Maximum temperatures in the southwest and the BBC

Every year when the summer starts in earnest and weather presenters start talking about how warm it is (particularly in the southeast), they start presenting charts with temperatures across the southwest of England several degrees lower than the rest of the country, and every year I can’t help writing a protest blog like this one! Today was a perfect example, take a look at yesterday’s forecast temperatures for 16 PM (15 UTC) today as presented by Jay Wynne on BBC 1 at 1.25 PM.

Figure 1 – Courtesy of the BBC

Now have a look at today’s actual values from across the country (fig 2).

Figure 2

As you can see it’s actually 20.8°C in Plymouth, 21.1°C at Culdrose and 23.0°C at Cardinham, so the forecast for Plymouth is too low, not only that, where the 19 label is drawn on the BBC graphics right on top of Exeter, where the temperature at 4 PM today was a hot 27.1°C, that’s a full 8°C higher than on the chart. The forecast temperature for Plymouth looked wrong yesterday, and it was wrong in reality today. For goodness sake, why don’t the BBC simply choose a city that’s more representative of the southwest, or perhaps display the maximum forecast temperature for the whole region.

  • I did a quick search and I can see I’ve already had a go at the BBC this year back in April.
  • It looks like its not only maximum temperatures that get me going.
  • And here is last years effort.

The one thing you can’t accuse me of is, and that’s a lack of consistency when it comes to temperature forecasts from the BBC.

A little bit too optimistic

Figure 1 – Courtesy of the BBC

There was a little bit more rain than expected in Spotlight forecast at 6.55 PM yesterday (fig 1), and the front was a lot slower and more reluctant to clear Cornwall, with some kind of secondary feature following on behind the initial band of rain (fig 2).

Figure 2

Now that I’ve seen the midnight analysis (fig 3), I now realise that the secondary feature is the cold front, and that the initial rain band is the warm front.

Figure 3 – Courtesy of the Met Office

Here’s a quick look back at yesterday’s estimated rainfall totals in the southwest (fig 4), most of it falling before 15 UTC if truth be said. It finally did brighten in the late evening, as the upper cloud finally began to retreat eastward.

Figure 4 – Courtesy of the Met Office

Dartmoor triggers intense shower

Data courtesy of the Met Office

In the last couple of hours, the high ground of east Dartmoor (possibly around Haytor) seems to have triggered a thundery shower that quickly tracked NNE across the west of the city of Exeter. It looks to have been quite an intense, but short sharp shower. I say thundery, but so far Blitzortung has only detected one flash of lightning from it.

What seems very silly to me, watching the latest forecast on the BBC News channel (which looks like it may have been taped), is how the weather presenters prefer to use NWP model rainfall rather than real-time weather radar images. In a dynamic convective situation like today, you would have thought that’s what they would be most helpful to the public. Likewise the BBC presenters never ever use real-time SFERIC data, or come to that rapid scan 5 minute visible satellite imagery. What’s the point of having all these various ways of observing the weather in real-time, and then not using them?

Certainly use NWP images for forecasting, but masquerading NWP forecast data as real-time observational information is wrong and just plain misleading.

I remember coming home from work on the 16th of August 2004, it must have been just before 4 PM, to watch Michael Fish prattle on in the forecast about how hot it had been recently in Greece. This too was a taped forecast that they used on the same BBC News channel. The news was later interrupted by breaking news about flash flooding that was occurring at that precise moment in Boscastle. It seems to me that little has changed since then in how we present the weather, if this intense shower that I noticed this afternoon had been slow-moving and a little larger, the same thing could have easily happened all over again.

WeatherscapeXT from MetraWeather

Figure 1 – Courtesy of MetraWeather

I wrote something a few weeks ago saying that the MeteoGroup takeover of the BBC contract was imminent. I wondered about what the graphics would look like but never thought to look at who MeteoGroup have teamed up with to supply the graphics. It’s a New Zealand concern called MetraWeather, who have as far as I know, been responsible for supplying the BBC with the current system for the last 10 years or so. Looking on their website they are all excited about something that they call Weatherscape XT black edition™ (fig 1). If the screenshot their promotional video is anything to go by it looks to be a step up from the existing graphics software in use at the BBC, isobars seem to be light blue on a dark blue sea, but the colours I’m sure are totally configurable. This is what they say about it courtesy of their website:-

  • Compelling graphics
    Weatherscape XT’s powerful graphics terrain engine and weather animation effects deliver beautiful rendered; full-screen graphics in real-time
  • Complete customization
    The powerful design tools integrated into Weatherscape XT enable your graphic designers to create unique weather presentations that reflect your programming style, and compatibility to both your brand guidelines and those of your sponsors and advertisers
  • High-quality maps
    Weatherscape offers designers both stylized and realistic location-specific maps at standard resolutions of between 15 to 300 metres ensuring crystal-clear, accurate rendering. The big question is – can they fix the projection that’s been shrinking Scotland for the last 10 years?
  • Rapid data updating
    At the core of Weatherscape XT is the capability to rapidly receipt and update data inputs from radar, gridded model data, satellite imagery, lighting data, local observations, typhoon tracks, and warnings and alerts provided by non-meteorological data providers warning of events such as tsunami, forest fires and volcanic activity. By the way it’s lightning and not lighting – I kid you not!
  • Powerful data editing
    Weatherscape XT is uniquely configured to enable broadcasters to edit and interpret meteorological data such as gridded model fields, observations, forecasts and warnings, or to create custom forecasts for special events including elections, sports tournaments and live performances. 
  • Data management
    MetraWeather customer engineers have vast experience in configuring Weatherscape XT to automatically and continuously ingest and manage data provided from both global and regional meteorological data suppliers 
  • Content for all screens
    Weatherscape XT is easily configured to automatically output weather graphics and content for all broadcast and online channels including highly successful weather presentation solutions for proprietary streaming media, kiosks, and connected mobile and tablet devices.

MetraWeather is of course a subsidiary of MetService, the now privatised New Zealand Met Office, and this weather graphics system was probably built by their own team of programmers. I’m guessing (and I could be wrong), that the Met Office didn’t do that with their new weather graphics engine VisualEyes™, they always seem to prefer playing it safe, and teaming up with an outside concern to do the development, rather than doing it themselves in-house, which is a great pity, especially for the IT staff. None of this kind of development is rocket science and hasn’t been done before, it just needs someone with a clear vision of what they want to achieve, and of course a team of talented developers to get it done.

Will the BBC graphics look like this?

One of my readers after reading my article about NWP Web viewers has kindly pointed me to one that the MeteoGroup are trialling at the moment. It’s called MeteoEarth and it’s very ‘Google Earth’ like with a spinning globe, although you can display ‘flat’ maps. There are a number of basic overlays available, but they are just that quite basic, and the contoured pressure overlay does need further work to look right. It might be early days for their web application, because that’s what it really is, perhaps this is some kind of test bed. They seem to be encouraging people to embed MeteoEarth into their own websites, and you can also hook up to social media through various buttons. It makes me wonder if it resembles in any way the graphics engine that they are planning to use when they take over the BBC contract very shortly.

Figure 1 – Courtesy of MeteoGroup

The info box makes interesting reading as well…

Figure 2 – Courtesy of MeteoGroup

They are making use of global ECMWF data in the MeteoEarth, wouldn’t that be a turn up if they did the same thing at the BBC! I can’t see this happening somehow, especially the political ramifications of using European NWP data in preference to Met Office NWP data when, we are just about to begin the Brexit negotiations.

BBC weather takeover imminent

The takeover of the BBC weather contract by MeteoGroup must be imminent if it’s to happen this spring as promised. I’ve read recently that Meteogroup are going to provide the graphics for the new service, and I wonder if they are of the same standard that they push out on their Twitter account. It seems that the Met Office are using all forms of social media including Twitter to show of their new graphics engine – Visual Cortex. If these couple of examples I have included from their respective Twitter accounts are anything to go by (figs 1 & 2), I think I much prefer the graphics and animations from the Met Office than those of MeteoGroup, which do look a little dated. Corporately, I think that the Met Office are still smarting after being dropped by the BBC, it’s quite obvious from how they’ve upped their game in the last couple of years on social media.

Figure 1 – Courtesy of MeteoGroup

Figure 2 – Courtesy of the Met Office

Of course MeteoGroup could surprise us, and have their own new bespoke graphics engine ready to go from the start, or maybe getting that in place and tweaked has delayed the takeover. It seems that all the various staffing changes concerning the presenters in London have now taken place, and all we are waiting for now is the big switchover.

Eleven of the best NWP viewers on the Internet

If you, like me, are fascinated by what the weather is going to do in a bit finer detail than you’ll ever get from watching the weather on the BBC, then you might be interested in some of the eleven websites that I’ve listed below. They all provide a quick and easy way of looking at NWP forecast data from various models, usually the American GFS, but not always. There have recently been a rash of new websites providing these kind of viewers, and they are increasingly becoming more sophisticated and professional. I’ve tried to put them into some kind of ranking, let me know if you agree, or maybe you know of a website that I’ve missed out completely.

#11 –

A basic GFS viewer from James McGregor in New Zealand, the graphics and maps used are simple and clear once you find them, as are the controls and available map regions, all set in a web page that offers visualisation such as satellite, weather radar and observational data.

Figure 1

#10 –

The Norwegian Meteorological Institute and the Norwegian Broadcasting Organisation have put this NWP viewer together between them. It’s rather Norwegian centric, rather like the Met Office website is with the UK, and as far as I know it’s their own NWP model that they use. Personally I just like the Meteograms, which in my opinion would be even better if they ran out to T+120 or longer.

Figure 2 – Norwegian Meteorological Institute and the Norwegian Broadcasting Corporation. © 2013 – 2017

Figure 3 – Norwegian Meteorological Institute and the Norwegian Broadcasting Corporation. © 2013 – 2017

#9 –

The Wetterzentrale site has been around for years and was recently was re-sited and the NWP viewer spruced up. I like the basic idea of the graphics that they produce, and the overlay of an MSLP field on top on a colour filled contoured chart of geopotential height seems to be a defacto standard. I also like the ability to access other models from around the world, especially the 20th century reanalysis data that the site provides as well. I think the controls are still a little bit basic and could do with a bit of a makeover.

Figure 4

#8 –

The OGIMET site is an old favourite of mine, and is were I go to download SYNOP data, and even though the guy that runs it has never replied to a single one of my emails over the years, it still displays a comprehensive array of GFS model data for various regions. I like the multi-layered graphics which are clean and simple, the functionality and controls are rather basic though.

Figure 5

#7 –

The weather outlook viewer controls are just a little bit more sophisticated than those on the wetterzentrale site, but the graphics are very similar. The controls make selection a little quicker, with flyover selection of model time steps by means of the mouse, with a good range of NWP data fields to access and display.

Figure 6

#6 –

The netweather site have a similar approach to that of weather outlook, but if anything the user interface is clearer and more organised. Again the NWP graphics are not that dissimilar to those produced by wetterzentrale and superbly clear.

Figure 7

#5 –

This site keeps a special eye on tropical cyclones, but that doesn’t stop them enabling their NWP viewer to be pointed at other regions from a variety of NWP models. The graphics offer multiple layers of NWP types which are sharp and clear, and you can view some quite sophisticated overlays on detailed outline maps.

Figure 8

#4 –

The earth viewer from Cameron Beccario has been around for a long while now, but has exquisite animated graphics. The controls are hidden away so as to get the full effect of the viewer when running in full screen mode. What else can you say other than it’s just beautiful.

Figure 9 – © 2017 Cameron Beccario

#3 –

The windytv viewer goes a little bit further than the earth viewer, and gives you an always visible interface, that’s not obtrusive and very tastefully designed, I bet behind the facade it probably uses the same vector graphics and mapping though. A really swish NWP viewer which I don’t use often enough.

Figure 10

#2 –

The ventusky viewer is from the same stable as the earth and windytv viewers by the looks of it. I quite like it, even with the slightly gaudy colours that they’ve chosen to use on their wind fields.

Figure 11

#1 –

The wxcharts viewer does just about everything that you could want of a NWP viewer in a web browser, and it must be the equal of many desktop weather visualization systems. It’s hard to believe that the developers have packed so much into the display, in which they’ve also managed to include meteograms as well as maps. You can tell that the team that’s put this together are real enthusiasts. The NWP graphics might be static, and not just quite as impressive as the animated maps from ventusky or windytv viewer, but I still think wxcharts as the edge on them in my opinion.

Figure 12

National Met service offerings in comparison

The DWD in Germany just like to keep it simple, why complicated things with those funny isolines and colour filled contours.

Figure 14 – Meteo France

Sophisticated stuff from the French, in this screenshot from the Meteo France website. I suppose for the general public this viewer is telling them all that they need to know – generally sunny.

Figure 15 – The UK Met Office

The Met Office are still using the same FAX charts that they’ve been using for the last 50 years or more, they analysis chart still has the same PPVA89 file designator that it’s always had, but at least nowadays it’s in colour. The Met Office have collaborated to produce the Visual Cortex graphics engine which they use on ITV and internally these days. They seem to have placed most of their effort into producing weather forecast videos which they use on their website and social media. It’s a shame that Visual Cortex can’t be used to provide NWP graphical output to a web viewer in a similar way to what ventusky have done, how hard can it be.

At least the Americans are now trying to update their viewers, and have recently started to use GIS web services offered by ESRI I notice.

It’s all very sad when you compare the quality and sophistication in the first eleven NWP viewers with those on offer from some of the worlds National Met Services. The difference is quite stark in most cases, it’s as if these organisations don’t even want to compete, after spending countless millions on the fastest supercomputers, and building the most sophisticated ensemble and deterministic NWP models, they just tire at the last hurdle and leave it to others to provide a web interface for the citizens of their respective countries to use, the question that springs to my mind is, why not?

Look back at Sunday’s maximum temperatures

I thought that I would look back at yesterday’s (9th April 2017) maximum temperatures forecast by the BBC. This is not a moan about the temperature contrast across the south of the country, their most certainly was a good contrast, it’s more to do with the sites that the BBC choose as representative of their region and label in their graphics. The one that infuriates me the most is Plymouth in the southwest. There is a method in their madness though, because the Met Office bonus is dependent on how these extreme temperatures are scored in verification, choosing a coastal site (which is difficult not to do on an Island) can pay dividends, because the extremes, and therefore any potential misses are not as large. I believe that yesterday’s maximum temperature of 20.0°C at Exeter was more representative for the whole of Cornwall, Devon, Somerset and Dorset, than a forecast maximum of 14°C was for Plymouth, and even here the temperature reached 16.8°C. Rather oddly many of the temperatures on the chart don’t reflect the colour filled temperature contours that they are overlaid on, for example, a forecast maximum of just 16°C at Birmingham was always going to be wrong but the background colour is a mid-orange colour.

I think that I would have the full support of the South West Tourist board if the BBC put a little more thought into how they forecast regional temperatures.

Figure 1 – Courtesy of the BBC

I couldn’t help myself and resist scoring the maximum temperatures on the chart, if anything it underlines the fact that the Met Office model seems to have as many problems underestimating maximum temperatures as it does overestimating low cloud at the moment.

Figure 2

Where would you go to catch the sun?

Figure 1 – Courtesy of the BBC

Where would you go for a day out and catch the sun? It’s a simple enough question, but I bet the majority of the people who saw this chart wouldn’t say the southwest of England.

I don’t know why this surprises me because I moan about it every time spring and summer come around. The BBC presenters in past years have done irreparable damage to the tourist industry of the southwest with misleading charts like this. It doesn’t matter how many times that you complain that a maximum temperature for Plymouth Hoe is just not representative for the whole of the southwest peninsula (especially in a SE’ly) it has absolutely no effect in the slightest.

To be fair weather presenters must feel a little left out and neglected once the named storm season and all that gesticulating and arm waving that goes with it have finished, so they return to their first love, extreme temperatures, and if they are occurring in the southeast of England so much the better!

Why is forecasting low cloud so difficult?

I had ago at David Braine this week for his cloudier day forecast for Thursday. It wasn’t anything personal, I think he’s a pretty good forecaster turned weather presenter, he like a lot of other presenters was only towing the party and following the model, it was all to do with the output from the Met Office mesoscale NWP model. It seems to have a problem with forecasting low cloud amounts, especially in anticyclonic situations when it seems to overestimate the amount. They know about the problem, and according to their new blog Why is forecasting low cloud so difficult? it has to do with thin SC layers being missed by the coarseness in the vertical resolution of their model (100 M), combined with the difficulty in assessing SC amounts over the sea, especially when it’s masked by higher cirrus layers in satellite imagery.

I think there’s also another problem that they are missing, and that it’s either not possible, or difficult to tweak the graphics engine used to produce the graphics on TV. So the forecaster may know full well that the model is wrong or misleading from the evidence of the latest observations and satellite imagery, but they can’t quickly adjust the graphics à la Photoshop. I use Krita which is a free drawing package, and it comes with dozens of different pens and tools to draw and edit with, it surely wouldn’t be difficult to configure a pen that dissolves cloud layers?

This problem won’t necessarily go away when MeteoGroup take over at the BBC, but that’s of course dependent on whether the BBC stipulate that they use Met Office model data. Perhaps the new weather graphics engine that the Met Office now use internally and at the ITV, Visual Cortex , does make it possible to edit NWP forecast frames, if it doesn’t it should.