reply to post by PuterMan
NO, the vertical axis units are not given in either your first or last plot, but only in the middle, small plot, so yes, I did have to infer what the
units were, because they were not provided in the plots I asked about. And since I was correct in my inference, I wasn't confused on that point. I
also now understand what the base average for 1951-1980 is about; thanks for explaining that because that was unclear -- to me at least -- in your
With reference to the issue of Antartic expeditions, there is a BIG difference between knowing some temperatures for a region from various expeditions
to the region, and having an accurate, regular, long-term record of measurements if one is trying to construct a consistent database. That is my
point. Nothing you have said since has undercut my point on this issue. Moreover, given that the base temperature averaged is for the period
1951-1980, and that these AWS's were not installed until the mid 1980's, there was still poor sampling of this region for the "basis" period of
1951-1980, and thus better accounting/correcting for these regions in the past with much sparser observations could well explain the differences in
the database that you see.
Another poster mentioned how these datasets get updated from time to time. Admittedly he only referred you to a technical report for the latest data
setupdate. Did you, however, look at the references for this report? There are both a in-press journal article and a submitted one. You might
contact the first author of the one that is in press and ask for a pre-print, or you can wait until it is published and check it out. I'd imagine
that it would detail how the dataset is established as well as provide references to some of the previous ones.
As to my point regarding satellite observations of temperature vs. conventional measurements, I believe the accepted orthodoxy now is that the remote
sensing of temperatures worldwide gives a more accurate estimate of mean global temperature, and that this value is different from that based solely
on conventional, ground-based measurements -- in no small part because many more and better observations of ocean surface temperatures can be
gathered. Keep in mind that about 3/4th's of the earth is covered by ocean. I remember reading and article about that in EOS some years back.
So, I imagine a correction was calculated to compensate for this difference when only conventional observations were possible. Why this correction
systematically very slightly reduces the values in the 1880's while increasing them in the most recent decades, I cannot tell you for sure, but
imagine it still has to do with station coverage. Like I said before, coverage in the southern hemisphere was likely far sparser than in the north,
back in the 1880's; and until the satellite corrections were added for to all years' data, the more recent, i.e. post-mid-1980's data were likely
depressed because of these additional Antartic stations that you mention.
Have you checked carefully what these dataset differences are at other periods, say in 1920 and 1960? Looking at the middle plot in your OP, it
appears that there is an inflection point in which each year's compiled dataset for the entire period flips its systematic bias; this appears to be
around 1943. WWII gave rise to a lot more weather stations and the information was crucial for military operations. I suspect that the systematic
biases your are seeing has to do with the effect of this very probable increase in station coverage at this point. I think it would be worth your
effort to make a plot similar to your bottom one in the OP for the periods of, say, 1930-1940 as well as another for 1945-1955; I suspect the
systematic trends will "flip" for these two periods.
Of course, you are likely to say well this is just more fudging of the date to show that earlier temperatures are lower than later ones, but I would
counter that this shows the effect of a change in the number of measurements for calculating the mean global temperatures because of a significant
increase in station coverage.
In any case, the lack of a response on your part to my point, which is that this database systematic bias is less than 7% of the signal that suggests
a 0.9 deg C. average gloabl temperature change over 132 years, is conspicuous in its absence. Even taking the worst case systematic bias, i.e. -0.06
deg C for early data and +0.06 for later data, there is still then a 0.78 deg C change over this period, which is still a significant increase, albeit
not as big as the one assuming the updated changes in the datasets are warranted, i.e. approximately a 0.9 deg C change -- or so I eyeball it from
your middle graph.
Have to continue my response in another post because of character count...