In the introduction, the goal of this project was set: to present the available global temperature data in a different yet correct way. You will have to trust the correctness blindly until the code is published in a follow-up post. For now, let’s focus on what’s different.

The data set is rather large. The largest file with unadjusted mean temperatures contains more than 6.7 million temperature measurements from 7280 different stations, monthly averaged. A human looking at the raw numbers in the file is not going to see the forest through the trees. So most representations simplify these millions of measurements into a single clean line. The data is averaged, clipped, normalized, homogenized, distributed, concentrated, smoothed, polished and massaged. A 90 minute symphony performed by a philharmonic orchestra is reduced to a 4 bar tune played on a plastic flute. It should be clear that a lot of information gets lost that way. In fact, so much disappears that it is possible to obtain almost any desired outcome.

While searching a hockey stick, scatter plots have presented both raw temperatures and the “anomalies” in almost all their variation and complexity. Almost, yes. Because the temperatures in the files are monthly averages, it made sense to plot the evolution of the measurements from all available stations and all available years … as dozens of individual graphs, one for each month.

Let’s combine these dozens into a cyclic animation. The first dozen shows the raw temperatures as they change during the year:

Yearly variation of raw temperatures

Note the different climate regions: the tropics, where temperatures are nearly constant year-round, the northern hemisphere where temperatures approach tropic temperatures in July/August and the underrepresented southern hemisphere with includes some isolated low, low temperatures in the antarctic winter.

In the next animation, a mean temperature was calculated for each of the 7000+ stations by averaging all available temperature measurements for each station individually. Fear not, we’re not throwing any information away: the animation shows the difference between the raw temperature data and these means:

Difference between raw temperatures and station mean

The tropics, northern and southern hemisphere are clearly visible as separate ‘blobs’. With some imagination, one can almost feel the wobble caused by the tilted axis of the earth, and the resulting 20C to 50C difference between summer and winter outside the tropics.

Finally, instead of a single mean, a dozen of means – one for each month – is computed per station and the differences plotted. This produces an animation of the anomalies as they change intrayear:

Evolution of anomalies throughout the year.

The anomalies are evidently centered around zero. But have a look at the wide range. In the winter of the well-represented northern hemisphere, anomalies of +5C and -5C are common with extremes reaching + and -20C. And think about it: in the NH the mean temperature of the 31 days in January in one year is likely 5C hotter or colder than in another year.

These three animations unveil earth’s climate as known from the direct but imperfect temperature record.

Where is the hockey stick ? Everywhere you look. Pointing up, down, left and right.


About this entry