Ever since "the future" was added to the radar slider, the regular "past" radar has degraded in quality. While it's playing, the radar data "wobbles", sliding by tiny amounts randomly over the map underneath.

Any chance of a fix for this? It's quite annoying.

As an aside: Please add an option to remove the "future" from the slider, I live on the US west coast, right on the edge of the radar coverage zone, so it can't see new rain coming in anyway, making the future prediction fairly useless.

I understand that forecasting the weather is difficult, but the future radar forecast is completely inaccurate.

Video example:

In this example, the rain system suddenly starting moving rapidly to the east in the forecast, even though it showed no sign of doing so in the past data.

If the forecast is this inaccurate, it should not be made available.

Can we at least have an option to disable it?

It is likely that the Trump/DOGE cuts are going to affect the rollout and quality of new NOAA/NWS models.

The RRFS data was moved off Amazon into NOAA infrastructure, where the team is much more resource constrained and the amount of data available has been reduced. The operational date also slipped repeatedly (from "early 2026" to August 31 to October 6, 2026), though the stated reasons were evaluation results and weather-delay contingencies rather than budget. On the research side, the National Severe Storms Laboratory, Warn-on-Forecast, and other research-to-operations efforts remain on the chopping block in the current request, and these feed the long-term RRFSv2/MPAS roadmap.

No model has been cancelled or clearly delayed because of the cuts, but the labs producing them have been under a standing threat of closure for two budget cycles, have absorbed workforce losses, and are running the RRFS transition leaner than planned. Whether the FY2027 fight ends the same way as FY2026 is still open.

When you're viewing a weather forecast of a local model, for example the HRRR in North America, when the limit of the forecast is reached it shows the GFS after it.

Would it be possible to add an option to a different local model instead of the GFS in the forecast? For example, after the HRRR, it would make more sense to show the NAM model for longer range forecasts.

@williamcorlin This really depends where you live. The ECMWF is the best global model, and the best default fallback when you have no better local model.

In the USA the HRRR is more accurate, but only predicts the next 48 hours.

In the UK the Met Office "UKM" model is generally the most accurate.

Other locations may have their own local high-resolution models.

There's a new "Meteoblue AI" forecast that isn't really a model but it is some kind of forecast-only thing. I'd like to see more accuracy data for that.

To be clear, I'm not talking about the future radar forecast here. (Although I did complain about that in another thread.)

I'm talking about the regular "past" radar. It never used to wobble around like it is now, and can be seen in the above video.

Windy just sent me an alert that says "Hurricane category 2 Marie expected to arrive in 5 days". I'm in the San Francisco Bay Area.

Isn't this really misleading? It implies that a Cat 2 hurricane will arrive in 5 days, but what it actually seems to mean is that the hurricane is Cat 2 now. Marie is currently forecast to stop being a hurricane long before it reaches land.

This alert is needlessly alarmist.

@Wheats The alert says a Cat 2 hurricane will arrive. Even if the forecast is accurate, that is not correct. A storm system will arrive, not a hurricane of any category.

It is likely that the Trump/DOGE cuts are going to affect the rollout and quality of new NOAA/NWS models.

The RRFS data was moved off Amazon into NOAA infrastructure, where the team is much more resource constrained and the amount of data available has been reduced. The operational date also slipped repeatedly (from "early 2026" to August 31 to October 6, 2026), though the stated reasons were evaluation results and weather-delay contingencies rather than budget. On the research side, the National Severe Storms Laboratory, Warn-on-Forecast, and other research-to-operations efforts remain on the chopping block in the current request, and these feed the long-term RRFSv2/MPAS roadmap.

No model has been cancelled or clearly delayed because of the cuts, but the labs producing them have been under a standing threat of closure for two budget cycles, have absorbed workforce losses, and are running the RRFS transition leaner than planned. Whether the FY2027 fight ends the same way as FY2026 is still open.

@williamcorlin This really depends where you live. The ECMWF is the best global model, and the best default fallback when you have no better local model.

In the USA the HRRR is more accurate, but only predicts the next 48 hours.

In the UK the Met Office "UKM" model is generally the most accurate.

Other locations may have their own local high-resolution models.

There's a new "Meteoblue AI" forecast that isn't really a model but it is some kind of forecast-only thing. I'd like to see more accuracy data for that.

@idefix37 The UK Met Office model is known to be weaker than the ECMWF for hurricanes, which is pretty much the gold standard at the moment.

The "NHC" source isn't a model, it's created by hand at the NOAA Hurricane Center.

Regardless, none of the sources predict a dangerous storm hitting the bay area, and certainly not a "category 2" hurricane.

The alert is extremely misleading.

Windy just sent me an alert that says "Hurricane category 2 Marie expected to arrive in 5 days". I'm in the San Francisco Bay Area.

Isn't this really misleading? It implies that a Cat 2 hurricane will arrive in 5 days, but what it actually seems to mean is that the hurricane is Cat 2 now. Marie is currently forecast to stop being a hurricane long before it reaches land.

This alert is needlessly alarmist.

This is still happening. The regular past radar (not the future forecast) is wobbling and sliding around the map.

To be clear, I'm not talking about the future radar forecast here. (Although I did complain about that in another thread.)

I'm talking about the regular "past" radar. It never used to wobble around like it is now, and can be seen in the above video.

I understand that forecasting the weather is difficult, but the future radar forecast is completely inaccurate.

Video example:

In this example, the rain system suddenly starting moving rapidly to the east in the forecast, even though it showed no sign of doing so in the past data.

If the forecast is this inaccurate, it should not be made available.

Can we at least have an option to disable it?

Here's a screen recording of the problem.

You can see the radar image wiggling back and forth.

Ever since "the future" was added to the radar slider, the regular "past" radar has degraded in quality. While it's playing, the radar data "wobbles", sliding by tiny amounts randomly over the map underneath.

Any chance of a fix for this? It's quite annoying.

As an aside: Please add an option to remove the "future" from the slider, I live on the US west coast, right on the edge of the radar coverage zone, so it can't see new rain coming in anyway, making the future prediction fairly useless.