Cold ground shows up as cloud in generated images

[wlbr]

Hi,

I understand the problem of the missing data for the polar regions. And the need for the substitutions of these gaps. Ont he other hand, most of Europe seems to be under clouds - all the time. I tried to compensate this using the cloud-threshold config value, but that does not really help. I am using Xplanets mercator projection and it seems to me, that something like 50% of the data is actually static. I think that this behavior had need been that way all the time, only since some months now.

Could that be correct or am I doing something wrong?

Regards, Michael

[matteason]

Hi @wlbr,

You're not doing anything wrong, no - there's no static data beyond the poles but I expect what's happening at the moment is that ground temperatures in Europe are so low that they're falling into the range of what the script considers to be cloud when it's processing the source infrared image (more on this issue at https://learningweather.psu.edu/node/23), and these temperatures are fairly consistent day to day so it looks like it's always cloudy. I do my best to compensate for this but ultimately it's a limitation of relying on infrared rather than visible imagery. If anyone is able to provide any suggestions on how this could be improved I'd be really grateful

[wlbr]

Thanks a lot for the explanation. Actually, I was not aware that you are not really using visual images of the clouds, but are kind of "guessing" for clouds because of the infrared data. Somehow fascinating, despite of the limitations.

Thanks again!

[matteason]

Ah yes that's actually something I should definitely make clearer in the readme. Basically it's using infrared as the 'base' data because it covers the whole Earth and then augmenting that with a visible light image, which adds more detail and includes some clouds that are completley invisible in the IR, but obviously only works in areas where it's daytime. One downside of that is that if cold ground is interpreted as cloud from the IR, the visible light data won't correct that (because it's only adding clouds, not removing them)