You are the creator of the Bortle Scale! It's an honor to "talk" to you. Sorry it took me another post to see where you are coming from.
As you say, the zenith sky brightness (i.e. light pollution maps) is a single metric of sky brightness and the Bortle scale is a multivariate metric of sky quality (i.e. takes into account zenith, close to the horizon and transparency as well). If I had to choose between a good Bortle site and a good light pollution atlas site, I would choose the Bortle because (1) it is multivariate index and (2) it is based on "on the ground" observations.
There are some important comments I'd like to make that may help clear up or at least help you better understand your issues with the light pollution atlases.
1. You are correct that the most widely used light pollution map is biased. The https://www.lightpollutionmap.info map, which is not my work, is systematically biased toward dark skies (i.e. it says the sky is darker than it really is). For one of my favorite observing sites in Wisconsin, lightpollutionmap.info say the SQM is 21.74. In reality, on excellent nights, I get a SQM of 21.5. The 21.5 is consistent with my maps, which show my site on the boundary between light and dark green. The original 2001 light pollution atlas (Cinzano et al 2001, not my work either) did not have this huge bias, and my model is based on that work. Lightpollutionmap.info, on the other hand, shows a newer 2015 atlas described in this paper:
Falchi, F., Cinzano, P., Duriscoe, D., Kyba, C. C., Elvidge, C. D., Baugh, K., ... & Furgoni, R. (2016). The new world atlas of artificial night sky brightness. Science advances, 2(6), e1600377.
The 2015 atlas changed the ratio of upward propagating to (nearly) horizontally propagating (increased upward propagating too much). Since horizontal propagating light is the main culprit, especially more than a couple kilometers from a light source, they biased the map dark. For some reason, this atlas became the most popular. Perhaps because it has the most "information", which gets me to my next point below.
2. When you click on the lightpollutionmap.info maps, you get the coordinates, the elevation, the SQM reading to the hundredths place!, and the Bortle scale. The site assumes that there is a one-to-one correspondence between zenith brightness and Bortle. I do not claim my maps are the Bortle scale and I intentionally do not mention the Bortle scale on my site. When I introduced the previous version of my maps (2016 version), I listed out 5 important points, the first one was short and sweet: "These maps are not the Bortle Scale". I wrote this because the widespread conflation of the two. I think this conflation started innocently with a local study by the Northern Virginia Astronomy Club (NOVAC). They tried to find the best correspondence between the Bortle Scale and the recently published 2001 light pollution atlas (Cinzano et al 2001). For their region, with its particular blend of cities, town and rural lights, the correspondence might have been pretty good. But this would not necessarily extrapolate to a different region where light pollution is dominated by a single large city, or where light pollution is the result of a more or less spatially uniform light sources.
For lightpollutionmap.info, the even bigger problem is that they took the 2001 maps/Bortle scale comparison and used it on the 2015 maps, which are biased. Hence, my favorite observing site above is "Bortle 3" according to lightpollutionmap.info, but the site is Bortle 4 on excellent nights. On the other hand, if you use the zenith brightness on my maps, then the NOVAC relationship says my site is Bortle 4, which is consistent. Nevertheless, the correspondence can never be expected to hold in general because the Bortle scale is more than just the zenith, so I do not even mention Bortle on my site. Lightpollutionmap.info, on the other hand, gives this extra "information", never mind that such a one-to-one correspondence does not exist.
3. Regarding the finger of dark skies in my maps: I think this is a consequence of the fact that horizons are more affected by nonlocal light sources than zenith. While my zenith maps are a smoothed version of the upward light sources, a map based the sky brightness closer to the horizon would be even more smoothed. So the dark finger may exist for zenith brightness but things are too smeared out for lower elevation brightness. Since the Bortle scale is affected my much more than zenith, the dark finger does not exist for a Bortle scale "map". Since many interesting objects do not rise to zenith, I agree that the Bortle scale is better than zenith brightness. I've thought about making a separate set of maps showing the brightness closer to the horizon, especially the southern horizon (for us northern hemisphere people). So far I've avoided the trouble of modifying my calculations because I assumed I could guess the brightness of the southern horizon based on the zenith brightness further to the south. Based on your observations this is not a good idea because the lower elevations are more sensitive to distance sources so you can't just "shift my maps north" to get a good idea of the southern horizon. I think I will be adding maps based on lower elevation brightness at some point in the next year (this is not my day job). While I'm at it, I guess I'll also do some northern horizon brightness maps for those lucky southern hemisphere people with all the good stuff and dark skies.
4. The darkening trend seen in the maps is almost certainly due to the change over to LED fixtures, which (1) tend to be much better shielded and (2) have a different wavelength spectrum. This could be a real thing, or it could artificial. After I updated previous maps, I've had people share opinions both ways as to whether there has been darkening due to LEDs or not (both private messages and on the forum). I do not live in a region with the "darkening skies" so I do not have an opinion. For sure, LEDs are better shielded. On the other hand, the detectors on the satellites are more sensitive to the spectrum of sodium lights than the spectrum of LEDs. So an LED could appear dimmer when it is actually not. Mostly likely, both of these competing effects are operating, but I'm not sure which effect "wins".
Edited by DaveL, 22 September 2021 - 11:22 PM.