While this is interesting, it really strikes me that there isn't that much data available. Would this pattern hold if you saw another 100 years into the past? Will it hold an additional 100 years in the future?
>Would this pattern hold if you saw another 100 years into the past?
Yes. In the 1700s The Earth was just beginning to emerge from 'the little ice age'. There is some controversy over whether the medieval warm period may have been as hot, or hotter than present.
Caldeira has apparently used the GISTEMP LOTI Global mean series, but this seems to have been smoothed to remove any peaks. I tried plotting this particular series at.
The pattern he describes in his tweet is not so clear with higher resolution data. In particular, there is a pronounced spike in the late 1930s, and another in the mid 1950s:
Furthermore, this does not seem to be replicable using other available temperature series. For instance HADCRUT4 global mean series from 1880 to 2013 seems to have a peak in the late 1940's that is not present in Caldeira's graph.
In short, unless I am missing something, his claim is based on the one outlier among temperature series which can be made to support it (if you squint your eyes real hard, I suppose).
That is a true statement on a 150 year timescale (or possibly even a millennial timescale) but on a geologic timescale it is very false. Also, it is misleading to attribute this deviation solely to the activities of man when solar output has been on the rise for nearly two centuries (http://en.wikipedia.org/wiki/File:Solar_Activity_Proxies.png)
Cute, but... doesn't seem very significant?