Glidemet

Some time ago—-might be last year now—-I asked people to write in with the wind speeds and directions in which their local ridges worked. (Yes I’m already thinking about winter forecasting—-well winter never really ended, did it?)

This is my list so far:

• Booker: 310-340, 15 knots
• Challock: 215-245, 15 knots
• Cross Fell (edensoaring): 180-300, 15 knots
• Denbigh: 230-300, 12 knots
• Dunstable: 200-310, 12 knots
• Edgehill: 270-360, 15 knots
• Keevil: 300-020, 12 knots
• Kenley: 185-215, 15 knots
• Long Mynd: 250-310, 15 knots
• North Hill: 270-340, 20 knots; 170-190, 15 knots
• Nympsfield: long-distance 250-330, local 270-030, both 15 knots
• Pocklington: 260-320, 10 knots; and 240-290, 13 knots
• Quantocks: 230-250, 15 knots
• Rivar Hill: 340-010, 15 knots
• Shobdon: local ridges (Shobdon and Wapley hills) 315-015; Wenlock Edge 290-350; Old Radnor to Telford SW 315-350. All 15 knots.
• South coast cliffs(!): 150-180, 20 knots
• South Downs: 340-040, 15 knots
• Sutton Bank: south 170-230, 17 knots; southwest 230-260, 12 knots; west 260-300, 15 knots
• Talgarth: Main Ridge 250-360 (8 knots), Cwm Du Valley 180-240 (10 kts), Hay Bluff 000-070 (10 kts), Pandy run 030-080 (15 kts), distance flights along Brecon Beacons 300-040 (15 kts)

Anything I’ve missed?

There’s a reason I’m putting everything in the format of nnn-nnn, nn knots (with some interpretation of what people have told me). We’re hoping to generate automatic ridge forecasts from RASP data, but to do that we need fixed ranges of directions and speed thresholds for the system to process. Should be a very useful tool if we can get it to work . So if you disagree with any of the numbers I’ve used, do tell me!

If you go down to the UK RASP website today you’re in for a big surprise. Paul Scorer (nephew of the “father” of wave science, R.S. Scorer) has put the upgraded RASP system live. I’ll take a few moments to talk through the changes Paul has made, but first I want to publically say thanks to Paul for all the work he does for us. Paul has taken RASP and pushed way beyond what the original developer, Dr Jack Glendening, envisaged. Wrangling cutting-edge meteorological models, ones exactly the same as used by the National Weather Service in the US, ain’t easy!

The changes
The most notable change is to the forecasts you see in the morning, the so-called “on the day” run. This was originally 12 km for the UK and 4 km for an area of eastern Scotland and northern England. Well now it’s 4 km for the whole of the UK! It’s worth noting that this equals the resolution of the Met Office’s own mesocale model, so it really is something.

The 4 km has been shown to forecast areas of wave activity well (John Williams used it to know that his famous “North Sea 1” turnpoint was reachable, which demonstrates the faith he has in the system). The resolution is not sufficient to forecast individual wave bars, but what you will see on the vertical velocity at 850, 700, and 500mb charts are what look like slightly fat “wave bars”. The wavelength of these plotted bars is much larger (2-3x usually) than what you’ll find in the air, but what they WILL show you is the position, extent, and timing of wave systems. For the first time, Wales and all of northern England will have accurate wave forecasts.

This 4 km grid takes some 9 hours to compute, even with custom-compiled (iFort) executables running on a 4 GiB Q6600 Quad Core 2. If we started this run with 00z data you wouldn’t see the charts until much too late, so instead we’re initialising the run from 18z “day before” data. This is a calculated exchange between resolution and accuracy of data: even though the initialisation data is six hours older, the 3x increase in the resolution of the model’s grid more than compensates. In the US an identical model is run for severe weather (tornadoes etc.) forecasting and the scientists concerned have found that even though they have to start the model twelve hours earlier than their 12 km one, their 4 km run is far more useful for forecasting.

As it stands, you’ll be seeing the output from our 4 km model at about 4am local. This was actually a request from John Williams so he could plan the longest flights possible using all the daylight he can!

You’ll also see an upgrade in the “next day” charts you see at around 7pm in the evening onwards. The resolution of these charts-for-tomorrow has been pushed to 5.1 km, starting from 12z data. The slightly larger resolution means the run only takes about three hours. We haven’t had a wave event during the time that the test system has been running in parallel with the operational but I believe it will be suffient for wave forecasting all over the UK.

All the remaining days—-from, +2, the “day after next”, through to +6—-will remain at 12 km. This is because fine-scale details lose their predicitability at these timescales—-no-one in the world runs a 4 km beyond 48 hours.

Nerdy bit
Skip this part if you want but do read the “caveats” section beneath.

You may wonder how the resolution of a model is chosen. There’s two factors—-the “nest ratio” from the parent model, and the meteorological impact of the resolution. In the case of RASP the parent/”driving” model is the GFS, which has a global resolution 36 km. A model “nested” within this parent will have a grid resolution defined by an integer ratio. A typical value is 3, which is why a common RASP resolution is 12 km—-36 / 3. The old “window” run was nested within the first RASP run, again with a ratio of 3, giving a resolution of 4 km. What Paul has down with the new RASP is push the first grid ratio to 9. This is steep, but is actually exactly what NOAA does with their high-resolution models I mentioned earlier (we actually consulted with the chief NOAA 4 km model-wrangler, Matthew Pyle, on this). For the “next day” forecasts the grid ratio is 7, giving the 5.1 km grid.

The other part is the meteorological side. A global model, running at a grid scale of 25-40 km, can’t see individual cumulus clouds—-they’re too small. Instead the model uses something called “parameterisation”, which is a kind of sub-model which approximates the effects cumulus has on the rest of the atmosphere without actually modelling them themselves. This is called, erm, cumulus parameterisation. This is generally used down to a resolution of 12 km. At scales of 5 km or less, the models can actually start beginning to model individual cumulus clouds. Between 12 km and 5 km is the “cumulus gap”, where parameterisation is too an inaccurate approximation, and the grid is too large to model convection explicitly, so you’ll rarely find models running in this middle ground (I think the Germans have one though. Those crazy Germans, eh?).

Now when I say “cumulus” I should really say cumiform—-we ain’t talking about individual little puffy clouds here (that would need sub 1 km grids), we’re talking full-on cumulonimbus storm clouds. In the US the 4 km models are used, with considerable success, to forecast supercell storms (the type that cause tornadoes—-and tornadoes have killed 98 people in the US so far this spring, the worst toll for ten years—-and the “season” runs until the end of June…). Here in the UK true supercells are rarity but we certainly get big storms, and 4 km RASP will model these. They’re visible as little diamonds (the intersections of the grid lines) on insolation charts, or little blue holes on boundary-layer top plots (the latter because the storm’s downdraught reduces the BL to zero height).

Notably the new UK RASP appears to be modelling cloud streets (technically “horizontal convective rolls”). There show up as closely spaced with-wind lines of different cloud bases, or strips of vertical motion. RASP is a little too eager to forecast these sometimes, but is by and large quite good with it.

Caveats
This is an important bit, so pay attention 007. In the US, the high-res models are used to forecast the “mode” of convection (supercell, multicell, line squall etc.) and the area where such storms can be expected. What they won’t do—-well actually they do do, but not often enough to be reliable—-is forecast the exact positions of individual storms. Same goes with the new UK RASP. It will forecast areas of wave, but it won’t be kilometer-perfect, nor will it show individual wave bars. It will often indicate areas of horizontal convective rolls, but it again won’t be perfect in where it puts them. It will show sharp changes in things like dewpoints and inversion heights, and those boundaries will be out there, but not exactly where RASP shows them. (One thing that it does do with remarkable accuracy is show sea-breeze fronts. It’s uncanny with those).

I hope this little run-down has been interesting and informative, and you’ll all be making best use of the new UK RASP  .

Tonnes of feedback on what was an interesting weekend, particularly Saturday.

Roughly in order of arrival:

• The first report, from Andy Miller at Keevil, also happened to be the 100th comment posted on this blog! Andy reports that a southeasterly wind is actually the best for wave off the Wiltshire hills over wave, giving flights to 3,000 ft for hours at a time. Another pilot reported wave as high as 4,000 ft there, and also encountered large directional wind shear above and below the inversion
• Mike Orme and another pilot from North Hill found lift to 2,000 ft started by cliffs to the south of the field
• Paul Fritche at Parham found wave to 3,600 ft just 500 m from the ridge crest. Paul’s post also sparked the first discussion within the comments, something I’ve been hoping for since introducing them
• Bruce Cooper found wave lift outside Dublin (somewhat outside my forecast area too, really!)
• Ray Staines at Dartmoor Gliding Society reports a two hour flight in Sunday’s easterly, in the lee of Dartmoor
• Peter Thomas fround a very little wave over Dunstable, and there was the odd thermal to be had below the low inversion
• Mike from Lasham also found a little weak thermic lift on Sunday.

Paul’s question about the closeness of the wave to the South Downs ridge gives me the opportunity to use my favourite wave picture:

This was taken during the 1950s in Owens Valley, California, when mountain waves in the Sierras were just beginning to be studied. The wind is coming from the right to the left and it is accelerating down the lee slope of the mountains, rushing across the valley floor picking up dust, and then rushing upwards on the left in the hydraulic jump, where you can see the lenticular clouds forming above the rising dust. Paul could well have been soaring a rather smaller (and less dusty!) but otherwise identical system coming off the South Downs.

The Herstmonceux sounding for Saturday shows what things were like:

The top of the picture is about 12,000 ft. As you can see, the air is very stable. Air flowing over hills would want to flow down the lee side, and then spring back up as a soarable jump, forming a lee wave.

Forecasts running up to the weekend were fine weather-wise, but I had not thought about the possibility of wave until Phil King pointed it out (thanks Phil ). I was thinking that with the southerly wind and high stability wave was unlikely—-until I thought about it some more! Phil’s post is exactly the kind of thing I want to encourage—-more discussion and critique of the forecasts. Many heads are better than one!

I was a little over-worried about stratus, but it’s easy to look very silly when having forecast clear skies all week, a day turns out to be dull and overcast. This particular high is proving to be mostly cloud-free, although stratus has now indeed formed over the eastern North Sea, southern Scandanavia, and the Denmark/northern Germany area. If the wind is from the north east, that stratus can easily blow over us, often thickening as it crosses the North Sea.

Well, it would seem I am not a plonker  .

There was good wave over Wales and downstream and a number of flights have been uploaded to the ladder. Several are short XCs, and it’s especially good to see flights from Talgarth, the Mynd, and Denbigh (though the Denbigh flights could have been uploaded as XCs, making it more obvious to any passing CAA types where they originated from…). Stand-out flights have to be Diana King’s 139 km on Saturday and Phil King’s 238 km on Sunday.

Although the wave had been well-forecast all week, the main problem was the strong wind on Saturday which I totally failed to mention. To figure out just how long XCs in wave can be done, I decided to look at profiles for John William’s three longest flights last year. (It took me a while to find a data archive!) The profiles have been drawn using the IDV.

The first is from 8 April 2007 when John flew 1,000 km on a north-south track across Scotland:

The classic bulge of a stable layer is there, and the wind gradient increases from 15 knots at the surface to about 40 knots at flying height. Above that, it slooowly increases to 70 knots at the tropopause. John averaged 130 kph running up and down the bars in his Antares (handicap 113, which helps!).

The next is from 30 August when John flew 770 km—-twice. At nearly 150 kph. A Libelle pilot did 500 km the same day too!

This shows a fairly small stable layer which is also quite high up. The wind gradient is slightly less than before, from around 10 knots at the surface to only 30 knots at height, with higher speeds of 60 knots or so not coming until near the tropopause. Speculating, I expect it was the lower wind speed that allowed gliders like a Libelle to cover such long distances.

Finally here’s the sounding from 10 September when John flew another 1,000 km, this time at near 140 kph:

Pretty damn big isothermal layer on that one! John flew most of the flight around the 700 mb level, which was at 40 knots again. This time the 50 knot winds came much lower, as far down as about 12,000 ft. I wonder if the stronger stability increased the strength of the wave to offset the extra wind?

Judging from those three examples it seems that the best wave in Scotland at least often sees around 30 knots at flying height. So if we look at Saturday’s profile for Wales:

You can see where things went a bit wrong, with winds of 50 knots at 6,000 ft! The stable layer isn’t all that strong either, so the wave wasn’t the strongest it could be. Here’s a sat pic from Saturday (around 1pm) showing the results of all that:

Looks nice and defined on the satellite (some high-level cirrus, as forecast, obscuring things a little), but of course you can never tell the windspeed or wave strength on a satpic, so it can be deceptive. You can see that Nymsfield also benefitted from the wave, with a couple of folk soaring for over 5 hours there.

Sunday didn’t look, or at least I thought it didn’t look, so promising for wave. Here’s the sounding for Sunday:

What worried me was that although the wind increased with height until around 850 mb, it then began decreasing. My (unspoken) suspicion was that it would result in breaking waves rather than trapped standing waves, and indeed the overnight F215 warned of severe turbulence in the area, while Sunday’s made no mention of neither wave nor turbulence! However, look at that stable layer! Much bigger than on Saturday. And more significantly, those winds are much lower—-max 30 knots on that sounding. The result was wave which looked like this:

Quite smaller bars (and also a lump of annoying cloud to the north). (Also note the with-wind roll clouds in west Wales.) The lower speed allowed Phil King to do his 238 km flight. (I did want to try over-laying Phil’s flight on a 250 m/pixel MODIS image, which has worked very well before, but hardware failures at NASA at the weekend mean that there’s no data available at the moment  ).

So in future I’ll be paying more to what the wind speeds actually are, rather than just what the gradient is doing!

One other boo-boo I made last week was not pointing out that Thursday’s windspeed and direction was just right for Nympsfield’s ridges. Indeed two 177 km flights were flown that day! One of the pilots also took a thermal to 4,000 ft —- not long now to the thermal season…

Well, it’s the time of the week for the review, and you’ll find it at the bottom of this post, but I thought I review something a bit different first for a change.

Last night I set my alarm clock for 3:30am and watched the last half of the final race at the Gliding GP live. The coverage was absolutely stunning and fully lived up to its billing. The real-time 3D graphics were excellent: you could see pilots flying in wave while down below others struggled on a ridge before thermalling back up. You could see every last turn of each thermal and the pilots dolphining in the cruise. At one stage they even moved the virtual camera behind a glider which while well out in front had become rather low. The graphics showed the glide angle back to home: it looked very marginal, and indeed the pilot (Bruce Taylor, race winner the previous day) soon landed out. The in-cockpit cameras worked a treat with views both back at the pilot and forwards from beside their heads so you could read in the instruments (and look at the stunning views!). There were also external air-to-air shots from a helicopter flying beside the last part of the course—-poor Bruce Taylor was filmed landing out live!

The GP format, a mass start with first person back the winner, is compelling viewing. Trying to figure out who would be first back—-the guy who took the thermal to 8,000’ or the pilot lower down but further along the course?—-was great fun. The commentary team were very good (even if Gavin Wills, I think it was, kept referring to Pete Jones and Steven Harvey the whole way round!). In the end Pete Harvey won the day, but Steve Jones was just edged out by Sebastian Kawa in a Diana 2, meaning Steve fell from second overall to fourth, and Kawa became world champion. It was very exciting; never have I watched gliders meaningfully racing for the line, and desperately rooting for Steve Jones to catch up with Kawa on the long final glide.

The GP has seen five race days out of six, but there nearly wasn’t a fifth. A cold front almost killed the end of day five (some guys only just making it back under the increasing top cover) and it was unclear whether the front would clear in time the next day for a final day of racing (which would have left New Zealander Ben Flewett champion and Steve Jones second). I’m going to take you through how I would have forecasted the final day.

New Zealand, probably as a result of its small population, has nothing like the meteorological resources available to us in Europe. No HRV, no spectral analysis, no XC Weather equivalent (there seem to be very few METAR stations on the South Island), and certainly no mesoscale weather models (including no RASP). So it’s a bit of a challenge. What is available is IR images from a Japanese geostationary satellite giving European-quality images every three hours and rather poorer (too poor to use for New Zealand really) images every hour, plus the same four polar satellite passes we get here (polar satellites cross the same point at the same local time every day). There are basic daily synoptic charts available from the Australian and New Zealand met services, and of course GFS data (the “G” is for Global, after all). You could use ARL to look at the GFS data—-and indeed I believe that’s what Oz and NZ pilots do—-but I prefer to use the IDV.

The very first thing to do though is work out exactly where Omarama is, so here’s a picture with a big arrow:

So, let’s say that after the flying on day 4 (23 December) I’d have looked at the data for the first time. First the most recent IR sat pic:

You can see the mass of cloud bearing down on the South Island from the west. Will it clear? Well the easiest way to see is to look at surface solar heating—-if the sun’s on the ground the sky is clear so there’s every chance of at least some thermals on the mountainsides, and maybe ridge and wave too if the wind’s right. At 5pm NZ time the latest available forecast would be the 23 Dec 00z run, which broked, leaving no 0.5 degree data available. However the 1 deg output did work, so here’s a rather blocky look at the surface heating forecast for 4pm on the final day, 24 December (all the following forecasts are for this time).

Why 4pm? Well Omarama is only about 35 degrees south, and of course it’s the height of summer for them, so if the sun’s out by then, gliders can play. The blue is shade while the red is maximum downwards radiation (about 1 kW/m2). This forecast shows cloud clinging to the north coast, but a bit of a chance for Omarama—-probably something like “broken cloud”.

On the morning of the final day, the cold front was right overhead. Available would be the early morning polar satellite pass, which in this case was taken at 5am local:

It shows the edge of the front approaching Omarama but by this stage its progress across the island was slowing, and there’s also a deck of cloud behind the front with only a narrow clear slot right behind the front itself. How would things evolve?

At that time in the morning the most recent GFS run available would be the 23 Dec 12z run, and here’s its surface radiation forecast (this time at proper 0.5 degree resolution) for 4pm:

This shows the Omarama being rather cloudier that would be liked. IDV can also generate soundings, with a very clever trick: in one window can be the selector for the sounding point (shown as the small coloured rectangle), while in another is the sounding itself. You can drag the sounding point around the map and watch the sounding change as you do in the other window, making it easy to find the height of the inversion, the wind with height etc. in various places. Here’s an example for the cloud near Omarama:

This shows the fairly thick cloud, though of course the bottom of the chart is sea level at the Southern Alps reach to over 6,000’—-around 850 hPa. One plus is that winds look to be at a good angle for the mountain ridges to work.

Another sounding shows the clear air to the west of Omarama with a stable layer above the boundary layer—-good for wave?:

The IDV can also show the cold front itself using a cross-section (the vertical line marks Omarama) which helps with figuring out what’s going on. You can see the colder, drier air to the west and the warmer air ahead, with grey cloud having formed along the boundary:

Here’s a simpler plan view of just the cloud:

So from these forecasts, it’s clear that it would be, er, borderline. How did it actually turn out? Well the day did clear enough for a race. Here’s the afternoon polar satellite pass, which as it happened was bang on 4pm local, the time all these forecast images have been for:

The forecast surface radiation and cloud predictions compare well to this image: a good performance by the GFS, I think. The thin cloud deck behind the cold front was, according to the sounding, barely higher than the mountains and dispersed in the Foehn effect downwind of them, as the descending air compressed and warmed.

And if glideomarama.com ever want a weatherman, I’ll be right there…

Meanwhile, back in Blighty
It’s wet and crap. Saturday was fairly well forecast but Sunday was in no way “brighter”, with the mist and fog only forecast properly the night before. This was a big boo-boo on my part—-I’d had my suspicions that cold, damp air might cool to dewpoint under clear skies (no, really?!) earlier in the week, but didn’t communicate that until I knew for sure. Which meant all the forecasts for Sunday were rubbish really.

Still, the fog did give a nice picture on Keevil’s webcam late on Sunday:

Though I’d rather be looking out on snow-capped sun-drenched mountains. I’ll be back on Boxing Day; have a Happy Christmas.