Cumulus Support

How does Cumulus determine wind chill temperature? I've found an error between what Cumulus calculates and what the National Weather Service calculates for wind chill temperatures. I've also found windchill calculators online at http://www.nws.noaa.gov/om/windchill/index.shtml and http://www.csgnetwork.com/windchillcalc.html which agree with Wunderground.


Example taken from my weather station readings:


Temperature(°F) Wind Speed(MPH) Cumulus Wind Chill Wunderground & National Weather Service Wind Chill

20.7°F 13.6 MPH 12.7°F 8°F
23.2°F 14.5 MPH 16.1°F 11°F

Thanks,
Milt

Assuming you're not using an old version of Cumulus, it uses the NWS formula, i.e. the formula under "North American Wind Chill Index" on this page: http://en.wikipedia.org/wiki/Wind_chill

The figures you're quoted for Cumulus aren't those that it actually calculates. If I put 20.7F and 13.6 MPH into the Cumulus code, it gives a wind chill of 7.7F. I think you must be looking at the gust figure, not the (average) wind speed.

I see now that it is using the average wind speed, not the latest wind speed, which is what I was using in the calculation. Over what time span is the average taken? It seems that Wunderground and the National Weather Service are using the latest, not the average wind speed. When I look at the "select-a-graph" I see instantaneous wind chill measurements. Why use the average? It appears to be inconsistent with what is being used by government and professional weather services.

Milt

Your average depends on whether you have Cumulus calculating it or the station. The station uses 48 seconds, by default Cumulus uses 10 minutes. I'd be very surprised if weather agencies are quoting wind chill figures based on gust speeds that vary second by second, what use is that? The NWS use sustained wind speeds measured over 2 minutes for most purposes, IIRC, and also 10 minutes? An online calculator will use whatever figure you give it, of course, as will Wunderground. The default in Cumulus is to send the 'latest' figure for wind speed, which is probably not appropriate for Fine Offset stations, and the option to send the average instead should be used.

Davis VP and VP2 weather stations use a 10-minute average wind speed to calculate wind chill. Here is their justification:

"The reason an average wind speed is employed in the Vantage Pro and Vantage Pro2 to
calculate wind chill is as follows: The human body has a high heat capacity, thus high wind
speeds have no effect on the body's thermal equilibrium. So, an average wind speed provides
a more accurate representation of the body's response than an instantaneous reading. Also,
"official" weather reports (from which wind chill is calculated) provide average wind speed, so
using an average wind speed more closely matches the results that are seen in weather reports."

If you can point me at some definitive document which says that wind chill should be calculated using gusts rather than speeds, I'd be happy to consider it as an option for a future version of Cumulus.

Thanks Steve for sending me on what is turning out to be a wonderful investigation!

I spoke with Mark Ewens of the National Weather Service in Grand Forks, North Dakota. He stated that they used to use a one minute average for all measurements, including wind speed, but with the newer ASOS stations, they now use a two minute average. The following is from a NWS Powerpoint presentation, Slide 23 (see attached document for entire .ppt presentation):

Wind speed and wind directions are recorded every five seconds.
ASOS reports a two-minute average of the five-second average of wind speed and wind direction once a minute.
Peak Wind – is the greatest five-second average wind speed exceeding 25 knots in the past hour.

What was noted in our phone conversation was that the 10 minute average would not allow for the rapid flesh-freezing wind chill temperatures to accurately reflect the very low wind chill temperatures.

The National Weather Service also discusses how the new (as of 2001) wind chill temperature index was developed. I see that there is still much controversy as to how the formula should be changed to incorporate the average of the entire body instead of using just the face.

Thanks again,

Milt