ACCESS-C coverage for Australia
-
-
Hi @David-Polášek, just wondering if there is any news with adding ACCESS-C into Windy? :)
-
@anthony10 At this moment, we have implementation of this model ready, so it should be out in the version 44, in the first half of the November.
-
@Suty amazing news to hear, thanks windy team.
-
@Suty With the upcoming release of version 44 & BoM ACCESS-C being implemented. Would you know if the devs are adding in CAPE (Convective Available Potential Energy) from the ACCESS models?
-
@anthony10 No, that is not planned now.
-
Hi @anthony10, according to documentation (for ACCESS-G and ACCESS-C, both linked from general NWP documentation), CAPE is not available as parameter. If it had been, we would have used it.
-
@Filip_K Just a side question, what about adding SWEAT (Severe Weather Threat) index? I can see that GFS, ECMWF & ACCESS-G all have this forecast.
Or would that have too much in common with CAPE index?
-
@anthony10
I'm not aware that ACCESS-G would provide such parameter (see link to documentation in my previous post). Neither am I aware of such parameter within GFS and ECMWF. Is it possible that it is some proprietary parameter that weatherzone.com.au compute from other parameters? -
@Suty I know that others and I eagerly awaiting the release of version 44 with the newly added ACCESS-C models.
Is there any updates on when that might be happening?
@Suty said in ACCESS-C coverage for Australia:
@anthony10 At this moment, we have implementation of this model ready, so it should be out in the version 44, in the first half of the November.
-
@anthony10 Hi the web app should be out in upcoming days, mobile apps, hopefully, few days after the release of the web app.
-
@Suty Thank you to all the Windy team members for adding in all the ACCESS-C models for across Australia. It’s an early Christmas gift for all Australian users of Windy and another big step to making Windy the #1 weather app across the globe!!!
-
@anthony10
I totally second this and so will a lot of other Aussie users (when they see it is added :) )
@Suty
Thanks to all of the Windy team.!!
Amazing addition especially in the midst of this current weather pattern. I am in the Hunter region NSW and after reviewing the C vs G model it is awesome. Will be very interesting to compare models today with severe thunderstorms predicted.
I would suggest an announcement to let all of the Aussie users know.
Kind Regards
Gary -
@vortexya I'm just going through all the layers with the ACCESS-C models and its very interesting that the forecast don't align closer with each other like I'd expect.
I'll have to do some investigation and figure out why BoM forecasts are so varied.
I'm looking at Canberra & Lithgow being different between the models.
Obviously, it's different parameters over the different regions but I would have expected the overlapping windows to be more similar in forecast.
It'll be interesting to see which model is more accurate in those overlapping areas.
-
@anthony10
Hey champ, how did you go with the investigation?
I also saw the difference, but to be honest the initial forecasts were pretty similar but as you went out to say 72 hrs they diverged a lot. As you know initial parameters that vary by minimal amounts when extrapolated diverge greatly
over time (the butterfly effect) . Also a bug , when switching from any other model to Access C syd over 24hrs to compare , the Access C Syd reverted to a 0500 timestamp. Anyway I was watching our local (hunter region ) weather to compare models today and apart from overestimating rain rates, the Access C Syd nailed the path and time of rainfall. Pretty impressed . Cheers! -
@vortexya I’ve email BoM for clarification, but until then I thought I’d ask ChatGPT and here is the answer.
The differences between the Bureau of Meteorology (BoM) ACCESS-C models for Sydney (NSW) and Victoria (VIC) in overlapping areas arise from variations in initial conditions, grid configurations, and tuning specific to the domains. Here are the main reasons:
-
Model Domain Setup
• ACCESS-C Sydney and Victoria are regional models designed for different geographic areas. While they overlap, their central focus areas are distinct. Each model is optimized to provide the best forecasts within its primary domain.
• Models can apply different resolution grids, meaning areas closer to the center of a domain are usually modeled with higher accuracy than areas near the periphery. -
Boundary Conditions
• These regional models derive their boundary conditions from larger-scale models (e.g., ACCESS-G). Differences in how the boundaries are ingested or updated can lead to variations, especially in overlapping zones where different data might be prioritized.
-
Initial Conditions
• Each model uses slightly different observational data for initialization, often prioritizing the data coverage and accuracy within their primary region of interest. Variations in satellite observations, ground stations, and assimilated data can cause discrepancies in overlapping areas.
-
Model Configuration and Physics
• Models are fine-tuned for the meteorology of their primary regions, meaning the parameterizations for processes like cloud formation, precipitation, and boundary layer physics may differ slightly between the Sydney and Victoria versions.
-
Resolution Differences
• The effective resolution in overlapping areas may differ due to how grid points align with the model’s core focus region. For instance, the Sydney model might better resolve features near its focus area, leading to different forecasts in overlap zones.
-
Temporal Updates
• The frequency of updates or runtime differences can also cause variations. If the models are initialized at different times, the forecasts for overlapping areas may diverge due to different starting conditions.
-
Numerical Artifacts at Domain Edges
• Overlapping areas are often near the edges of each model’s primary focus. Numerical artifacts or reduced accuracy in these regions can result in slight discrepancies.
Example Scenario:
A cold front approaching from the west might have different timing or intensity in the Sydney model versus the Victoria model, depending on how each model initialized the system and resolved its dynamics.
Key Takeaway
The differences reflect the inherent limitations and optimizations in regional modeling, highlighting why overlapping forecasts should always be interpreted with an understanding of these constraints.
-
-
This is a huge addition to Windy app. I haven't used the web version for a week or so and didn't notice ACCESS-C until today, as we have severe thunderstorms in our area.
Can't wait for the app to include ACCESS-C.
The ACCESS-C predictions for Wednesday are a good addition to ECMWF as we can see ACCESS-C has much higher estimates for showers. This is of course, an indication of convection and storms and not necessarily indicative of what my region will see but the potential is there in heavier storm activity.
-
I jsut found ACCESS-C yesterday and its so far very good for predicing rain. I hoped the ACCESS original when added would be good but it was almost always beaten by ECMWF. ACCESS-C is tracking rain exceptionally well at my home location. Just waiting for it to be added to the app.
-
Thank you so much for implementing the ACCESS-C coverage Windy team! Very much looking forward to it being available in the mobile apps.
An additional feature request for me is being able to set this model as the default model, as this will be my primary model for cloud forecasting going forward. Many thanks.
-
@tscharke I was just playing around with the widgets and see we can't select the new Australian models their either as that could have been a work around. The first request I find for setting defaults in the app and website seem to be 2018. It seems like a simple request but something must be hard, likely the different weather models available in different countries.
