Hi Ingemar!
That particular scale is my own. Here’s what it means: first, urbanization has the well known effect of raising temperatures. Maybe a bit less obvious (but still well-studied) is that it raises nighttime temperatures more than daytime temperatures, especially in relatively clear, calm wind conditions. I’ve also noticed that the temperature drop in cities is more gradual, but also a bit more steady during the night, compared to outlying areas in the “countryside”, which tend to have a dramatic temperature drop around and shortly after sunset, and then gradually slowing during the night.
This is mostly related to the failure to form a significant temperature inversion over the slowly cooling concrete and pavement of urbanized areas, while open fields and grass in rural areas can develop strong nighttime inversions, which also tend to break up quite quickly around mid-morning. Not only is the diurnal temperature cirve of an urban area less extreme (lower amplitude) - it’s also a somewhat different shape (with some time-lag or phase-delay).
When I assign urban heat island effects, much of that decision is based on looking at the actual amount of urbanization (buildings, houses, parking lots, roads), but it’s also a result of testing and tweaking to try to match the customer’s weather station’s actual characteristics. I’ve found that another factor comes into play heavily: the elevation relative to surroundings. Local high-elevation spots, either because the ground is locally high OR because the thermometer is mounted high above the ground (on a roof, for example), behave a lot like more urbanized locations. For this reason, I’ve often given a remote, non-urbanized location a somewhat “urban” heat island effect, like 60 or 70, because the thermometer is relatively high up compared to most of the surrounding area.
In professional mode, it’s possible to play around with this number. However, my recommendation is always to keep it like it is unless I can revisit the situation, because I’ve already gone through the process of trying different values to see what works best.
So, again, it’s my own creation, and it matters a lot! 
Tom