It's 90°F in Phoenix. In March.

March 2, 2009 12:42 AM

WeatherFilter: When "record highs" (or lows) are from the early 20th century, does anyone stop to consider that device accuracy and methodologies might have contributed to those records?

I mean, I know the basic mechanics of a thermometer have been relatively stable for a long time, but with my lovely town of Phoenix hitting near the record high of 90°F, I'm just a bit curious if 1921 was a bit of an odd year, or if it's more likely there was an aberration in the reading, and how that's taken into consideration?

Were the temperature recorders of yore sworn to a solemn, well-understood duty to ensure that those data points they were collecting would be cast in stone, for future generations to see, examine, and Ask Metafilter about? Or was it a bit more slipshod or haphazard, or might someone have been prone to exaggeration? Where would I find more on the practice of temperature logging from back in the day, to determine just how seriously "they" took it. (They, in quotes, simply because the groups responsible must vary wildly throughout the country.)

posted by disillusioned to science & nature (13 comments total)

Yup. I'm heading out so haven't got time to give references at the moment, but it's fairly well-trodden ground. Amongst other things, temperatures from both ground-based stations and satellites have been (and are still being, in some cases) corrected for both systemic errors and other effects such as 'heat islands' (the tendency for towns / cities / etc to be hotter than the surrounding countryside).

There's also the (possibly apocryphal) story about the rainfall records for Tully, Australia being ... ahem, 'tainted' ... due to the fact that the official rain gauge was located at the train station. The story goes that train drivers found it a handy place to relieve themselves, leading to some rather exaggerated readings ;-)
posted by Pinback at 12:59 AM on March 2

Interesting...

I also *have* to believe that the lows in Phoenix could be heavily skewed based on how much heat is trapped in asphalt that simply did not exist decades ago, or near airports with planes putting out lots of heat...
posted by disillusioned at 1:09 AM on March 2

I'm not sure if I'd consider that a skew or an accurate reading of the effect of the city on the local climate, though.
posted by hattifattener at 1:13 AM on March 2

Does this NYT piece help? It certainly offers the kind of grounding and the conceptual framework to google onwards: phenology is the subject, phenological correction is the process of dealing with anomalies and recalibration.
posted by holgate at 1:28 AM on March 2

yeah I've wondered this for a long time. I'd be grateful if you actually contacted someone in an atmospheric science department and asked them for a reference to a serious article documenting global warming which takes this into account. I mean, if this hasn't been taken into account, the whole global warming thing is a fucking joke, right? So it must be in the literature somewhere.
posted by metastability at 4:28 AM on March 2

How often is there a new record? It mostly depends on how long records have been kept.
Now, daily weather records go back about 100-150 years. There are 365 days/year, so there should be 2 or 3 record days each year everywhere, just by chance.

It's the same with sports records, which have about the same length of history. There are a few records broken each year.
posted by hexatron at 6:31 AM on March 2

Something tells me that no one was keeping weather records in Phoenix 150 years ago. The town wasn't founded until after the Civil War.

When weather people refer to records and average temps, etc., they typically use a fixed span of time as their baseline. That will vary from location to location. Get in touch with your local folks and ask.

BTW, the lows in Phoenix are afffected by the urban sprawl. (Airplane heat... not so much.) There's no skewing going on. The nighttime temperature simply doesn't drop as low as it would if Phoenix wasn't there.
posted by justcorbly at 7:38 AM on March 2

There's a chain of government entities that have been responsible for monitoring the weather through the years. Today it is the National Weather Service, which morphed out of the Weather Bureau, which came out of the USDA and the Army Signal Corps back in the 19th century. All have taken weather measurements seriously. There are training programs and rules to follow for volunteer cooperative observers, for example so that a person in Phoenix measures the max/min, rainfall, etc. in the same manner as someone in North Dakota or Boston.

That said, weather measurements were not initially intended to be used as climate records. Through the years station locations have changed, urban areas have grown, instruments have changed, measurement times have been altered, etc. That makes trickier, but not impossible. A mercury thermometer is going to have slightly different characteristics than the thermistors currently in use. The NWS is smart enough to run both types in parallel so that they know what the correlation is between the two.

Similarly, the urban heat island can be statistically accounted for by looking at the correlation between affected stations and those in nearby rural areas. The definitive paper is Urbanization: Its Detection and Effect in the United States Climate Record by Karl, Diaz and Kukla (1988) - (article may be behind a paywall).

The longest record that I found for Phoenix ran from 1895 to 1999. Sky Harbor's record started in 1930 and continues today. If you dig into those records you'll see that the National Climatic Data Center has a detailed history of each.

One thing you can do with these records is compare nearby cities for the same day in 1921. Did other cities in Arizona also have record, or near record highs on this day? If so, maybe the Phoenix record is correct. If not, maybe there was an error with the Phoenix observation.
posted by plastic_animals at 9:04 AM on March 2

As far as global warming goes (mentioned above) the mass amount of measurements climatologists use would make a small percentage of bad measurements (be it equipment or human error) negligible. The idea here is regardless of local changes, like moving the thermometers around, putting an air conditioners next to one, putting down asphalt, planting trees, erecting a large building next to the sensor, etc when statistically analyzed there should be a change thats not only in the local region but global.

I'm hesitant to link to the blogs of climate change deniers because of the large amount of distortion there, but their criticism has changed from conspiracy to bad statistics, especially when it comes to adjusting and averaging all this historical data. I think the issue of accuracy is now a statistical/methodology issue and its been well established that surface temp readings arent perfect. The models that predict climate have already taken this into account. In science you rarely have pristine data from the exact same tools by the exact same people using the exact same methods. This is why statistics and analysis are so important. Its old news, but its new in the right-wing blogosphere right now for some reason.

Lastly, the ice in the arctic is thinning and collapsing. It seems odd to me that this observable physical phenomonen can be ignored while climate change deniers grouse about temp sensors too close to concrete or ac units. Not to mention historical satellite temp and co2 data.
posted by damn dirty ape at 10:38 AM on March 2

Also I should mention historical sea level changes. It seems inconceivable to me that even these would be wrong as well as temperature and no one can make heads or tail of the data. Instead you can analyze this stuff and see a real trend towards warming and sea level change. That's really statistics 101.
posted by damn dirty ape at 10:42 AM on March 2

I mean, if this hasn't been taken into account, the whole global warming thing is a fucking joke, right?

I honestly have no stance on global warming, but this statement is ridiculous. Global warming is about trends over the centuries and millennia; it has nothing to do with the precise temp. in 1921 and whether the urban heat island might have thrown measurements off a half a degree.

The temperature that ice melts is a (relative) constant. The temperatures and conditions under which certain plants and animals thrive or die out is also known and can be assumed to be relatively constant. These are the types of thing that are relevant to climatologists.
posted by coolguymichael at 11:50 AM on March 2

I also *have* to believe that the lows in Phoenix could be heavily skewed based on how much heat is trapped in asphalt that simply did not exist decades ago

That's called the Urban Heat Island. Asphalt, concrete, and other building materials absorb heat more readily (and release it more slowly at night) than undeveloped land. Often you will have a weather station that 50 years ago was surrounded by complete wilderness, and now it's surrounded by a parking lot. This might appear like steadily rising temperatures when really it's reflecting steadily sprawling development. So you have to use multiple temperature readings and a variety of locations to accurately determine what's going on.
posted by jschu at 5:37 PM on March 2

Sweet, thanks all.
posted by disillusioned at 6:33 PM on March 2