The satellite indices from UAH and RSS try to measure temperature in the lower troposphere by a very complicated process of deduction from microwaves emitted by oxygen. Recently, they indicate temperatures that are rising more slowly than surface measures. So they are popular with naysayers, who gloss over the fact that they are actually measuring (or attempting to measure) very different things.
A prominent recent example was Ryan O'Hare, in the Daily Mail, headline:
"2015 may NOT have been the hottest year on record after all: Satellite data shows temperatures were lower than first thought"
Totally inaccurate, of course. Satellite temperatures do not contradict the surface record. They say (less reliably) something about the temperature in a different place. But this seems to be being worked up as the standard distraction from the 2015 record temperature. For last year's record, they tried Oh, but we can't be sure. This time the margin will be too great for that, so satellites will have to do.
There is a WUWT post here which tries by comparison to suggest that surface temperatures are trending differently to troposphere, and are therefore wrong. It's one of many, and graphs like this crop up:
I drew attention here to the huge adjustment that was made to UAH this year, in going from V5.6 to V6. In fact, the V6 plotted above is still in beta, and V5.6 is still maintained. So recently I restored it to the datasets that I collect and plot, and you can see it in the active plot. So I thought I would add V5.6 to this comparison. And sure enough, V5.6 is far closer to the surface measures than it is to V6 (or RSS). Below the fold is a snapshot, where the anomaly base is 1981-2010 for all:
This is monthly data smoothed with a running 12-month mean. The land/ocean indices GISSlo and NOAAlo are very close. RSS and UAHv6 less so, but yes, clearly diverge from surface. But again, UAHV5.6 is much closer to the surface measure. Now it may be that there are good reasons why UAH made such a big change only a few months ago. But it's a very weak basis for saying that the surface measures are wrong. Of course, the greater weakness is that they are measuring different places. But if UAH sometimes agrees, sometimes not, it could hardly ever be a convincing counter.
I find the "trendback" mode of that plot also revealing. This shows trends from a varying start, shown on the x-axis, to present.
Now you can see that UAH V5.6 behaves very like the surface indices. V6 and RSS are different.
There is an interesting aspect of that graph. At WUWT, Lord Monckton posts a monthly series on the Pause, allegedly of 18 years duration. At least, the series was monthly, but missed this month. It may expire. The "Pause" is measured where the green RSS trend plot first crosses the very faint zero trend line. You'll see that the crossing is now very brief, and with a couple more warm months, will go away. There is a second crossing near 2001, but that won't last long. So even by cherry-picking the least of the troposphere readings (and ignoring the actual surface), the "pause" will shrink, on that measure, to about seven years. That is, from an interval tilted by the 1998 El Nino, to one tilted by the 2010 version.
Update. UAH V5.6 still shows 2015 as the third hottest year. The first of my plots shows why. The troposphere really is a different place. Temperatures respond very strongly to an El Nino event, but late, in the year following. That is when records are set. So 1998 and 2010 were hotter than 2015. They won't be hotter than 2016.
Update See also Kevin's SKS article and comment