Why Satellites are Better at Measuring Global Temperature

There has been a lot of discussion about the difference between the satellite and station methods of measuring the global temperature and which is better.  This is one of the more contentious aspects of the global warming debate.  It gets people very agitated.  There are serious problems and benefits with both methods, but my main method has been to compare their response to climate events.  In that regard there is no question that the satellite method registers a greater response to events.  This is shown in the following two posts.

Detecting the Global Warming Temperature Signal

Proving the Problem with the Station Data

I will not repeat any of the arguments from those articles here.  Instead I will focus on the difference between the two methods and why I feel that the satellite method is more accurate and responsive at detecting changes in the global temperature.

Station Method

The biggest difference to me is one of resolution.  The station method uses thermometers that are located around the world and they measure the minimum and the maximum temperature each day.  This is then averaged, adjusted and combined into the global average.  Here is the list of stations used for the CRU temperature set.  It contains the location of the 4,138 stations used to measure the global temperature.

Earth’s Surface Area = 510 million km2

CRU Stations = 4,138

Each station measures 123,265 km2

This leaves each station measuring an area of North Korea or the state of New Mexico depending on which one means more to you.  While this is not a major problem by itself, these stations are not uniformly spread around the world.

In fact the real problem is that 46.7% of the worlds stations are located the region that covers only 4.8% of the worlds surface.  North America is the region in the world that has enough stations that each station is covering less than 20,000 km2 per station.  That means that outside of North America, the total station to area exceeds 230,000 km2.  That is getting into the range of 1 station per area the size of the United Kingdom.  It is even worse than that though.  In many, many places there is not even a single station in a region that large.

Generally for the station method the Earth is broken into 5×5 grids of longitude and latitude.  According to the MET office here is the spatial coverage of the Earth.

The Inconvenient Skeptic

(White) grids without a single station to measure temperature.

This does not tell the whole story either.  This map shows ocean temperatures that have been interpolated from the HadSST2 data.  This map shows the data that has not been interpolated for the ocean records.

The Inconveneint Skeptic

(White) locations without any ocean measurements.

There is also one final spatial problem with the station method.  Here is the global breakdown of stations by continent.

The Inconventient Skeptic

Global breakdown from the official CRU station list. The top 4 continents have 80% of the stations, but cover only 19% of the Earth's surface. There are only 827 stations to cover the other 81% of the Earth's surface.

That means there is exactly 1 single thermometer to cover an area the size of Spain.  That is less than one thermometer to cover an area larger than the state of California.  The warming in Antarctica is based on 60 years of records from only 23 stations on the entire continent.  Antarctica is larger than the United States.

This is not a problem by itself.  Such coverage can give an idea about the global temperature, but it will not be sensitive at detecting changes.  An area the size of Spain can have entire storm systems that only cover a small percent of the area.  It is not possible to accurately measure that much area with a single thermometer.  This is the cause of the lower sensitivity that is statistically observed by using the station method.

Satellite Method

The satellites have much greater precision.  They have resolution at a scale of 4 km2.  That is comparable to having a thermometer every 2km in every direction.  This coverage is from 85N to 85S.  Or an area that covers 99.6% of the Earth’s surface.  That is the UAH coverage at least.  The RSS is less as they cover from 80N to 70S.  The resolution of both data sets is identical.

This allows the satellite method have the equivalent of 127,033,686 thermometers uniformly spread across over 99% of the Earth’s surface.  Instead of big blocks to cover the Earth, they provide a fine picture of the Earth’s temperature.  Here is the daily example of the Earth’s temperature.

The Inconvenient Skeptic

Daily temperature map of the Earth as provided by the satellite measurements. The black areas exist on a given day, but change each day.

Even the shapes of the current storm systems show up clearly in the satellite data.  Such resolution is absolutely impossible by using the station method.  This is why the satellite method is so sensitive at detecting changes in the Earth’s temperature.

What they measure is different.  Instead of measuring the temperature at a single point a few feet off the Earth’s surface, the satellite method obtains the temperature of the bottom few kilometers of the Earth’s atmosphere.  So not only does it measure with much finer resolution, it measures the temperature of a column of air.

The result is an anomaly map that is a much better representation of the atmosphere’s temperature.  Here is the latest anomaly map as provided by the RSS.

The Inconvenient Skeptic

February 2011 RSS anomaly map for the Earth. Unlike the station method, this picture shows less resolution than the actual data.

The net difference is that the satellite method has the equivalent of 30,700 thermometers to each station method thermometer.  When it comes to detecting changes in the Earth’s temperature, this is a truly significant difference.  One that is easily detectable when comparing the response of each method to climate events like ENSO and volcanic eruptions.


Many, many problems can be found in any method of measuring the Earth’s temperature.  Long articles have been written on extensive problems for both methods.  I am sure that people will try to refer to those articles in the comments section.  In the end, the satellites show a greater ability to detect changes in the Earth’s temperature.  To me, those results matter.

If a group of scientists were to know nothing about the why that a measurement was made, but only knew the difference in resolution and saw the data that showed greater sensitivity, there is no question which method would be chosen.  The only real problem that people have with the satellite method today is that it does not show as much warming in the past 30 years as they want it to.  It doesn’t match up with their expectations, so they disregard it and use a litany of reasons to justify their dislike.

I will bet that if the satellite method showed twice as much warming as the station method with all other things being equal, the warmists would mock anyone who was skeptical of the satellite method.  Since that is not the case, it will continue to be derided until the Earth is well into the next glacial period.

Posted in Anomaly and Science Overviews by inconvenientskeptic on March 29th, 2011 at 1:04 am.


This post has 5 comments

  1. Thx John, good post. The reliability of the station data seems to finally be getting the attention it deserves (Berkley Earth Project for example). I cannot respect any scientist who argues that the station data is the “right” dataset to use!

  2. Let’s think about this in a more complete way.

    8 satellites or so measure 95% of the 5×5 grids for one minute say twice per day (i.e. 2 minutes). That’s 0.95 x 8 x 2 = 15.2 observation-minutes a day per grid. A whole storm can sneak through a grid in the 23.75 hours that the satellites aren’t looking at it.

    4138 surface stations also measure say 70% of the 5×5 grids for one minute 1440 minutes per day. That’s 0.70 x 4138 x 1440 = 4171104 / 1296 grids = 3218 observation-minutes a day per grid average. The number 4138 should probably be doubled to take into account marine readings but I don’t want satellites to look too bad.

    It looks like ground thermometers provide 85 times more coverage than satellites despite their spatial shortcomings because they have much better temporal coverage.

  3. inconvenientskeptic Mar 29th 2011

    Are you trying to say that the station data uses 1440 measurements a day?

    Or does that method in fact use the min and max for each day. That is 4138*2 = 8276 measurements per day.

    That assumes that each station reports the data each month for each day. As this is not the case there is yet another variable in the station data.

    Again, the issue is with detecting changes to global temperature. The coverage is only the reason why satellites detect change more accurately.

  4. 4138 is an exageration.

    CRU claims Masset as a station.

    GISTEMP says neither Masset station has data past 1968 or 1981.


    Kamloops is on the CRU list. GISTEMP says no Kamloops data past 1990.


    Barkerville – 1990 is the latest data.

    I think the list of stations reporting current data is closer to 1000 than 4138.

  5. inconvenientskeptic Mar 31st 2011

    I won’t take the time to study the data from each station simply because nothing of use could be learned.

    A study of the results shows that the station data is less sensitive. I am satisfied with the reasons why they give less reliable results.

    But that gives me an idea. 🙂

Web Design & Dev by

Mazal Simantov Digital Creativity