Much more information about where this graph comes from and what it means is on the web page http://sks.to/escalator.
It was therefore appropriate to make a scatterplot of this data (since one of the RVs was time, this kind of scatterplot is called a timpelot, as we have said in class).
The explanatory variable is the year, the response variable is the temperature.
The red and blue lines clearly show different ways of putting regression lines through the given data. The red line seems to be the line through the entire scatterplot, while the individual blue line segments are best lines (presumably the least squares regression lines) through five successive limited time-windows of this data.
This graph demonstrates beautifully the principle that what data you choose to include in your analysis strongly influences your results. In some sense, for this particular case, this is the fact, which we have discussed in class, that regression lines are highly sensitive to scatterplot outliers — think of data outside each time-limited as outliers with respect to that window.
Another famous thing we have studied which is relevant here is extrapolation: if you had done this analysis in 1997, with data you had that just happened to date from the last decade, you would calculate the third blue regression line in this figure. But to use that line to predict the next decade of global temperature would have been gross extrapolation, and the real would data for that next decade was very much off the 1987-1997 prediction line: so much so that today's climate change deniers have to put in two more downward sloping steps so that they can say, at any moment in time, "the trend line today is for decreasing temperature" although they need several short steps, each higher than the previous one, to make that true.
[BTW, this deceptive interpretation of the global temperature data has been described as going down the up escalator.]
The axis labels are fine for this graph.
The first negative thought I have is that it would be nice to report the correlation coefficient $r$ and particularly its square $r^2$, so that we could see how much of the variation of the temperature is described by these regression lines as a function of time — if $r^2$ is too big, we really shouldn't have any confidence in the lines.
Also, digging a little deeper into the information provided on the descriptive web page is quite frustrating: the precise definition of to what the temperature anomaly is relative is unclear; it is also unclear how the global temperature is defined — average of all weather reports on the planet? also research stations in the Arctic and Antarctic? what about the oceans? etc. While the main page about this escaltor doesn't have to have all the details, it should be possible to figure out the details, without an enormous amount of trouble, from references and links.
Clearly, however, this document serves a polemical purpose, not a scientific one. It would be nice to cite particular publications on the other side of the debate which make the mistake of claiming the blue lines are the right ones to use. As Wikipedia always says, [citation needed]!