Eric Berger writes (excerpt below) at the Houston Chronicle “SciGuy” blog about the wicked hot record breaking temperatures people were getting in Houston on their iPhones. Turns out the whole issue is over station siting. While he located the station, doing some further investigation, I’ve got shots of the actual equipment and error analysis.
Eric Berger – Houston Chronicle
Over the last several days I have seen a lot of reports on Twitter and elsewhere about very warm temperatures in Houston, a few degrees warmer than those being reported by the National Weather Service. They have come from people with iPhones.
For example the “Houston” location on your iPhone’s generic weather app reported a temperature of 109 degrees yesterday at 2:50 p.m. when in fact the official reading at Bush Intercontinental Airport at the time was 105.
Indeed no major weather station in the Houston area reported such a high temperature at the time. It got TechBlog‘s Dwight Silverman and I wondering what was going on. Where was Apple pulling this data from?
We started speculating on Twitter and didn’t begin to solve the mystery until Travis Tubbs chimed in.
After that Dwight archived the conversation in Storify, and it appears
below (above), starting with Dwight’s initial tweet about the temperature getting back to a record 109 degrees Monday. (Officially, it didn’t. Monday’s high topped out at 107.)
So the mystery was solved, sort of. Apple gets its weather data from Yahoo!, which in turn gets it from The Weather Channel.
And for whatever reason the default iPhone weather data point is the Houston/Hermann Medical Center weather location. Whatever that is. It’s not an official National Weather Service site.
Full story here
No, it isn’t an official NOAA station, it serves the heliport. But it is a MADIS station, registered with NOAA, and used for other purposes such as MESONET.
Te lat/lon of the station is listed at the Gladstone CWOP metadata site as 29.7140 -95.3950 and that bring up this image on Google Earth:
Clearly the lat/lon places it on the roof in the helipad, but even at max zoom, I can’t make out any weather station. I was however able to see a helicopter on the pad and a windsock on the Bing Maps image and I theorized the weather station might share the mast with the windsock.
Fortunately, certain people like to photograph helicopters in action, and I was able to locate this shot from a photographer’s FLICKR collection that showed the station:
If you enlarge the photo, you’ll see the anemometer and wind vane, and what looks like a gill shield for the thermometer, though I can’t be certain. Even if it isn’t, the thermometer will be nearby to serve the helipad.
What I found most interesting though was the temperature analysis provided by the NOAA MADIS website. Apparently the problems with this station are well known.
|24 hours||Day time||Night time|
|Average temperature error||-2.7 °F||-4.9 °F||-0.5 °F|
|Error standard deviation||3.3 °F||3.4 °F||0.9 °F|
|Worst average temperature error||-3.8 °F||-6.5 °F||-1.0 °F|
|Worst standard deviation||4.3 °F||6.2 °F||1.5 °F|
NOTE: If the error above is POSITIVE, then it means that the analysis temperature is HIGHER than the reported temperature. This means that your sensor is reading COLDER than expected
Worst daytime temperature error of 6.5F …wow.
A look at their graph is also telling:
As is the note below it (red color theirs):
Your readings indicate a solar heating problem. This means that the sun can shine directly onto the temperature sensor (or it’s housing) and increase the temperature recorded significantly. This is often caused by the lack of a radiation shield, or the sensor is poorly sited. For information on radiation shields, see CWOP Radiation Shields, and information on siting, see CWOP Station Guide.
With error magnitudes like that, perhaps the Weather Channel Might want to use a different source of station data for reporting Houston temperatures.
Again, siting is the issue. The station might be fine for reporting helipad conditions, and it might very well be that hot of the roof of that building. Knowing that accurately is important for the safety of the helicrew, especially on days of high humidity and high density altitude that sap rotor lift efficiency.
But it also underscore a point I’ve made time and again: rooftops aren’t a representaive location for climate monitoring, as we’ve seen a couple of days ago with this rooftop GHCN station # 2076548 photographed in Tampico, Mexico:
As we’ve seen time and again, GHCN stations are often sited at airports, and while runway and helipad conditions are imperative to know, especially when they differ from surrounding ambient, such stations sited in odd places don’t translate well for other applications, be it long term climate monitoring or a simple iPhone app for infotainment purposes.
If people reading iPhones can spot such problems and know the temperature data isn’t valid, and a citizen run CWOP program can use MADIS analysis of the data sight unseen and note it’s FUBAR, why can’t our meteorological experts at The Weather Channel figure it out?