Cold, cloudy, and “first furnace day” in California

Well, on a cold rainy day in California (that ought to be sunny and warm) thoughts turn to warmer places… like Phoenix Arizona.

It’s cooler than average there today, too, with a bit of overcast / clouds, but the temperature is ‘near 80 F’. Much nicer.

(Wunderground reports the “normal” for San Jose as 74 F instead of the 59 F we’ve got and the record as 98 F (from 1903), so being 40 F below a “very hot day” and 15 F below normal is something you notice… Though at least we’re warmer than the record low of 30 F (from 1893). Those “climate chaos” guys have a ways to go to match that kind of chaos…

But Back To Phoenix

So I took a look at Phoenix Sky Harbor Airport. Back in warmer sunnier times, it was a LOT warmer than the surroundings. I was curious if that was holding up as they went into a cooler cloudier circumstance.


Right now ( 11:20 AM Pacific Time) it’s 81 F on the ASOS and 79 F at “Sky Harbor Airport”.

Yup. Same place. Yup, ASOS reading 2 F high (probably due to the ’round up’ we found as part of their spec earlier; though perhaps just a bit sunny near the runway).

Chandler, a bit south, is cooler at 78.4 F:

Sunrise, Chandler, AZ 78.4 °F 52 °F 40% NE at 3.6 mph
17.72 in 0.00 in / hr 79 °F 1243 ft 7 min 26 sec ago Normal

Mesa (a bit north and then East of the airport) is cooler too at 78.8 F:

Mesa, AZ 78.8 °F 27 °F 15% North at 0.0 mph
27.22 in 0.00 in / hr 78 °F 0 ft 41 min 14 sec ago Rapid Fire

So we’re continuing the pattern of hotter at the airports in the sun and hotter still on the ASOS as it rounds up.

UPDATE 19 Oct 2010

In looking at the generic Wunderground page for “Phoenix Arizona” and looking at the station list at the bottom, this rather nice example showed up. There are a couple of interesting things in it. First off, The stations around Phoenix are about 68.x F while Sky Harbor ASOS is 72 F, so we’ve got a couple of extra degrees of warmth.

Secondly, the ASOS station is NOT in whole degrees C. Don’t know if that is an artifact of the process (what happens on the way to Wunderground?) or if Sky Harbor is now reporting the ASOS without rounding… Might be an interesting development… Or perhaps each individual station can be set to use a rounded or non-rounded feed? Who knows… But in Wunderground, under “page preferences” at the top, you can chose “both” for the temperature display and see if things are in ’rounded F’, or ’rounded C’, or not rounded.

Which brings me to the third interesting point. Notice that the APRSWXNET site is in whole degrees C and with a fractional “.0” on the 68.0 F; while Coronado Historical, with the same 20 C has a “68.7 F”. 68.7 F is 20.3888 C. ( 68.0 F is 20.0 C, an odd quirk). But clearly some odd rounding is going on here.

For Mesowest Durango, that 71 F ought to give 21.6 C (that would round to 22 C) or going the other way the 21 C would give 69.8 F. Don’t see how you can have 71 F and 21 C together unless they are truncating the 21.6 C to 21 C.

Coronado Historical District, Phoenix, AZ 68.7 °F / 20 °C 56 °F / 13 °C 64% North at 0.0 mph / 0.0 km/h / 0.0 m/s
30.08 in / 1018.5 hPa 0.00 in / 0 mm / hr – 1211 ft 0 sec ago Rapid Fire

APRSWXNET Phoenix AZ , Phoenix, AZ 68.0 °F / 20 °C 58 °F / 14 °C 69% NW at 0.0 mph / 0 km/h
28.72 in / 972.5 hPa 0.00 in / 0 mm / hr – 1115 ft 9 sec ago MADIS Website

ASOS_HFM PHOENIX/SKY HARB, AZ, Phoenix, AZ 72 °F / 22.2 °C 55 °F / 13 °C 57% East at 3 mph / 4.8 km/h / 1.3 m/s
29.90 in / 1012.4 hPa 0.00 in / 0 mm / hr 76 °F / 25 °C 1105 ft 33 min 55 sec ago MADIS Website

MesoWest Durango Complex , Phoenix, AZ 71.0 °F / 21 °C 59 °F / 15 °C 65% North at 1.0 mph / 1 km/h
30.00 in / 1015.8 hPa 0.00 in / 0 mm / hr 75 °F / 24 °C 1049 ft 33 sec ago MADIS Website

36th St and Camelback, Phoenix, AZ 68.0 °F / 20 °C 58 °F / 15 °C 71% WNW at 0.0 mph / 0.0 km/h / 0.0 m/s
29.94 in / 1013.8 hPa 0.00 in / 0 mm / hr – 1214 ft 5 sec ago Rapid Fire Website

43rd Ave & Camelback Rd, Sevilla Neighbo, Phoenix, AZ 68.1 °F / 20.1 °C 57 °F / 14 °C 67% SW at 0.0 mph / 0.0 km/h / 0.0 m/s
29.82 in / 1009.7 hPa 0.00 in / 0 mm / hr – 1108 ft 1 sec ago Rapid Fire Website

That 43rd and Camelback does a nice job as 68.1 F gives 20.0555 C that would round to 20.1 C. Though going the other way, 20.1 C gives 68.18 F that ought to round to 68.2 F and we’ve got 68.1 F instead. My guess is that it’s recorded in F and converted to C.

So all of this leads me to believe that we’ve got various equipment set to record F or C, but the C sites are being converted to F at some point for the USA, then in Wunderground, back to C for this display. And all along the way we’re collecting various conversion, rounding, and precision errors.

I’d further speculate that the automated equipment is reporting / recording in C, that it’s converted to F (in the USHCN? ) on it’s way to join the other USA data (that are mostly recorded in F in the first place) and that the conversion back, here, for display shows the degree of accumulated error. I suspect there is a modestly large “dig here” here.

End of The Update

The Graphs:

KPHX: The Phoenix Sky Harbor record:

Phoenix Sky Harbor

Phoenix Sky Harbor

MKPHX: The Phoenix Sky Harbor ASOS record:

MKPHX Phoenix Sky Harbor ASOS Record

MKPHX Phoenix Sky Harbor ASOS Record

San Jose in the Drizzle

The good news is that at SJC San Jose International under overcast and with drizzley skys, the temps match. 59 F at SJC, at the ASOS, at Moffett Air Station and at Palo Alto Airport. There is about a 1 F “jitter” in nearby stations with some each way.

My surmise from this is that the major impact at Airports is solar heating of the tarmac and brown surfaces, when compared to trees and lawns in nearby urban areas. Further, I’d speculate that the best time to observe the impact on historic temperatures would be during months of sunny skys. (I’ve yet to test the impact of snow clearing at airports. I’ll have to find a good set to compare. Perhaps Denver airport and something nearby but snow covered? Or Kansas City in winter?…

This could go a long way to explaining the ‘by month’ variations in ‘trend’ found with the dT/dt method. A nice “dig here” would be to compare the warming months with the cooing months against cloud cover and / or precipitation. (One would also need to control for “Percent ASOS” in the record as the non-ASOS thermometers are dropped and more weight is put on the Round Up ASOS records…)

Is it “All About The ASOS?”

Globally, the “Percent Airports” goes from nearly nothing, to about 80-90%+ today (depending on country). The ASOS are used in the QA Process for daily data for “nearby” non-ASOS stations, so even the non-ASOS may have some ASOS derived bias introduced via the “QA Process”. All in all, the inherent upward bias from a Round Up ASOS on the record, coupled with the increasing percentage of airports in those records that are kept, argues for them being a significant source of error (from a “climate science” point of view. Pilots need the error to be toward the warm side, so for aviation it is correct to ’round up’.)

The onset date of the ASOS system also jives pretty well with the onset date of the “hockey blade” found with dT/dt, and even the magnitude is about right, with the 1/2 C of “AGW” being roughly the same as the known bias from rounding up in whole degrees of C; that being 1/2 C.

I think I need to find out what ASOS data feeds go into GHCN et. al. If the ASOS is used for the METAR and CLIMAT reports, that would pretty much nail it. At least one site:

treats METAR and ASOS as near synonyms with the heading:


It goes on to describe that the temperatures are reported in whole degrees C which implies it is the “rounded up” temperature and not the 1/10 C direct from the instrument.

Further, this link:

from NOAA that appears to be the ASOS users manual is rather interesting too.

In section 2.4 says that the ASOS provides METAR temperatures in whole degrees C. So I think it’s pretty clear that starting about 1990 we’ve got a consistent rollout of ASOS with an ever increasing percentage of the METAR data being “rounded up by whole degrees C”. And at the same time we find 1/2 C of “Global Warming”.


This “ASOS Manual” goes on to describe in great detail how it can calculate all sorts of things, including a monthly average, based on lots of 1/10 C data it stores internally.

But I’m left with the suspicion that it does all these nice things, but it’s the plain old METAR that got glued into the existing GHCN / GISTemp process for creating a ‘monthly mean’…

When pressed about “dropped stations”, the folks in charge of GHCN were all over the place saying how all they did was take in METAR reports and if someone didn’t send them in, well, it wasn’t THEIR problem… (Somewhere I’ve got a reference in one of the postings from that time about how the process works and how the data “flow” roughly twice a month over an old TTY-Era-Design communications system. No mention of just picking up a nice accurate 1/10 C monthly calculated by the ASOS internally… just “Pick up the METAR” and use it.

I’d never believe anyone could be that broken, and have a stupidity error of calculating it ‘right’ inside, then tossing that, and using a rounded up METAR to compute a monthly value that’s warm by 1/2 C (at least)… but these are the folks who’ve twice been unable to keep metric and English units straight and crashed multi-million ($Billion?) dollar satellites as a result.

Wonder if anyone saves those ASOS internally computed Monthly Averages anywhere? It would be very interesting to compare a few to the GHCN ones…

Along the way, I ran into this paper:

That has an interesting discussion of some of the “issues” they ran into with the ASOS as a climate instrument in the ‘early days’. It does confirm the timing of the rollout as matching the ‘hockey blade’ and brings up some other odd issues they ran into along the way.

One that caught my eye was that they said the ASOS was found to read a bit low compared to the prior instrument… which was an HO-83. Which IIRC was found to have a high reading tendency due to sucking it’s own exhaust (the air flow out from the heated bits could be pulled back in via the fan…).

One can only hope they did not “correct” that “low error”…

In Conclusion

Quite a bit to dredge through here. It smells like a smoking gun to me, but the best proof would be to run the numbers end to end. Start with a country somewhere that has ASOS style reporting. Look at actual temperatures and monthly means as computed internally in 1/10 C. Look at what flows to GHCN. Compare. Compute the impact of the increasing percent ASOS over time for that country. Subtract from GHCN values. Compare to magnitude of “AGW”. Light fuse…

Just need to find where the “raw”er data are hidden…

About E.M.Smith

A technical managerial sort interested in things from Stonehenge to computer science. My present "hot buttons' are the mythology of Climate Change and ancient metrology; but things change...
This entry was posted in AGW Science and Background and tagged , , . Bookmark the permalink.

6 Responses to PHX ASOS SOL SOB

  1. GregO says:


    Great work – really and I feel bad for not doing more as I am here in Phoenix. My lame excuse is that I own a company and am busy keeping us in business.

    Let’s talk about that ASOS station at the airport. Arizona State University relies on it for much of their work on urban sustainability and they rely on that ASOS sensor at the airport to track temperature trends. A friend of mine works for the university and is responsible for inputting the values from that sensor.

    I just got off the phone with her telling her about your post and it is not the first time I have warned her about possibly bogus or exaggeration on the side of warming from that particular airport sensor. Incidentally, she has asked me to use her personal email and not her university email as it makes her paranoid to get skeptical of AGW emails at her university account. Weird but true and I think a sign of the times.

    Truth will eventually prevail – I mean we are talking about thermometer readings and how long can those be jacked with until enough people just buy their own …

    Speaking of.. I have been doing my own UHI research. I live in SE Phoenix and mount a sensor on my car and drive to the adjacent Indian reservation just south where there is no urban development. I am working on getting a full year of data; but so far I have found that in non-summer I get a UHI delta T of around 4 deg F and a summer delta T of 7 deg F. Leading me to conclude that in desert-like environs like Phoenix we have an excellent opportunity to examine in detail UHI because of the exaggeration of delta T and the lack of confounding due to factors like humidity and cloud change.

    Food for thought.

  2. E.M.Smith says:

    @GregO: Interesting way to do a UHI study. I have a thermometer in one of the cars, maybe I ought to start looking at it ;-)

    Wunderground is a great way to do an ersatz ‘study’ as you get a whole lot more stations in an area than are in GHCN or USHCN. Pretty easy to ‘bracket’ an area. I can’t do it for Phoenix just because I’m not that familiar with where everything is located.

    But someone who knows the place ought to be able to pick a dozen stations that transect the city E/W and N/S and basically show the heat bubble and how it drifts with wind.

  3. E.M.Smith says:

    I’ve added an update.

    In Wunderground I set the preferences to “both” for temperatures. The F and C do not agree in the small fractional parts. I think this indicates some accumulated rounding and conversion errors. In particular, I suspect that the ASOS is reporting whole C, it’s being converted to “whole F” somewhere (for USA locations) then getting converted back to C in Wunderground.

    Other stations have different artifacts. But all in all it’s pretty clear that the 1/10 degree position of temperatures that have taken a type conversion is not very clean.

  4. mark albright says:

    I just checked Bapchule, south of Phoenix on the reservation. On the morning of 19 Oct they had a low of 61 while Phoenix Sky Harbor had a low of 70. Bapchule is at nearly the same elevation as the airport, but in a very rural environment. You can cut and paste the latitude/longitude below to see the location relative to Phoenix Sky Harbor. BPCHL is a weatherbug or schoolnet site. This map for 6 AM MST 19 Oct 2010 shows the temperatures centered on KPHX:

    “BPCHL” => “33.1364,-111.8725,,AZ,US,St. Peter Indian Mission School/Bapchule”,

  5. E.M.Smith says:

    Well that’s going to be about as close to natural as you can get.

    Sky Harbor is in about the worst place possible for taking a natural temperature. Hottest day of my life was spend driving near / past it. (Radio reported that the tarmac on the aprons was melting so the airport was being temporarily shut down… I think they said it was 125 F at the airport and 140 F + just above the tarmac…) The car A/C just could not keep up and when we parked for some reason, you came back to a solar oven and it took half an hour to reach “just too warm”…

    I was more comfortable in Death Valley in August ( I think it was only 119 F then… and no car exhaust heat nor black tarmac….)

    I remember driving through the sandy desert as comfortable in comparison.

  6. Alexej Buergin says:

    There is nothing odd about 68.0°F being equal to 20.0°C, that is just the definition of the Fahrenheit scale:
    32°F is defined as equal to 0°C
    and a difference of
    18°F is equal to 10°C.

    0°C=32°F, 10=50, 20=68, 30=86 etc
    So not even -40°F=-40°C is odd.

    (Today, all the ancient english units are defined using the “metric” SI as a base)

Comments are closed.