An illustration of the trough-to-peak trick often used to state a false increase in the “trend” of global temperatures.
Innocent sounding periods like “the latter half of 20th c. ” or “last 50 years” are used to give the impression that it is an arbitrarily chosen round number, in reality it is a choice which determines the result.
The title of the graph is a quotation of a “finding” of a recent peer-reviewed paper  but, as is shown here, a simple cosine which has no long term “trend” can be used to produce the same result.
There is a roughly repetitive pattern that is a significant part of NH temperatures, with about this period and timing.
This shows how totally false conclusions can be drawn after fitting “trends” with a suitable choice of the time periods used.
Fitting a linear “trend” model to data from such a system, that is anything but linear in its behaviour, is invalid. Yet this practice is ubiquitous in climate science, even in IPCC reports and peer-reviewed literature such as this:
 Compo et al 2013: