Some cities at different zoom level seem to have different geopositions
-
I've reported an issue similar to this a year ago, when temperature changed between closer zooms. Now it's between 7 & 8, and I couldn't find anything on the topic in the search.
Here's zoom 7, Almaty city shows 19°, apparently from a position in the mountains.
https://www.windy.com/-Temperature-temp?temp,43.309,76.864,7
Same place at zoom 8: Almaty city temperature is now 25°C from a place seemingly much lower. Also notice Kapchagay temperature change the other way around, from 31°C to 25°C.
-
@Dmitri-Lebedev thank you for your feedback. Can you please let me know from which location did you try to access Windy? Also which device and browser do you use, please?
I am not able to replicate it on our end, so that is the reason why I am trying to get further info about your case.
-
@Ondřej-Šutera I'm accessing this from Almaty city itself. The browser is Firefox 113.0.2 on Ubuntu 22.04.
-
@Dmitri-Lebedev I checked it right now, and also can't see the problem.
The screenshots I posted are from morning time, when temperature rises quickly.
I wonder what's the dataset that is displayed? I bet the dataset the browser displays is different for different zooms, to reduce downloads.
If the positions of the city labels don't alter between zooms, then maybe the ds for lower zooms got cached for longer? (Because in the morning the temperature was rising quickly as the sun was rising, and temperature could possibly grow this fast, if the modelling point (or station), that gave the results, was close to mointains.)
-
@Dmitri-Lebedev I will monitor this issue, so I can get further info about this. Please let me know, if you meet the issue again.
-
Just checked the map right now (08:35 UTC), another city with the same glitch in Tashkent (local time was 13:35).
zoom 5, temperature too low, seems like from night time
Tashkent city is far from mountains, so this error wouldn't have happened even if city coords were jumping. So it seems the zoom 5 dataset has temperature from night time.
-
@Dmitri-Lebedev
There could be a small variation of the predicted temperature shown below city labels when zooming. But not with such a big discrepancy as you show. Or it is a recent bug.
It is not a different geoposition of the city.
See similar possible small differences:
https://community.windy.com/topic/9382/forecasted-values-on-map-change-depending-on-zoom-level?_=1686564204589 -
@idefix37 Seems to be the case, indeed (except that the variantion isn't small, it's whole 12°). The glitch is happening only in cities close to mountains, at particular zooms: Almaty is right next to them, so the glitch happens at closer zooms. Tashkent and Bishkek are farther, and the transition happens at zooms 5-6.
Tashkent temperature changes from 26°C to 38°C, Bishkek from 20° to 32°C.
-
@Dmitri-Lebedev
Further to your post, yesterday and today, I have checked the temperature predicted at Tachkent, at different zoom levels. I could not see any difference. Right now 37°C is displayed both at low and high zoom levels (on mobile app, I have not checked with browser version).
It should be nice to see the timeline on your screenshots to check they show the same time stamp.
-
@idefix37 The glitch disappeared couple of hours later for me as well.
As a software developer, I bet there's nothing to do with the browser, because it's not CSS glitch, and you don't do modelling on the client side. For my years of frontend development, never Firefox has displayed same data differently than, say, Chrome.
When the glitch happened, the temperature looked like from much earlier hours (+38 was at 13:00 local time, and +26 was a perfect temperature for night time).
Your data is stored in JSON tiles, and I bet they're generated on the backend and stored statically, to reduce server load. Here's Tashkent's tiles:
Zoom 6: https://tiles.windy.com/labels/v1.4/en/6/44/23.json
Zoom 5: https://tiles.windy.com/labels/v1.4/en/5/22/11.jsonI suppose the files in
/5/get cached for longer than in/6/(or/5/is updated rarer than/6/). A simple Python script that would get these 2 files every hour and compare the values would show when this happens.(edit: looks like it's not direct values of temperatures but indice of an array which is loaded elsewhere.)
