Image talk:Global Cooling Map.png

From Wikipedia, the free encyclopedia

[edit] Sincere flattery

Well that didn't take long. Imitation is the sincerest form of flattery, right?

Could you look over the text of the image description again though? The dataset "available here" link needs updating, and I already fixed one mention of "temperature increases" to "decreases" in the text. Also, in trying to find the right map, I get a temperature shift of only -0.10 not -0.18, but that may just be that I haven't duplicated your settings quite right.

Also, could you double check that you color scale is centered? The zero seems ever so slightly yellow, but that may just be my mistaken impression. Dragons flight 14:07, July 13, 2005 (UTC)

I hadn't noticed where the dataset link went to. Fixed. The -0.18163 temperature shift is the mean of the 14,347 non-undefined values. (SEWilco 16:41, 13 July 2005 (UTC))

Okay, with those map settings, I think the value ought to be -0.11. Keep in mind that the Mercator projection that they start with overemphasizes high latitudes. One of the advantages of an equal-area projection like Mollweide is that if you average over the surface of it, you get the same value as a true average over the surface of the Earth. Also, I would strongly encourage you to recheck your color scale. Having just plotted it I see a lot more weak negative values (i.e. light blues) over the oceans than your figure shows, including in areas where you seem to have light yellows. Dragons flight 18:12, July 13, 2005 (UTC)

You're getting an average from the image? The average is from the data points, except undefined points (value 9999). Yes, I now see the color scale is off-center, I'll apply a new one. (SEWilco 21:20, 13 July 2005 (UTC))

To be clear, I am averaging the data after performing the projection onto a Cartesian grid in the Mollweide space in order that the average would accurately reflect an average over the surface the Earth, rather than the distorted Mercator projection which overemphasized high latitudes. This approach was technically convenient to me, but depending on your approach may not be easy for you. You can verify however that if the temperature anomalies, T, are reported on the Mercator (λ, φ) space then: , where φ is the latitude and this compensates for the overcounting at high latitude to give an accurate average over the whole surface of the Earth. I imagine this approach is probably the more direct and easier to understand. Dragons flight 21:53, July 13, 2005 (UTC)

The projections have nothing to do with the problem. The high latitudes are distorted due to the longitude lines getting closer together, and samples being every two degrees of longitude. Your math is better than your definition. (SEWilco 08:43, 14 July 2005 (UTC))

An equal-area projection has the property, by definition, that an average over it's surface is equal to an average over the actual surface of the Earth. This is why it is sufficient to work in the projected space for the purpose of removing the distortion. Admittedly, I may not have explained it well above. Dragons flight 13:47, July 14, 2005 (UTC)

[edit] Antarctic bit?

Some of this is a bit odd, in particular the Antarctic bit. I'm pretty dubious that there are many/any stations there for the period concerned. William M. Connolley 17:01:34, 2005-07-13 (UTC).

Which would, for example, explain why nearly all of Antartica is grayed out for lack of data. Dragons flight 17:40, July 13, 2005 (UTC)

PS. The smoothing scale associated with GISTEMP is up to 1200 km, so I imagine that in at least some cases the infilling of data is fairly generous. Dragons flight 17:43, July 13, 2005 (UTC)

I meant (of course) the bits that aren't greyed out. William M. Connolley 18:03:18, 2005-07-13 (UTC).

You know how few data stations there are in the Antarctic. All the data there is too sparse for many uses. This dataset also shows focused splatters around the peninsula on the Image:Global Warming Map.jpg. Both images are showing other expected patterns, with the changed phase of the Pacific decadal oscillation and the Siberian high chilling China during the middle of the century. (SEWilco 09:01, 14 July 2005 (UTC))