The author's model doesn't really say that Reykjavik is a good option. From my understanding, the graph shows the cities which have less variation than all cities colder than them.