been trying to work out what all the different wind measurements presented by Cumulus correspond to in terms of the data output by the Fine Offset weather station. I have discovered some inconsistency in the naming between the Cumulus main page and the graph pages. At least I think I have. If anyone thinks what I report below is wrong please say so! Note that the information below is consistent with FAQ item on wind speeds but adds some information about the terms used on the graphs pages.

The weather station seems to output two values at 48 s intervals. These are the average speed over the 48 s measuring interval and the peak speed over that period, which it calls gust speed on the Fine Offset display.

On its main page Cumulus lists the following wind speeds:

Latest - this is the gust output from the FO
Gust - this is peak gust over the past 10 min of data from the FO
Average - this depends on which of the "Use 10 min wind average" and "use speed for avg calculation" boxes are ticked in the Station Settings:
No boxes ticked: 48 sec average output from the FO
"Use 10 min wind average" box ticked: 10 min average of the gust output from the FO
Both boxes ticked: 10 min average of the average output from the FO

On the graphs page:

Wind speed: average speed as above (i.e. depends on which boxes are ticked).
Wind Gust: Gust output from FO (same as Latest on the Cumulus main page).

On the Select a Graph page:

WindSpeed and Wind Gust: as on the graphs page.
Peak Gust: 10 min peak gust speed (i.e. same as Gust on the cumulus main page)

Thus the gust output from the FO is referred to as Latest on the Cumulus main page and Wind Gust on the graph pages. The maximum gust speed over 10 min is referred to as Gust on the Cumulus main page and Peak Gust on the Select a Graph page.

If you only tick the "use 10 min wind average" box then the average speed is the average of the gust readings. There does not seem to be any scientific case for an average of gusts. It doesn't have any physical significance and is also very instrument dependent. However in a typical urban environment the gusts probably correspond better to the wind aloft so the average gust is probably closer to the clear air 10 m wind speed. It depends on what you want to measure.

[steve]

Wind Gust: Gust output from FO (same as Latest on the Cumulus main page).

If you have 'plot latest gust' selected. If not, it's the peak gust value from the last 10 minutes.

Thus the gust output from the FO is referred to as Latest on the Cumulus main page and Wind Gust on the graph pages.

Not necessarily; see above.

Wind speed naming is a nightmare. Cumulus has to cope with different types of station reporting different measurements using different names. I've done the best I can. I think you'll find that overall the naming is reasonably consistent (given the various options), but you have allow some leeway for the fact that a gust doesn't have a universal definition, and even taking the common usages, the thing that the Fine Offset calls a gust isn't really a gust.

[brassing]

OK - I hadn't spotted that tick box.

I am interested in your comment on what Fine Offset means by a gust. My guess was that the head unit takes several wind measurements over the 48 s reporting interval. These would be made by counting pulses over a particular period. Might be 6 sec which would give 8 readings over the 48 s interval. The average output would presumably be the average of the individual readings and the gust output would be the maximum of those readings. Is that what you understand?

I am not sure if there is any official definition of the gust speed. It is all tied up with the frequency spectrum of the wind speed. If the instrument time constant is less than the shortest period of the wind speed variation then the instrument would record an instrument independent gust value. For longer instrument time constants the reported gust value would be instrument deoendent.

My understanding is that the wind speed actually has some quite high frequency components which could only be recorded with a sonic anemometer or similar, so any definition of gust speed would need to specify the instrument time constant to be meaningful.

[steve]

I am interested in your comment on what Fine Offset means by a gust. My guess was that the head unit takes several wind measurements over the 48 s reporting interval. These would be made by counting pulses over a particular period. Might be 6 sec which would give 8 readings over the 48 s interval. The average output would presumably be the average of the individual readings and the gust output would be the maximum of those readings. Is that what you understand?

That would be my guess too. I had thought that perhaps it was taking six 8-second values or eight 6-second values, but of course 48 seconds has so many factors that there are lots of possibilities. Unfortunately Fine Offset haven't seen it necessary to document what either of the wind speed values represent.

I am not sure if there is any official definition of the gust speed. It is all tied up with the frequency spectrum of the wind speed. If the instrument time constant is less than the shortest period of the wind speed variation then the instrument would record an instrument independent gust value. For longer instrument time constants the reported gust value would be instrument deoendent.

The typical definitions of a gust that I've seen have been along the lines of "an increase in the wind speed above the current average by at least X, lasting for at least Y seconds".

It would be much better if FO had done what Davis do, and simply supply a single wind speed calculated by counting the number of revolutions over a given short period (in the case of Davis this is about 3 seconds) and then left the software to do sensible things with it. But that would require the sensors to transmit much more frequently and the batteries wouldn't last very long.

[Gina]

There are a number of things about Fine Offset stations that could be improved. They have improved some things between the first one I bought a couple of years ago and the second I bought a few months ago. The rain gauge, for instance, is more accurate and a matrix of drain hole rather than one large one each side helps stop spiders etc. getting in and should help with the problem of wind blowing in the bottom and tipping the buckets. I've found the temperature data to be more accurate too. OTOH there was no change in the wind sensors whatsoever.

[brassing]

On the subject of how many wind readings the Fine Offset makes during the 48 s reporting interval - it is much easier for a simple microprocessor to divide by 8 (or 4) than by 6 (its just a simple register shift). So my guess is still with 8 readings. I have thought of a way of finding out. First bring the sensor inside so there is no wind, then pulse the wind sensor manually by flicking the cups and stopping them after a second or two. Wait till the wind reading updates on the display then do the same thing again. Make a note of the gust and average values. Do this several times then work out the ratio of gust to average for each one. Most of the time this ratio will be the number of readings per 48s. Occassionally your pulse will be split across two readings which would make the ratio lower so take the highest ratio which should also be the most frequent figure.

I can't try this myself yet because my sensor is up a mast on our chimney. However I have just bought another unit for a friend so I'll try it on that sometime. I'll have to do it out of radio range of my existing one though!

[brassing]

I've now done the experiment with the following results:

Gust Avg Ratio
29.50 1.10 26.82
24.50 1.10 22.27
28.10 1.10 25.55
33.10 2.50 13.24
27.00 1.10 24.55
28.10 1.10 25.55
35.60 3.60 9.89
30.60 2.50 12.24
33.10 1.10 30.09
35.60 1.10 32.36
29.50 1.10 26.82
34.20 1.10 31.09
25.60 1.10 23.27
29.50 1.10 26.82

Not quite what I was expecting. It looks like it takes at least 16 measurements for every 48 s interval. The ratio gets as high as 31 which might suggest 32 measurements per period but I doubt that. Its probably down to rounding errors. You will notice that the only values for average speed are 1.1, 2.5 and 3.6. It could be that it reports all values between 1.1 and 2.4 as 1.1 which would explain the high ratio values.

If it is 16 that would indicate a 3 s measurement period for gusts. A very approximate calculation suggests that 1 rev/s of the anemometer = 1.1 m/s wind speed (anemometer cups normally move at between 2 and 3 times slower than wind speed so you can work it out based on the anemometer diameter of 140 mm). The internal photos elsewhere on this forum show that the anemometer has only one magnet so it gives one pulse per rev. This means that 1 pulse in 3 s corresponds to .37 m/s which is close to the known 0.3 m/s (=1.1 km/h) resolution which we notice in the reported speeds. Probably not a coincidence.

[steve]

Signatrol seem to think it does 24 2-second samples.

If your figures are in kph, it might be better to do it in m/s, which is the station's native unit.

[Gina]

it might be better to do it in m/s, which is the station's native unit.

Yes, I wondered that.

[brassing]

Is signatrol another forum member - couldn't find any posts by him when I searched on author? Do you know how he arrived at the 24 / 2s ? It is certainly consistent with my data but it seems a bit strange to choose 24 because it makes division to get the average more difficult. If the division is done by a register shift that means all values are rounded down so would be consistent with my suggestion that all values between 1.1 and 2.4 are reported as 1.1.

My unit came set to kph so I left it like that as I assumed that was its preferred unit. Unit conversion would be done in the display unit which would have a more powerful processor so should be able cope reasonably accurately with the conversions. Or maybe not!

At least we seem to be in the right ballpark with a 2-3 s gust measurement period.

[steve]

Is signatrol another forum member

A Fine Offset retailer, but they're also on the forum in some guise.

http://www.signatrol.com/product/35_wea ... tions.html

My unit came set to kph so I left it like that as I assumed that was its preferred unit. Unit conversion would be done in the display unit which would have a more powerful processor so should be able cope reasonably accurately with the conversions. Or maybe not!

I know it's an easy conversion from m/s to kph, but it still introduces another level of rounding.

[brassing]

I checked with Signatrol and they say they got the 2 s gust measurement period direct from Fine Offset so I guess that settles it. Not sure why they chose 2 s. In the USA at least, the standard gust measurement period seems to be 3 s.

The 2 s period leads to a bit of a puzzle about the anemometer factor (the ratio of wind speed to cup speed) but I'll start another topic on that.

[PaulMy]

Didn't this come up in another topic?

For users who require more advanced software, the excellent Cumulus software offered by Sandysoft is compatible with the WH1090. Cumulus software is available for a small donation and can be downloaded from here: http://sandaysoft.com/products/cumulus

[steve]

I had an email yesterday from someone who says he asked Fine Offset about its wind measurements and how they work, and got the following answer. As we're already led to believe, it makes 24 2-second samples over the 48 second measuring period, and the highest of these it reports as 'gust'. It then averages the 24 samples. And here's the strange part: it doesn't report that value as the 'speed'. Instead it adds that value to the 'gust' value and divides by two, and reports that as the 'speed'.

I find this hard to comprehend, or even believe. If correct, it effectively means that the highest of the 24 samples counts 25x more than the others, and the 'speed' can never be less than about half of the 'gust'. Odd.

[Gina]

I had an email yesterday from someone who says he asked Fine Offset about its wind measurements and how they work, and got the following answer. As we're already led to believe, it makes 24 2-second samples over the 48 second measuring period, and the highest of these it reports as 'gust'. It then averages the 24 samples. And here's the strange part: it doesn't report that value as the 'speed'. Instead it adds that value to the 'gust' value and divides by two, and reports that as the 'speed'.

I find this hard to comprehend, or even believe. If correct, it effectively means that the highest of the 24 samples counts 25x more than the others, and the 'speed' can never be less than about half of the 'gust'. Odd.

I agree, unbelievable!! I had to read that two or three times to be sure I was reading it right. I'll see if I can set up a test rig to confirm or deny this. I don't think it can be right though as my tests have shown that a steady wind speed always produces at least one unit difference between gust and speed. ie. unless in motionless air the gust is always one or sometimes two units above the speed. A unit of speed is 0.34m/s. I guess the difference could be caused by rounding errors if the processor always rounds down, but this difference always seems to be there whatever the air speed - I tried varying it by small amounts.