Wednesday, January 6, 2010

Quality of U.S. Temperature Data

Photo (courtesy of Anthony Watts, Dr. Roger Pielke Sr., and

This is a photo of a surface weather station in Hopkinsville, KY. See the white bird house looking thing hanging off the building? That's the thermometer. Notice that it comes off of a chimney of a brick house, it's just above a black asphalt driveway, that big white thing is an A/C compressor, and it's directly over a Weber grill.

The data from this thermometer becomes a part of the US Historical Climatological Network (USHCN) Station of Record for Hopkinsville, KY. Initially, this seems rather humorous. They are literally "cooking the books" at this station. But then it sinks in that this station data is being compiled in with the rest of the data from the country, and it is from these stations that we determine the change in temperature over time.

You may wonder if stations such as this one is unique. Sadly, it is not. There are many similar graphics available here of weather stations that would have their local environment affecting the data.

Have a look at this weather station below at an airfield in the US. A great number of Australian weather stations are located at airfields like this one. About ten years ago a vast majority of these weather stations were made more “user friendly” for the techs collecting the data. This was done by laying a bitumen pad around the weather station. Duty of Care was the so called reason so the techs weren’t wading through mud to collect data!

Do you think that siting a weather station on a black bitumen pad would affect temperature readings from that station?

Damn! Here we go again. A dumb arsed retired truck driver who had to sit for his Intermediate Certificate twice knows that it is a lot hotter on a tarred road then it is in a field!

You think our climate scientists don’t know this?

The official record of temperatures in the continental United States comes from a network of 1,221 climate-monitoring stations overseen by the National Weather Service, a department of the National Oceanic and Atmospheric Administration (NOAA). These surface stations are supposed to be setup in accordance with guidelines given by the NOAA's Climate Reference Network (CRN).

In 2007 a project was initiated by Watts to visually inspect and document the quality of the surface stations. So what did they find? Some of the stations that they found are pictured at the link above. We can see the thermometers located next to asphalt parking lots, located next to air-conditioner exhausts, located on brick buildings, located next to BBQ pits, located next to trash burning barrels, located on rooftops and near sidewalks. All of which absorb and radiate heat. These all influence the thermometer to read higher than it would otherwise.

Let's look at the numbers.

There are 1,221 active climate-monitoring surface stations in the USA. Of those 1,221 stations, 948 have been evaluated by Watt's Surface Stations project. Of these 948, 90% have been found to be very poorly sited. Poorly sited means that they are likely reporting higher or rising temperatures. This means the overwhelming majority of stations in the USA have been reporting bad data. (Please note that the USA data is generally considered the best data in the world. Mmmm!)

I wonder just how bad the Australian weather reporting stations would rate sitting on their little bitumen pads?

See the link to Anthony Watt’s site here:

So, now we know that according to the NOAA CRN's guidelines, 90% of the surface stations in the USA are not capable of giving good data. They have errors greater than the amount of global warming we have supposedly experienced in the last one hundred years. It would be easy to assume that the other 10% give good data all the time.

Unfortunately that is not the case. Let's see why.

To collect surface station data site observers are given a card on which they are supposed to fill out the maximum and minimum temperatures each day. At the end of the month they are supposed to submit their card to National Climatic Data Center so that they can be compiled into the national database.

Well, what happens when the site observer doesn't work on a particular day? No readings occur. What happens when that observer is off because of a holiday? No readings occur. In fact, in Marysville, California, at Chico University Mr. Watts found that the temperature form for July 2007 had only 14 of 31 days completed. That is less than half a month's worth of data.

Missing data is not a rare phenomenon. Many sites have missing data. So even if a site is rated as capable of providing good data by the NOAA CRN, the data still must be read by someone. If that person is missing half the month, then you're not getting a significant quantity of good data.

So what happens when stations have missing data? From Watts' report: "[There exists] a data algorithm used by NCDC called FILNET, short for Fill Missing Original Data in the Network, that is used to “infill” missing data using interpolations of data from surrounding stations. After reading about it, I came to the conclusion that NCDC uses FILNET to create “missing” data where none was ever actually “ measured."

Basically, if data is missing from a certain site then data is taken from near-by sites and adjusted to fit the missing data site. Wow.

So, we see that 90% of US stations are not capable of providing quality data, and then we see that sites which have missing data have data filled in from surrounding sites by the FILNET program. So even a site that is capable of producing good data, if data is not read for a day, could have bad data filled in for it by this FILNET program. This means that even the 10% of sites that are capable of providing good data are contaminated by the 90% of sites that have bad data.

So now that we see the quality of data from the stations is overwhelmingly poor, let's look a little more at the numbers and types of stations.

In April 1978 there were over 6,000 surface stations in the US being used, that has now dropped down to around 1,200. So, we know somewhere around 4,800 surface stations dropped out of monitoring temperatures. The vast majority of the stations that dropped out were rural stations.

This is unfortunate because the rural stations are the ones that tend to give the most accurate data (because of less urban heat contamination.)

I wonder if this has also happened in Australia?

The majority of content in this post came from:

No comments:

Post a Comment