Are you using the mobile application or web? In the mobile application are currently some issues, we know about that, sorry for any difficulties.
If you add favourites in the web browser, is should always add to favourites list.
We will try to fix it soon.
Note that for US discussions (and other NWS products), you'll probably want to stick to using the new API as described at the last link above as the first two links will, eventually, not work.
The latest Service Change Notice regarding NWS forecast URLs is here:
Hi! Sorry for my so late reply !
I confirm @siim statement: after disabling ublock, the red message is gone.
And ublock reports this url in its log: https://node.windy.com/node/geoip?source=testConnection
geoip is highlighted in red in ublock logs, so I guess this is the trigger.
It's not critical for my personnal use, I can just disable ublock for my page.
But in a general way, it may be a good idea to modify this keyword as suggested by @siim
For reference if needed:
uBlock Origin v1.18.16
Rain areas are more stable within the Rain-Thunder display but rain can move quickly, or else just start and stop randomly.
But with regard to thunderstorms, they are short-term events and their occurrence is often very patchy and local. So they'll rarely 'move' smoothly as what you see displayed is a series of separate storms that are forecast to form and decaying in less than two hours (usually).
But the model moves foward in three-hourly steps. So it's unlikely to become smoother until the model's time steps are also about 1-hour long. The AROME model has 1-hour steps now but its coverage is very limited.
The other problem is that the forecast location and formation of any thunderstorm can not be very accurately predicted by a model. The formation process is more chaotic than a model can reliably predict. What you therefore see within the display is only giving you an idea of how likely a storm is within that area and within that time frame. And to indicate the likely storm intensity.
Models can forecast the likely thunderstorm intensity, if a storm does form in an area within a given time. But a model can not tell you accurately if one will form or where.
So even if the display was made smoother, it still would not be reliable, it's just a guide to what's likely to occur in the area if a storm forms.
I'm agreeing with this Radek-M, the legend scale needs to be linear, or at least a lot less radically non-linear, and to cover a wider display range for most of the current overlays, as they're all fairly poorly served by the legend's nonlinear compression, and incapacity to show a full spectrum of colors in the overlay (i.e their shading is not equal). Hoping this can be remedied.