REX WEATHER FORCE For MSFS

[Bottle 663]

21 minutes ago, rstough said:

Metar is real data reported from an observer.

Are you saying Meteoblue doesn't have access to METAR data? If so, can you evidence that? It would seem very odd if they didn't have access to it considering what their business is.

[versus 61]

2 minutes ago, rstough said:

We hope they do fix it so it is no skin off our back.  We built WF because we saw that there was definitely some issues.  In addition, we also wanted to provide archive as well as custom dynamic weather scenarios.

I believe that what Asobo has done is amazing, but there is definitely a difference between metar and model data.  Model data is based upon a grid which covers the world.  Most global models cannot get to that level.  There are some local models that can, but again they are at best forecasts.  Metar is real data reported from an observer.  So WF is built from metar data.  If  it says a thunderstorm you will get a thunderstorm.

Finally we got rid of the "human observation"  weather depiction which creates many different weather bubbles (metar stations)  with fake interpolation between them. This approach is outdated as most of the weather (except the one at the airport) is fake. Moreover human observation has very limited parameters compared to model depiction. Metar is generally meant for humans to interpret not for computers to compile. Don't see how can you improve weather  depiction with this. 

[rstough 531]

Just now, versus said:

Finally we got rid of the "human observation"  weather depiction which creates many different weather bubbles (metar stations)  with fake interpolation between them. This approach is outdated as most of the weather (except the one at the airport) is fake. Moreover human observation has very limited parameters compared to model depiction. Metar is generally meant for humans to interpret not for computers to compile. Don't see how can you improve weather  depiction with this. 

Human observation is better than model's.  Models rely upon the observation to make their forecasts.  Metar data is the foundation, but model data cannot replace it.  The best computer models cannot replace the metar report.

[rstough 531]

3 minutes ago, Bottle said:

Are you saying Meteoblue doesn't have access to METAR data? If so, can you evidence that? It would seem very odd if they didn't have access to it considering what their business is.

I am NOT saying that at all. I am going off of what I have tested numerous times and compared to models that I have access to.

[rick celik 1,251]

5 minutes ago, rstough said:

Human observation is better than model's.  Models rely upon the observation to make their forecasts.  Metar data is the foundation, but model data cannot replace it.  The best computer models cannot replace the metar report.

I agree. Personally, i think this is great news. 

Edited September 16, 2020 by Ridvan Celik
quote

[omarsmak30 4,068]

3 minutes ago, rstough said:

Human observation is better than model's.  Models rely upon the observation to make their forecasts.  Metar data is the foundation, but model data cannot replace it.  The best computer models cannot replace the metar report.

Sure but correct me if I am wrong, isn't the model gets improved over time with more data fed in? I mean, normally ML models need time and data in order for the prediction data to be improved. Sure, maybe now the model has a lot of noise data or the prediction model is not that good but as with every ML model, if they are actively improving the model, this will eventually improves in the future, isn't? 

[jcomm 6,009]

IMO, there's surely room for 3pd contributions to weather in MFS.

ASOBO should fix the most screaming limitations / bugs, but still, talented 3pds can do a bit more of Magic in, for instance, find the best possible merge between 850+ Hpa model data and observations...

 

[rstough 531]

3 minutes ago, omarsmak30 said:

Sure but correct me if I am wrong, isn't the model gets improved over time with more data fed in? I mean, normally ML models need time and data in order for the prediction data to be improved. Sure, maybe now the model has a lot of noise data or the prediction model is not that good but as with every ML model, if they are actively improving the model, this will eventually improves in the future, isn't? 

No.  Give you an example.  May 20 2019, all the models painted one of the worse severe weather days for Oklahoma. Models were off the charts with the potential.  Every storm chaser was flooding to OK.  Tornadoes did occur, but NOT even close to the model data suggested. What the models didn't see was the storms over Kansas causing an outflow boundary moving south cutting off the rich supply of moisture and instability for the storms.  It was basically a bust.  The only way to have known that was going to happen was metar data.  Observations showing a shift in winds and significant drops in temperatures.  

Edited September 16, 2020 by rstough

[wim123 765]

great news, i would like to see the weather as precise as is possible, besides historic weather, i would like to see also

other different dynamic scenarios.  hope its a oneprice thing, wont do a subsription.

[AdiR 27]

Reed, any chance for SIGMET implementation? Cheers!

[rstough 531]

3 minutes ago, wim123 said:

great news, i would like to see the weather as precise as is possible, besides historic weather, i would like to see also

other different dynamic scenarios.  hope its a oneprice thing, wont do a subsription.

No subscription.  The initial release includes Live weather and Dynamic Weather Scenarios.  The next update will include archive/historic weather.  The third major update will include community weather presets.  Where you can control the weather elements individually like in Environment Force.

[rstough 531]

3 minutes ago, AdiR said:

Reed, any chance for SIGMET implementation? Cheers!

We will be including a mapping system later in development.  We didn't include it now because we are looking into a better system than what we had in Sky Force.  It would be our hopes we can add Sigmet data to this. If that is what you are asking about.

[mjw25 42]

Reed, good luck, I hope things work out for everyone.....

[rstough 531]

Just now, mjw25 said:

Reed, good luck, I hope things work out for everyone.....

Thank you.  

[rayharris108 259]

Reed makes good points. METARS, as well as being human observations and up to date and accurate, are also the funadamental data real work pilots depend on. Surely you can't get better than that? I am sure in conjunction with the global models it will be worth it to have the REX add on. After all if you remember, REX Environment Pro was one of the most ground breaking addons for P3D.

I know MSFS is different of course, but one thing lacking as already mentioned earlier, is high-altitude cloud modelling above FL200. MSFS is basically using 2/3? layers of variations on cumiliform clouds at all levels right now albeit in amazing detail and light. 

I am waiting to see feathery alto-cirrus, cirro-stratus, alto-cumulus formations and alto stratus. Also really good anvil thunder heads. Env Pro gave us that in 2D in P3D can we get it in 3D in MSFS?

That's the question for me.

Ray