Best posts made by Filip_K

More information on how to read Thermals is provided under "info" icon

@idefix37 JMA was the first place I went to regarding researching models for Japan. LFM has only 10 hours of forecast so I was thinking more about using MSM.
Final choice depends on demand though, as LSM has much finer grid (2km vs 5km of MSM)

@gavin-q Consider picker just a preview tool, detail forecast is much more precise. When zooming, for example, picker data change quite a lot as different data tiles are used.

Example (speed and direction changes):
closest:

further away:

far away:

Unfortunately, Copernicus prolonged data outage. No exact date is given, only that data should be back by end of December. We will monitor the situation and get forecast back ASAP

@Wojciech-Dobrowolski The scale is non-linear and pink color in default scale corresponds to 500mm. See color settings for rain accumulation below

@idefix37 There is something seriously wrong with merging AROME and ECMWF. Notice that after 5 days (that's when forecast step of ECMWF changes from 1h to 3h) forecast goes haywire. It is more visible with 1h forecast, but 3h forecast is wrong too. Multiple days (notice temperature going up and down - "red hills") get mashed up into one. We will investigate and fix this, thank you for the report.

3-hour step

1-hour step

@Skip-Freedom According to Copernicus, forecast data is expected to be back up and running by next Tuesday; on 2022-12-13

I see. The *_best_track.kmz (which is our primary source for hurricane tracks) is unfortunately updated quite late. As you mention, around 3h late. I'll look into this to make sure we get as up-to-date data as possible. Thank you for noticing this

direct link to the plugin is https://www.windy.com/plugins/windy-plugin-sailcloud

@_AVX_ Raw data or simple images are both good. Aviation portal you mentioned (https://aviation.weathersa.co.za/#showSAWSRadar) is only for aviators so we can't legally use it. Even if we could, there is a lot of clutter in the images provided there (city labels, land, storms outlines and storm forecast outlines, ...). Long story short, the clearer data the better.
Other link (http://www.weathersa.co.za/home/radarst) only shows storms, not radar scans directly. That also greatly complicates incorporating the data to windy.

If you know of any other source or someone in SAWS/AfriGIS who might be more willing to cooperate, feel free to let us know. We'll be very happy to add South Africa coverage.

NOAA GFS model has not been updated since yesterday 12:00 UTC due to connection problem to NOAA servers.

Update: Connection to NOAA servers has been restore and GFS model is updated again.

@JoeBFLL Latest hurricane position, wind speed and pressure is now obtained from NOAA advisory so it should be as up-to-date as possible.

Thank you again for pointing this out to me

@erick3195 Thank you for your insight. Our source is v2 dataset which still seems to be unavailable (no recent data): https://mf-models-on-aws.org/#arome-france-hd/v2/

@hoffam-au Greenvale and Richmond radars have been added, thank you for the report

before

after

@tanpat Hello, it might be that windguru is using old v1 dataset for 1.3 km Arome but v2 for 2.5 km Arome. That is why 1.3 km Arome on windguru works and 2.5 km doesn't.

Since we're using v2 dataset (which is recommended by MF and is also documented - v1 isn't) we unfortunately won't be getting any data updates before v2 dataset is back and running.

Hi @gavin-q, can we request full-screen screenshot next time you observe the discrepancy? I've tried to replicate your issue with no luck. Are you sure you've selected wind at the surface (which is shown in the basic forecast)?

@vicb Sorry, that is something we want to keep to yourselves yet.
We're open to discussion on what should be forecasted under which conditions though.

@gavin-q @Korina I was able to reproduce same error with clean Firefox installation (and other chrome-based browser as well)

https://www.windy.com/-25.486/30.301?-25.504,30.322,13,i:pressure,m:cWGagZF

While detail shows wind from around 260deg, picker and particles show wind from around 280deg. It might be that in low wind speeds, discrepancies between map (and picker which uses map data) and forecast detail happens.

As map data go through transformations during import (from source projection to web-mercator), slight differences happen and are most visible with low values (similar thing happens with low precipitation). Point forecast detail has slightly more accurate data while map is (due to these technical limitations) intended more for general situation overview.

I'll investigate this issue further now that I have something to lean on

@balloni55 I was able to replicate your issue on 1920x1080 screen when I set up browser zoom to 130% (see bottom right corner on the screenshot)
Is it perhaps same in your browser? Have you tried changing browser zoom?

@Ondřej-Šutera https://www.windy.com/60.860/8.550?60.825,8.550,11