[Rasp] RASP Beta Inaccuracy vs old RASP 1.3 and 4km Res models

Sun Jul 13 21:27:13 SAST 2025

Hi Ria, Ian,

Nice to hear back from you and all makes sense.

One thing I forgot to mention (if not come across before) is that due to the way numerical forecast grids work, the actual forecast for a given spot location on the grid may not be anywhere near the spot on the ground for where we want it. 

At 1km resolution the grid corner could be 300-400m away from the top/ridge which for a ridge or mountain top can be lower, or different. Winds on one side of a ridge within 300-400m can be completely different. The bigger the grid the larger the chance a desired spot is not forecast.

So modelling grids for peaks/ridges is tricky. Due to their narrow nature, they may be missed by both the grid and the underlying terrain model RASP uses (smoothed SRTM 10m terrain data).

While the UK does not have the terrain range of southern Africa, the Wales model at 1km is ok across the region for wave, but spot forecasts for peaks/ridges are difficult to pick out. See the attached pic as an example showing the tops not on a grid point.

Fixed wing pilots are more ok with forecasts across an area as they fly longer and further. Whereas paraglider pilots are more keen to know about a spot location. I had a similar conversation with UK paraglider pilots in Scotland and Wales on this topic when they said the forecast was not fully spot-on the ridge they fly from.

See how you get on and stay in touch 🙂

Regards,

Darren

-------------- next part --------------
A non-text attachment was scrubbed...
Name: iu.jpeg
Type: image/jpeg
Size: 34420 bytes
Desc: not available
URL: <http://lists.zsd.co.za/pipermail/rasp/attachments/20250713/035fc603/attachment.jpeg>
-------------- next part --------------

> On 13 Jul 2025, at 11:33, Ria Moothilal <ria at airschool.co.za> wrote:
> 
> Hi Ian and Darren
> 
> Thanks for the replies and feedback - much appreciated.
> 
> Darren - I was unaware of the multiple surface wind forecasts with
> RASPBeta, so I'm now making sure to compare the 10m surface wind
> forecasts for the different RASP resolutions. In terms of weather
> forecasting strategy across various platforms, I'm with you on the
> multiple platform consensus resulting in a higher confidence for
> decision making - RASP was one of the models that usually gets it
> right even when others are wrong, and this is specific to the surface
> wind remodelling around the mountains which is why we're so passionate
> about it. Comparison to Metars will not be very helpful, as the Metar
> originates from Cape Town International, whose location is mostly
> unaffected by the effects of our mountains - most platforms/models
> will get the forecast for the CT International location correct. Our
> paragliding takeoffs are on the mountains. I can understand how a
> higher resolution model performs well in the UK, with its lack of
> mountains :-)
> 
> Ian - thanks for pointing out the Zulu time issue. I'm now making sure
> to pull the Zulu time from the data file headers and use this to set
> the local time before making any comparisons. This has made a
> difference to the results.
> 
> Model Comparisons and Actuals:
> I've set up a script to fetch and compare the wind speeds and
> directions for the 1.3, 2, and 4km res outputs for a single point on
> the west side of Signal Hill. This, together with a live weather
> station reading, should be helpful for a less biased comparison. It's
> accessible here:
> https://docs.google.com/spreadsheets/d/1ItHNxkOPhUQeHtgA4dx1q-J1WRzp5GWSknVMpxgKaFo/edit?usp=sharing
> 
> I'll look into adding a few more locations where we have live weather
> station data, e.g. Signal Hill takeoff, Hermanus, Porterville, and Du
> Toitskloof
> 
> Many thanks
> Ria Moothilal
> Chief Flight Instructor
> +27 78 095 8136
> www.airschool.co.za
>> On Sun, Jul 6, 2025 at 11:09 AM rasp.admin at stratus.org.uk
>> <rasp.admin at stratus.org.uk> wrote:
>> 
>> Hi Ria,
>> 
>> Many thanks for taking the time to look at data quality.
>> 
>> Based on looking at the images below, the old plots are at 10m/30ft and the new plots are at 2m/6ft. That could explain a difference.
>> 
>> To do a fair comparison (beyond a single location), I’d suggest using METARs across a wider area. I have done this for the UK and find ‘on average’ the forecasts are acceptable based on feedback. Note that is not perfect nor correct.
>> 
>> Then there is the comparison over time. The guess is that is a regular problem, as a  single day for a single point is probably too small a sample to conclude one way or another. Even government-funded forecasts do not guarantee it’ll be correct every time, all the time, at all locations. Here’s the UK quality comparison URL:
>> 
>> https://rasp.stratus.org.uk/index.php/daily-data-quality-checker
>> 
>> I have often advised to use RASP output with at lest two other sources to confirm concurrence. If they are all the same, then you can probably trust RASP. If not, then it’s up to the pilot to use experience on what is going on and likely to happen.
>> 
>> On reusing the old models, there would be a significant technical redevelopment effort required to retro fit. There is also the problem of software support as the 2.X WRF binaries and NCL 6.X plotting subsystems are no longer supported. Python-based plotting is now the supported path, but needs someone to rewrite the plotting core of all this to make that work. It’s not impossible, but needs someone to volunteer and do.
>> 
>> Regards,
>> 
>> Darren
>> 
>> On 6 Jul 2025, at 08:22, Ria Moothilal <ria at airschool.co.za> wrote:
>> 
>> 
>> Hi
>> 
>> I'm giving feedback as requested. Today, the RASP Beta was wrong again, and interestingly, the old RASP 4km model was more accurate.  Below are the forecasts for today, the 5th July at 11 am, as well as the actuals for Signal Hill taken by 3 live weather stations. Both the 1.3km and 4km old RASP models show that the north-west wind slows down before it reaches Signal Hill / Table Mountain. The RASP Beta 2km does not. This suggests that it's not just a resolution difference on the RASP Beta, but also either a source data or model parameter change that has taken place. The old RASP was accurate when compared to actuals, but the RASP Beta was completely wrong
>> 
>> The old RASP was the most accurate weather forecast for the Western Cape. Nothing came close. I think it would be sensible to merely recreate these same models on the new user interface
>> 
>> 1. RASP Beta 2km
>> <image.png>
>> 
>> 2. Old RASP 4km
>> <image.png>
>> 
>> 3. Old RASP 1.3km
>> <image.png>
>> 
>> 4. Actuals from weather stations at Signal Hill and Sea Point:
>> <image.png>
>> 
>> 
>> Ria Moothilal
>> Chief Flight Instructor
>> +27 78 095 8136
>> www.airschool.co.za
>> --
>> Rasp mailing list
>> Rasp at lists.zsd.co.za
>> https://lists.zsd.co.za/cgi-bin/mailman/listinfo/rasp