12z Model Comparisons Results

[Thundery wintry showers]

Model Comparisons Results for Friday 31st march 2006.

985mb low to the west of Britain. 1000mb low over Cent Scotland, westerlies elsewhere, changing to SW in the west.

GFS: Had lows too far west at T+144 and southerlies. Had main low over N Scotland at T+120 with showery westerlies, but the positioning of lows completely wrong, and Scotland was certainly wrong (westerlies as opposed to the 1000mb low). Good at T+96 though, with the lows in the right places and the right wind arrangement, maybe the Scottish low slightly further north than actually turned out.

Because of the good T+96 this gets 5/10, but T+144 and T+120 were not great.

UKMO: Had a low index pattern at T+144 with a low over the north and one over the south. Same mistakes as GFS at T+120- low focused to the north, one secondary depression to the SW, westerlies correct but low positionings completely wrong. As GFS, accurate at T+96.

Similar to GFS, so another 5/10.

ECMWF: 980mb Icelandic Low extending into Scotland at T+144 with westerlies over Britain- much closer than UKMO or GFS to the correct outcome. Not bad at T+120 either- had south-westerlies over Britain (low to the W slightly wrong shape) but lows in pretty much exact correct positions and the general weather pattern was correct. Accurate at T+96.

Some quibbles over the shape of depressions but overall a very consistent- and this time accurate- run from ECMWF. I give it 8/10, as the weather forecast using ECMWF would have been accurate at all three timeframes.

NOGAPS: Had everything too far north at T+144, with a ridge of high pressure for the E and SW winds in the W, so correct wind patterns but lows/highs in the wrong places. Good at T+120- slightly underdid the low to the W, but the correct pattern was there. Same at T+96.

This one gets a 7/10. Not bad at all, especially by NOGAPS's recent standards.

JMA: Lacking detail at T+144- large Icelandic Low and SW'lys. Had the westerly pattern at T+120 but the wrong positioning of lows (too far N and E, secondary low to the SW that didn't come off). Pretty good at T+96- had the main low to the W a bit far east, but otherwise accurate.

A bad T+120 brings this down to 6/10, not too consistent, but the general W/SW pattern was correctly shown.

GEM: Had low over the British Isles at T+144, with little resemblance to the actual outcome although it correctly suggested that lowest pressure would be to the W. Had 1000mb low off SE England at T+120 and light winds except in the south (westerly). Low still over Britain at T+96 with a low index pattern.

Some vague resemblances to what actually happened, but there's no way any forecast would have been accurate at any three timeframes. Probably a 3/10.

Today's runs reflected Evo's point above quite nicely- certainly, it was a good run from the ECMWF which was the best of the models today. At 6.5/10 the ECMWF heads the overall standings with the UKMO second on 6.0/10.

model_comparisons.xls

[Thundery wintry showers]

Updates to the Comparisons Results after having been away in Amsterdam for the weekend.

Saturday 1st April 2006: 990mb low over W Scotland, WSW winds over all but the north of Scotland (light/variable) and a mix of sun and showers. There was, unfortunately, no T+96 in this comparison.

GFS: Had us already in a NNW regime at T+144. T+120 was good though, the low perhaps a little deeper than actually turned out, but everything else correct.

Based on T+144 and T+120, this gets 6/10- a poor T+144 but a good T+120.

UKMO: Had the main low too far north at T+144, with a westerly pattern, and a phantom secondary low off SW England, but better than GFS. T+120 was, however, pretty good- again, the low a little too far north though, but closer than at T+144.

This also gets 6/10. The T+144 was better than the GFS but the T+120 was slightly worse than GFS.

ECMWF: Top-notch at T+144, everything pretty much spot on. Not as good at T+120 with a low in the North Sea shown that didn't materialise, but the rest of the pattern was accurate.

This run gets 7/10- T+120 wasn't great, but T+144 was excellent.

NOGAPS: Lows more or less over Britain at T+144- pretty poor. Better at T+120, but lacking in detail, with westerlies over England and Wales.

This one gets a 4/10.

JMA: Not bad at T+144- it identified the WSW pattern correctly, but had a secondary low off the E coast of Scotland that was rather overdone. The main low was too far west at T+120, and consequently our winds were given as showery SW'lys.

Another 6/10.

GEM: Very good at T+144, but fell down badly at T+120, with northerlies pushing in across the north, and low pressure across the south.

For the poor T+120 it has to be marked down to 4/10.

Again the ECMWF just shaves it into first place for the 1st April.

Sunday 2nd April: 1000mb low over NE England, with winds from the W in the south, and the N in the west. Sadly no T+120 available for Sunday 2nd April.

GFS: Overdid high pressure moving into France at T+144 with everything too far north, and WNW winds. At T+96 it was very good though.

A 6/10 in view of the good T+96, but T+144 was quite poor.

UKMO: Had slack westerlies at T+144 and everything too far north. Everything was still too far north at T+96 although it was nearer the correct outcome, it had westerlies over everywhere bar the far N of Scotland.

Just a 5/10 for UKMO.

ECMWF: Once again, a good performance at T+144. It was also accurate at T+96. The small low over NE Britain was correctly modelled both times also.

Hence, a good 9/10 for the ECMWF.

NOGAPS: Only T+144 available, so no rating.

JMA: Kept low pressure over NW Scotland at T+144, so quite a fair bit out. Closer at T+96, with a big 995mb low over S Scotland and strong westerlies over England and Wales, still quite a disappointing output on the whole.

A 4/10 for the JMA.

GEM: Overdid the northerlies at T+144 with northerlies across the whole of Britain. Better T+96, pretty accurate output with the lows in the correct places.

A 6/10 for GEM.

An excellent output from the ECMWF; other models had trouble at T+144.

Sunday 3rd April: High of 1025-1030mb to the west and NW, 1005mb Scandinavian Low, and NNW winds bringing sun and wintry showers. This time T+144 is unavailable.

GFS: Accurate at T+120, with everything in pretty much the right place. However, poor at T+96. It had a 1010mb low off SW England introducing warm southerlies for the south.

Only 4/10 for the GFS, in spite of the good T+120, it was awful at T+96.

UKMO: Poor at T+120- had a strong Bartlett/Euro High stopping the northerlies from coming in, and bringing relatively warm westerlies except over N Scotland. Not bad at T+96, with north-westerlies, although again overdid high pressure to the S.

Just 4/10 for UKMO also.

ECMWF: Overdid high pressure to the S at T+120, overall very similar to the UKMO's T+96 output. Very accurate at T+96 though.

A 8/10 for the ECMWF.

NOGAPS: No output at all!

JMA: Too low-index a pattern at T+120 with overemphasis on a Euro High, but otherwise pretty accurate with northerlies over Britain. Very accurate at T+96.

Also an 8/10 for JMA- no worse than the ECMWF.

GEM: Tried to introduce easterlies for the south at T+120, but had northerlies elsewhere. Had the same pattern as UKMO at T+96.

Overall a 6/10 for the GEM: not too good at T+120, okay at T+96, but both outputs weren't too bad.

Another good ECMWF, which seems to be modelling this showery W/NW episode very well, JMA also doing well for the 3rd April.

In the overall standings the ECM now has a clear lead, while remarkably, JMA is marginally ahead of GFS.

model_comparisons.xls

[Thundery wintry showers]

After falling behind somewhat following the Amsterdam trip, the rest of the comparisons will be given today:

Tuesday 4th April: 1030mb high ridging into W Ireland, 1005mb low over Scandinavia, northerly winds over the British Isles.

GFS: Had the high too far north and east at T+144, with anticyclonic northerlies around its periphery, although on the whole not too bad. Not good at T+120 though, showing easterlies over Britain. Very good at T+96.

A 6/10 for this GFS run. It had a bad wobble at T+120, but at T+144 and T+96 it was reasonably good.

UKMO: The high was too far NW on the UKMO, with easterlies for the south and northerlies elsewhere, and not as good as GFS at T+144. Accurate at T+120, but at T+96 it overdid the high, giving us anticyclonic NW winds.

Only 6/10, due to the poor T+144 and indifferent T+96, but it was good at T+120.

ECMWF: Had the high too far west at T+144 and consequently overdid the northerlies, but a good attempt. Continued to overdo the northerlies at T+120, but again locked into the right sort of pattern. Accurate at T+96.

Once again the ECMWF pulls out a good run, which gets 8/10- the forecast would have been accurate for all three timeframes had the ECM been used.

NOGAPS: Only T+96 available, so no rating.

JMA: Had us in easterlies at T+144. Worked towards the right pattern at T+120 but still had easterlies over the south. Closer at T+96- had us in northerlies, but not as good as the other models.

Overall 4/10, it didn't really click until T+96.

GEM: Had easterlies in the south at T+144. Northerlies were shown over Britain at T+120 but the low positioning was wrong with highest pressure to the SW. Still had easterlies in the south at T+96.

Overall 4/10 also.

Model Comparisons for Wednesday 5th April: 985mb low just E of Iceland, 1025mb high with ridge into S England, with westerlies in the north. No T+96 available due to Amsterdam trip.

GFS: Had ENE winds over Britain at T+144- not very good. Had a trough over Scotland that didn't materialise at T+120, and NW winds in the south.

A 2/10 for this output- never near the correct outcome, though it at least had a high to the SW at T+120.

UKMO: Showed northerlies at T+144, but showed signs of bringing in the toppler at T+120.

A 5/10 for the UKMO.

ECMWF: Similar to UKMO- if anything overdid the toppling at T+120, with rain indicated for the north.

Also 5/10.

NOGAPS: No data.

JMA: Had us in N/NE winds at both timeframes, so only 2/10.

GEM: Had a toppler showing, but underdone at both timeframes with N winds over Britain, so 4/10.

Thursday 6th April is unavailable.

model_comparisons.xls

[anvilhead]

great thread. cheers tws

[Thundery wintry showers]

The past three days' model comparisons, bringing everything up to date at long last, so next week can resume on a "situation normal" basis.

Friday 7th April

Only T+96 available for Friday 7th April.

There was a 990mb low to the north, with strong westerly winds over Britain.

GFS: Not bad at all at T+96- had the low to the north a bit far south, but otherwise quite accurate, had WSW winds over the British Isles of northerly origin.

The T+96 output gets a 8/10, pretty good.

UKMO: Had the low in the correct place, underdid the Greenland High somewhat, and had milder air in the south than actually occurred. Got the pattern right for the north.

I'd say a 7/10 for the T+96 UKMO.

ECMWF: A 9/10, looks like it had it near spot on.

NOGAPS: There was no high over Greenland shown, the mild air was too far north, but the general pattern was there. 7/10.

JMA: 9/10, very accurate.

GEM: Had the main low to the north of Scotland too deep and too far north, so docked a mark for that, but otherwise very good. 8/10.

Thankfully, none of the models had any real problems at T+96.

[Thundery wintry showers]

Saturday 8th April

There was a 995mb low just off W Norway. 1030mb Mid Atlantic High. Westerly winds (strong) in England and Wales swinging N'ly further north and west.

Only T+96 and T+120 available.

GFS: Underdid the cold air at T+120, with Britain under W/NW winds, rather than just the south, due to the low pressure to the N being too far west. It was still slightly too far west at T+96, with westerlies changing to north-westerlies over Britain. However at both timeframes, the general pattern was accurate.

This run only gets a 7/10. The general pattern was correct, specifics such as positioning of the low were less accurate and would have had effects on the weather in northern Britain.

UKMO: Had Britain in a col at T+120, so not very accurate at all. At T+96 the pattern was correct, but the shape of the low was wrong, so all parts had NW'ly winds.

This run only gets 5/10 because of the poor T+120.

ECMWF: Accurate at T+120, but had the low too far west at T+96- indeed more erroneously so than the GFS, with westerly/WNW winds across the whole country.

This run gets 7/10- it was better than GFS at T+120, but worse at T+96. General pattern was good.

NOGAPS: Had Britain in a col at T+120, and at T+96, had the pattern correct, but the low too far west, so only 4/10.

JMA: Low way too deep at T+120, the general pattern was correct but winds were given as being stronger than actually happened. At T+96, it was pretty spot on.

This one also gets 7/10.

GEM: Had the low way too far west at T+120, with W/SW winds over Britain, but was very accurate at T+96.

This one therefore gets a 6/10.

Sunday 9th April:

All three timeframes available. 1030mb mid Atlantic high ridging from the Azores, 1000mb large Scandinavian Low and northerlies was the actual outcome.

GFS: Underdid the toppling at T+144, with the Scandinavian Low positioned too far west with another centre 200 miles to the north that didn't come off. However, it did have N/NW winds over Britain. At T+120, it was very good, and also very accurate at T+96- impressively so, in fact.

This run gets an 8/10. T+144 was relatively dodgy, but T+120 and T+96 were top notch.

UKMO: Well out at T+144- had a low pushing towards southern areas giving NE winds. Overdid the toppling high at T+120 but otherwise accurate, and very accurate indeed at T+96.

This run is marked down to 7/10 because of the poor T+144. Overall not quite as good as the GFS.

ECMWF: Similar to GFS at T+144- had the low too far west, so the toppler underdone, but it did have N/NW winds over Britain. Top notch at T+120. Went off a bit at T+96, the high too far west and the low too far north, with a col over most of eastern England.

The ECMWF gets a 7/10, because it was not as good as GFS at T+96, and was similar at T+120 and T+144.

NOGAPS: Had an E/NE regime at T+144. No T+120. At T+96, it was similar to the actual outcome, but lows were too far north resulting in more of a NNW regime.

Only 5/10, it's difficult to mark this one because of the lack of a T+120, but it wasn't up to the standards of the GFS, UKMO or ECM.

JMA: Quite good at T+144, but again committed the ECM/GFS error of having lows too far west. Accurate at T+120. At T+96 similar to ECMWF, had lowest pressure too far north and a col over eastern England.

7/10 for this one, similar to ECM overall.

GEM: Poor T+144 with Britain in a westerly regime. Quite good at T+120 but had a low to the N that didn't materialise and underdid the toppler. Poor at T+96, had southern Britain under slack westerlies, and slack northerlies for the north, with the Scandinavian Low several hundred miles too far west.

Only 3/10, it had quite a good T+120 but was very poor at the other two timeframes.

A poor performance from NOGAPS and GEM today, but good solid runs from the GFS, UKMO and ECMWF. After a long period of ECMWF domination, GFS just shaves it as today's most accurate model.

model_comparisons.xls

[Thundery wintry showers]

Monday 10th April

A pretty beefy 975mb low W of Iceland, 1030mb high to the SW, north winds in the south, westerlies in the N from a toppler. Very slack regime.

GFS: Underdid the toppler at T+144, with northerlies still in full sway over Britain. At T+120 underdid the Scandinavian Low and had WSW winds penetrating a lot further east than happened, so northern Britain was under a pronounced WSW flow though southern areas had an accurate projection. At T+96 it was almost faultless.

This run gets a 7/10, quite poor T+144, but good T+120 overall, and excellent T+96.

UKMO: Quite good with the general pattern at T+144, but had too pronounced a NW flow due to the high being the wrong shape and grossly underdid the Icelandic Low. At T+120 the general pattern was again good, but everything was too far north. Again very good at T+96.

Overall 7/10, similar to GFS in accuracy.

ECMWF: Got general pattern right at T+144 but the toppling high too far north, so we were all in NNE winds bar the extreme north. At T+120 and T+96, had the high too far north and east, giving us a rather more anticyclonic outlook than actually happened, and at T+120 still had northerlies over the whole of Britain, underdoing the low to the NW. However the general pattern remained correct.

Another 7/10, the T+144 was a bit better than GFS and UKMO, but T+96 and T+120, if anything, were less good.

NOGAPS: Had a high to the NW, which threw off the output at T+144 and gave NE winds for most. High too far SW at T+120 but general pattern correct. Very little to fault at T+96.

This run also gets 7/10, again the model was about as accurate as the first three.

JMA: Very good at T+144, and again at T+120 and T+96. A solid 9/10 for this one, very little to fault at any of the three timeframes.

GEM: Decent at T+144 but overdid the westerlies. The toppling high was then way too north at T+120 and even at T+96 as well, with NE winds over southern Britain. T+96 was so poor, that I'd have to give this a 4/10.

Overall a decent set of outputs from GFS, UKMO, ECMWF and NOGAPS, but JMA was easily today's most accurate run. The GEM seems to have been faltering somewhat recently, and today was no exception, it being easily the worst of the models.

model_comparisons.xls

[Thundery wintry showers]

Tuesday 11th April

985mb Icelandic Low, WNW winds over Britain, 1030mb high way out to the SW.

GFS: High pressure too dominant and flat at T+144, with strong W'ly winds. Very accurate at T+120, slight quibble with low positioning over Scandinavia but the flow pattern was spot on. Worse at T+96, with too flat a westerly pattern, but again had fronts and airmasses more or less correct.

Overall 7/10.

UKMO: Poor T+144 with high pressure over Britain. Better T+120 but still had the high pressure much too close. Again getting closer at T+96, but still struggling with having the high too close and the warm sector too far north.

Only a 5/10 for the UKMO. It over-egged the high pressure.

ECMWF: At T+144 the high was too far N and E and the low too far W, with SW winds over Britain. High too close, and too flat a pattern, at T+120, but certainly getting closer. Almost faultless at T+96.

Also 7/10, it generally had the high too close at T+144 and T+120, but was excellent at T+96.

NOGAPS: Quite a way off at T+144 with SW'lys. Accurate at T+120, only minor quibble being the low a bit further west than actually happened. Worse T+96 with the low too far south but general pattern correct.

Overall not quite as good as the GFS so only 6/10.

JMA: High pressure too dominant at T+144. General pattern good at T+120 but with the low too far W. Similar at T+96, with high pressure to the east overdone.

Overall a 7/10 also.

GEM: High pressure over Britain at T+144. It was still there at T+120, but a bit further south. T+96 was closer to the correct pattern but still had the high much too far north.

Only a 3/10 for the GEM.

Some more solid 7s from the GFS, ECMWF and JMA.

The UKMO seems to have slipped a bit recently, performing worse than the GFS, which has granted the ECMWF (seemingly the most consistent model so far) a healthy lead. GEM is really struggling in the current mobile conditions, while NOGAPS has improved but still isn't at the current GFS/ECM level. JMA, surprisingly, appears to have been every bit as good as GFS/ECM recently.

model_comparisons.xls

[Thundery wintry showers]

Wednesday 13th April

GFS: Poor T+144, had a Euro high trying to ridge in from the SE, with southerlies filtering into the S. Still poor T+120 for similar reasons, had WSW flow and a dominant Euro High. Accurate at T+96 though.

This run gets a 5/10, because T+96 was good, but the other two timeframes were rather poor.

UKMO: Okay at T+144- a bit too flat a westerly pattern, but had the lows and highs in the right places and pretty good, especially for a UKMO T+144. Poor T+120 with secondary low situated over S Scotland. Impressively good at T+96.

This run gets 7/10- good T+144 and T+96 but not so good at T+120.

ECMWF: The low to the N was way too deep at T+144 but overall quite a good stab, with westerlies dominant. It was still iffy at T+120 with the low situated too far south. T+96 was quite good but had slightly too flat a W'ly pattern with the low underdone.

This one also gets a 7/10, it was okay at all three timeframes, though by no means brilliant.

NOGAPS: Poor T+144 with low pressure in the wrong place, to our NW. Unavailable at T+120. T+96 had too flat a westerly pattern with the high too far north and the low underdone and too far north.

So only 3/10 for the NOGAPS run.

JMA: Low too far west at T+144, with a Scandinavian High on the chart, though it did have westerlies over Britain. Closer at T+120 but still committing the same errors, low way too far west. Still not much improvement even at T+96, with the low too far west.

I will give this a 4/10 because it did have westerlies over us, but almost everything else was wrong at all three timeframes. Not a good JMA.

GEM: WSW winds at T+144- well out, a very flat pattern also. Better at T+120 with cyclonic W'ly but with low too shallow and too far south, and lacking detail. Lack of detail, and low too far E, were the problems at T+96 but still had WNW flow over Britain.

This run just about scrapes 4/10, it couldn't have been used for forecast purposes but the general pattern was close at T+96 and T+120.

Apologies for the inconsistent data outputs at the moment, I have been having major issues in getting the damn things saved to my hard disk!

model_comparisons.xls

[Thundery wintry showers]

Thursday 14th April

We had 990mb low to the N, 985mb Icelandic low, 1015mb high to the S and slack to moderate westerly winds over Britain. T+96 is unavailable for this comparison.

GFS: Not terrible at T+144, the low to the N too deep and too far E, giving lower pressure and more of a NW oriented flow, but characteristics of general flow pattern correct. Flow patterns to the N at T+120 were faulty giving rise to a stronger WNW flow than actually happened and lower pressure and no high to the S and SE.

Because the W/NW pattern was correctly identified this run just about manages a 6/10, but I can't give it more than that as many of the specifics were wrong.

UKMO: Too cyclonic at T+144 with strong westerlies, everything too far east, but alignment of lows and highs not dissimilar to actual outcome. Very good T+120 though, again westerlies a bit overdone, but everything was similar to the actual outcome.

I rate this as 7/10 because the poor T+144 was followed by a good T+120- certainly better than the GFS effort.

ECMWF: Similar T+144 to UKMO. However T+120 was spot on, everything almost perfect.

This one gets 7/10 also, though as a whole it was perhaps marginally better than UKMO. Let down by a poor T+144.

NOGAPS: Nothing available.

JMA: Well out at T+144 with low pressure firmly in charge and very stromg westerlies. Poor also at T+120- the low to the N was hopelessly done, we had a strong NW flow over Britain.

Only 2/10 for the JMA.

GEM: Too cyclonic an outlook at T+144 with low pressure too far south. Same issue at T+120, though it at least had WNW winds correctly over Britain.

I give this 3/10- it was better than JMA, but overdid low pressure at both timeframes.

Due to insufficient data I will have to give 15 April a miss, but 16 April will be done. The JMA and GEM were poor, the UKMO and ECMWF the best runs.

[Thundery wintry showers]

Sunday 16th April T+144 unavailable. Hopefully, situation will be normal tomorrow.

990mb low to the north of Scotland, quite strong W winds in Scotland, slacker WNW winds further south. 1005mb secondary low over Holland.

GFS: Couldn't get any data for GFS at all- this was when the wrong runs were saving to my hard drive!

UKMO: T+120 was poor, with high pressure too dominant, and the low to the N was missing. T+96 was better but the low was still underdone, and a secondary depression at 1010mb was shown in the south, which didn't materialise.

Only 5/10.

ECMWF: Better than UKMO at T+120, had the low too far north, but at least it was still there. Slack pressure gradient shown over Britain. However, worse than UKMO at T+96, with too much of a NW flow, and the high to the south too far W.

This gets a 6/10, overall it was a marginally better run than UKMO but not a great one.

NOGAPS: Not enough data, this time the fault of NOGAPS itself.

JMA: Low too far north at T+120 but at least it was there, similar to ECMWF with too slack a pressure gradient over Britain. At T+96 it was similar to ECMWF again, but better as the high to our S was more accurately positioned.

Overall 7/10.

GEM: High pressure too dominant at T+120, with the low too far north, though it did correctly have us in westerlies. Apart from a lack of detail, it had everything right at T+96.

Overall this run gets 7/10- lacking in detail but the pattern was probably the most consistent/accurate of all the models.

For once a decent GEM run, while UKMO and JMA did okay but not brilliantly. It's a shame the mess up with the GFS saving to hard drive occurred, though I vaguely recall that the GFS had us in showery northerlies around this time, so wouldn't have done particularly well either.

model_comparisons.xls

[Thundery wintry showers]

Monday 17th April.

990mb low to our north and a strong, cyclonic WNW flow over Britain, though with high pressure close to the south.

GFS: I can't rate the GFS today as only T+96 is available, but I should be able to rate it tomorrow as both T+96 and T+120 will be available. For the record, it wasn't very convincing at T+96 with everything too far east.

UKMO: The characteristic poor T+144- everything was completely wrong, and it had high pressure over Britain. T+120 wasn't much better, the lows way too far north, and a slack northerly over Britain. T+96 was very good, however.

I give this run 4/10 because of the good T+96- the only good thing that could be said about the run!

ECMWF: At T+144 it had high pressure over Britain and the lows too far north, so not a good start. T+120 was excellent though, with only niggly differences (high to the SW marginally further north than actually happened). T+96 also excellent.

This run gets a 7/10 because of the poor T+144, but T+96 and T+120 were top notch.

NOGAPS: T+144 and T+96 unavailable, so I cannot give it a mark. T+120 was okay, with everything a bit too far north, but the WNW pattern over Britain correct.

JMA: Everything too far north at T+144 but not as bad as UKMO and ECMWF, with the WNW pattern correctly identified over Britain, albeit more high pressure oriented with a slack gradient in the south. T+120 had westerlies and a secondary depression off northern Scotland in addition to the main one (which was too far east), but again the correct general pattern. Considering the success so far, T+96 was a little disappointing, with too slack a WNW regime over Britain.

Overall 7/10 because the pattern was about right on all three timeframes, but T+96 and T+120 not as good as ECMWF.

GEM: High pressure dominant over S England at T+144 with everything too far north, about as bad as ECMWF. T+120 was excellent, with very little to fault at all, but T+96 was relatively poor, with an elongated low to our north and strong westerlies over Britain, a bit out from the current pattern.

Hence, only 5/10 for the GEM, with poor T+144 and fairly poor T+96, but a good T+120.

Overall, JMA was perhaps the day's most consistent model, but ECMWF was the clear winner at both T+120 and T+96.

model_comparisons.xls

[Thundery wintry showers]

Tuesday 19th April

990mb low off N Scandinavia, 1005mb Icelandic Low and moderate westerlies.

GFS: T+144 unavailable. T+120 had the lows in the wrong places, but the pattern over Britain was spot on, with moderate westerlies over Britain with SWs to the west. T+96 was similar- the lows were overdeepened but in the right places this time, with the flow over Britain spot on.

Given that only T+120 and T+96 were available I can't be too generous with GFS, I give it a 7/10. The general pattern over the UK was very well spotted but apart from that, the charts looked different to the actual outcome with, in particular, the lows to the N too deep.

UKMO: T+144 had the high to our SW too far NE, with the high over southern England. T+120 had the orientation of the flow wrong (WNW instead of WSW) but quite close, and T+96 was very good.

Overall, this also gets a 7/10.

ECMWF: Had a showery NW'ly at T+144, so not the best. T+120 was excellent, as was T+96. However, because of the poor T+144, this only gets a 7/10 also.

NOGAPS: Showery NW'lys at T+144. No T+120 or T+96 so I cannot rate it, but it wasn't looking good.

JMA: Showery NWs at T+144. Similar errors to UKMO at T+120 with more of a WNW orientation and colder air. T+96 was good though.

Overall just 6/10, it was similar to UKMO at T+120 and T+96, and worse at T+144.

GEM: At T+144 had the general flow pattern over Britain close to correct, but the low in the Atlantic well overdeepened. T+120 worse, with the overdeepened Atlantic low causing more of a strong SW flow to develop, but T+96 was excellent.

I give this 7/10 as well, it was quite good for GEM.

[Paul]

Ian, can I ask, are you using the 0.5 degree GFS for this as per nw extra and some of the extra charts on wz, or are you using the 1 and 2.5 degree standard charts on wetterzentrale?

[Thundery wintry showers]

Wednesday 19th April

A rather slack synoptic sitation with pressure at or just below 1010mb, a low over northern Britain, and a 1000mb low west of Iceland.

GFS: Way off at T+144, had a strong deep low to the NW and strong to gale force westerlies. T+120 was poor too- a very deep low over N Scotland and gale force westerlies. T+96 was better, but it still had the pattern largely wrong- a low to the N/NW and moderate westerlies over Britain. Lows and highs were generally otherwise in the right places.

A bit of a poor run, this one- in view of the recognisable chart at T+96 I will stretch it to 3/10, but really, a poor quality set of outputs from the GFS for today.

UKMO: T+144 was similar to the GFS T+96- the main low pressure a bit overdeep and too far north, resulting in moderate westerlies, but a better attempt than GFS's T+144. T+120 was still off, with a WSW flow and a deep low to the north-west. T+96 was very good, with the slack pattern over Britain correctly identified.

This run gets 5/10 because of the good T+96, but T+144 and T+120, particularly T+120, were rather poor.

ECMWF: T+144 similar to GFS's T+120 with a deep low over N Scotland. T+120 similar too. T+96 was better but with high pressure from the Azores too dominant across the south, and light to moderate westerlies.

This run only gets 4/10.

NOGAPS: Unavailable except at T+96- which was very poor, with low pressure situated over the British Isles.

JMA: Low pressure to the NW and strong SWs at T+144. It had low pressure to the north at T+120 and strong WSW winds. T+96 was similar to GFS, with low pressure too far to the NW.

Overall, only 3/10 for the JMA.

GEM: Had SSW winds at T+144. 1000mb low over Scotland and SW winds for most at T+120. Even T+96 was poor, with low pressure to the NW and south-westerly winds. Only 2/10 for the GEM.

A rather poor set of runs for today. There was a tendency on all models- particularly GFS- for lows to be overdeepened, and all models seemed to position low pressure too far north. Only the UKMO, at T+96, was close to the actual correct outcome.

model_comparisons.xls

[Thundery wintry showers]

Thursday 20th April.

We have a 1000mb low off W Ireland and cyclonic/slack southerly winds over the British Isles. Weak 1015mb highs to the E and NE.

GFS: Too progressive a T+144, with the low to the W overdeepened at 990mb, and SW winds filtering as far NE as Scandinavia. T+120 had the low of approximately correct intensity, but too far east. Remarkably the GFS was still out at T+96, having the low too far east and a SW flow over Britain.

It was too progressive all the way through, and I don't think 3/10 is unjustifiably harsh here. T+120 and T+96 had similarity with the correct outcome.

UKMO: Way overdeepened low at T+144 and strong to gale force WSW winds- worse even than the GFS! T+120 too progressive, had a secondary low that didn't materialise and the main low a bit too far north, but better than T+144. T+96 was poor again- way overdeepened low and strong WSW winds.

Again only a 2/10. Only T+120 had a glimmer of resemblance to the correct outcome.

ECMWF: Unrecognisable at T+144, overdeep low to the SW and low to the N. T+120 had a strong ridge from the Azores anticyclone and the low to the W was completely missing. Even T+96 was poor, with a low to the N of Scotland, westerlies, and a secondary low to the SW.

Only 1/10 for ECMWF, never near the correct outcome.

NOGAPS is available this time at T+120 and T+96, unfortunately for the people at NOGAPS. It had a Scandinavian Low and strong NW winds at T+120, but performed better at T+96 with the low slightly overdeepened and further north, and SW winds, but quite similar to what actually happened.

At a push, this gets 3/10, the T+96 output wasn't too bad.

JMA: Deep 985mb low to the W and SSW winds, however it had the low positioned correctly so a better T+144 than the other models. At T+120 had our low way too far east, over the British Isles. At T+96 overdeepened the low ridiculously (965mb) and had it way out to the NW, with strong to gale force westerlies over Britain.

JMA also gets 2/10, I was hopeful that it would be better than the other models, but only T+144 was remotely decent.

GEM: Low pressure right over Britain at T+144 plus a Scandinavian Low. 995mb low right over Britain at T+120. At T+96 had us in a flat westerly pattern with lows to the north.

Only 1/10 for the GEM.

I don't know what on Earth happened with the models for today; they were nothing short of atrocious. The ECMWF, the most consistent model to date, was a particularly big surprise. GFS and NOGAPS (!) had the "best" runs of the day, which managed scores of 3/10.

model_comparisons.xls

[Paul]

Ian, can I ask, are you using the 0.5 degree GFS for this as per nw extra and some of the extra charts on wz, or are you using the 1 and 2.5 degree standard charts on wetterzentrale?

Ian, not sure if you spotted this q?

[Thundery wintry showers]

I didn't spot it; I am using mainly the GFS charts from

http://www.wetterzentrale.de/topkarten/fsavneur.html

at "500hPa Bodendruck", but am not sure what the resolution of these charts is.

If the ones at N-W Extra are higher resolution then I will start using those instead.

[Paul]

Hi mate, yes those ones are at 1 degree upto 180 hours, the nw extra and datacentre ones are at half a degree. Will be quite interesting to know how much of a difference that makes :lol:

[Thundery wintry showers]

Friday 21st April

There was a slack pressure gradient over Britain with light NE winds across England and Wales, which resulted in rather dull damp conditions around the North Sea, but warm sunshine in the SW. A weak 1010mb low over the near Continent and 1020mb high over Scotland.

Note: This hasn't reached the point at which the high resolution GFS is used. I will make a note when this time is reached, as there is a 4-6 day delay between me saving the runs to my hard drive, and them being used in the comparisons results thread.

GFS: Not terrible at T+144, but had the low to the south too deep (at 1000mb) and northerlies over Britain, so another example of the model being too progressive. T+120 was poor, with the low not only strongly overdeepened (990mb) but also situated off northeast Scotland, giving a strong NW flow. T+96 was quite good, but the low was still overdeepened and too far north, so we had northerlies over Britain.

This run gets just 4/10, the general pattern wasn't bad, but the low was consistently overdeepened and positioned too far north, with T+120 especially poor.

UKMO: Low overdeepened and in North Sea at T+144, giving northerlies. It was similar at T+120, and this time gave rise to NNW winds. At T+96, was slightly better than GFS, with the low of the correct intensity but too far north, giving a slack easterly flow for all of Britain.

This was marginally better than GFS so gets 5/10, but certainly not a good run.

ECMWF: High pressure right over Britain at T+144, which to be fair wasn't too bad a stab, especially given that the low to the NW was positioned correctly, the only error being the lack of a low in the near Continent. However, T+120 was poor, with the low overdeepened and in the North Sea, giving NNW winds. T+96 was about as good as the UKMO, with the low positioned correctly but overdeepened, meaning slack NE winds for all.

This run was slightly better than UKMO, so I give it 6/10, but still not great.

JMA: Had 1000mb low in the North Sea and strong northerlies at T+144, then at T+120 was horribly wrong, with a 975mb low to the NW and strong to gale force WSW winds. However, T+96 was similar to that of the ECMWF.

Overall, this run gets a 4/10, with T+120 being very poor indeed.

NOGAPS: At T+144, it wasn't too bad, having a low over the near Continent, the main error was to have the weak Scandinavian Low too far west, such that it influenced the weather of Britain, and meant weak northerlies instead of weak easterlies. T+120 was poor, with low pressure almost on top of us. T+96 unavailable, so I can't really give this one a rating given that the other models scored well at T+96.

GEM: Overdeepened low too far north at T+144, with northerlies. Low on top of us at T+120, then went back to the T+144 evolution at T+96, with overdeepened low and northerlies.

Overall only a 2/10 for GEM, it didn't even become correct at T+96.

Overall, a slight victory for the ECMWF, but really, the models have performed poorly lately.

[Thundery wintry showers]

Saturday 22nd April.

980mb low E of Iceland, strong SW winds for most of Britain, but high pressure holding firm across the SE. 1030mb Azores High, slack high over the Continent.

GFS: A poor T+144 with lows in charge, and high pressure well out to the SW, and westerly winds. There was a 995mb Scandinavian Low, which kept the Euro High away. However, T+120 was very good, with the high a little too far south, but everything close to the correct outcome, and T+96 was also very good.

I give this run a 7/10, marking it down for the poor T+144, but it was certainly one of the GFS's best runs of late.

UKMO: Same errors as GFS at T+144. T+120 was very good, however- the low over Iceland a bit overdeepened, but everything more or less correct, and T+96 was also very good.

UKMO performed almost identical to GFS, and so gets a 7/10.

ECMWF: At T+144 had the Icelandic Low way overdeepened and the high too far south, but closer than on the UKMO and GFS. However, T+120 was almost identical, and thus worse than the UKMO and GFS. T+96 was accurate.

Overall, the ECMWF scores 6/10, being marked down for the relatively poor T+120.

NOGAPS: Only T+144 was available, but it was as poor as UKMO and GFS with low pressure in charge, and a Scandinavian Low.

JMA: Poor T+144 with northerlies and a Scandinavian Low. T+120 was similar to ECMWF- the high was too far south, and everywhere had strong WSW winds. T+96 was accurate.

Only 5/10 for the JMA.

GEM: Low pressure over Britain at T+144. Northerlies, with a low over France, at T+120. T+96 was better, but still had the low over the Continent holding on too strongly, with NE winds for the south.

Overall, only 4/10 for the GEM, it was close at T+96, but a poorer product than the other models.

Overall, the GFS and UKMO were today's winners, with poor T+144 outputs, but they were very accurate at T+120 and T+96. In the overall standings, ECMWF is clearly in the lead, with UKMO, GFS and JMA vying for second, and the GEM and NOGAPS performing rather poorly, though NOGAPS seems to be unreliable in actually providing 12Z outputs at the moment.

model_comparisons.xls

[Thundery wintry showers]

Sunday 23rd April

A pretty anticyclonic picture over the British Isles, a displaced Azores High over southern Britain, at 1020mb, and lows at 970mb over Iceland and to the NE of Iceland.

GFS: Pretty good at T+144, the high a bit further south than actually turned out, but overall not bad. T+120 was worse, with a 980mb Icelandic Low and south-westerlies shown. T+96 was also quite poor, with too much of an Icelandic Low, and SW winds over all but the extreme south, the high too far south.

Overall just a 4/10, which is a shame because T+144 was pretty good. The model fell off the rails at T+120 and T+96.

UKMO: Worse than GFS at T+144, but the high too far south, with westerlies for all. Same mistake as GFS at T+120, too cyclonic an outlook. T+96 was better, with high pressure over the south, underdone slightly.

I give this 5/10, at least it was good at T+96, but T+144 and T+120 were out.

ECMWF: At T+144 had secondary lows over the south, and a slack outlook but with the high too far south. T+120- same mistake as GFS and UKMO. At T+96, same as UKMO, pretty good, but high underdone.

This also gets a 5/10.

NOGAPS: Unavailable.

JMA: High too far south at T+144, with WSW winds over Britain. At T+120 still had the high too far south, and an overdeepened Icelandic Low, like the other models. T+96 was pretty good with a high over southern Britain.

Overall, this also gets 5/10.

GEM: High way too far south at T+144, showing westerlies over Britain. Again, same mistake as the other models at T+120. And at T+96, still had the high way too far south.

Only 3/10 for GEM, it didn't rectify its errors at T+96.

Overall, not a good performance by the models.

[Thundery wintry showers]

Monday 24th April.

A deep 980mb Icelandic Low and SW winds pushing into most of Britain, barring the extreme SE. Slack 1015-1020mb high over Europe, over SE England also.

GFS: Too cyclonic at T+144, with 980mb low to the N of Scotland and cyclonic WSW winds, high too far south. Excellent at T+120 though, pretty much nothing to fault. Not quite as good at T+96, but the general pattern of high over the SE, SW winds pushing in, was still there.

Overall a 7/10.

UKMO: Too cyclonic at T+144- low just to the N of Scotland and westerly winds. Excellent though at T+120- similar to both GFS and the correct outcome. Also very good at T+96, though not as good as at T+120, with the low over Iceland the wrong shape.

Another 7/10. To get higher than 7s, these models need to start improving at T+144.

ECMWF: Better than GFS and UKMO at T+144. The Icelandic Low was too far east, with westerlies pushing in, but the high was correctly shown as holding on in the SE. Good at T+120, and again at T+96.

The models have so far all produced good outputs at T+120 and T+96, but ECMWF was best of the three at T+144, and so takes a mark of 8/10.

NOGAPS: Unavailable.

JMA: At T+144 had too flat a westerly pattern, but the high was correctly shown as holding on in the SE. Good at T+120, but worse at T+96- the low way too far NW and high pressure over the entire British Isles.

Only a 6/10, a bit of a mess-up at T+96.

GEM: Not bad at T+144- lacking detail with a rather large 985mb Icelandic Low, and too cyclonic, but it did have the high holding on in the SE. T+120 was okay, but with too much of a pronounced westerly flow for the north. T+96 was okay, but had a NE flow over the far SE for some reason, that didn't materialise.

I give this 6/10, it had the general pattern quite good, but there were some errors in all three outputs.

Overall, another good performance by the three main models, GFS, UKMO and ECMWF, with the ECMWF again shaving it as the most accurate of the models.

[Thundery wintry showers]

Tuesday 25th April

A 975mb Icelandic Low, and strong WSW winds in Scotland, but with high pressure close to the south, rather slack high at 1015-1020mb over continental Europe and the Azores. Rain moving east.

GFS: Not bad at T+144- the low a bit too far east, and had a more pronounced frontal system, but everything else was approximately in the correct positions. Quite impressive for a T+144. T+120 again quite close, but with the shape of the low slightly wrong, so we were shown in WSW winds instead of W winds. Unfortunately, the T+96 was sufficiently erroneous to keep it from scoring an 8/10, becuase the low was way overdeepened, and we were still shown as having WSW instead of W winds.

Quite good, but the orientation of the low was never really correct. 7/10.

UKMO: Had high pressure too dominant at T+144- the low too far W and the high clinging onto the SE, front kept out to the west. Same mistake as GFS at T+120, and again at T+96.

Only a 5/10 for this one.

ECMWF: At T+144, had the high holding on too strongly in the SE with everything too far west and SW winds. Same at T+120. And again at T+96.

Only 4/10 for this one. It was consistent with the outlook- but consistently wrong, with the low and its attendant fronts too far west.

NOGAPS: Available only at T+96, but it had too SW'ly an outlook at T+96- I cannot rate the NOGAPS on only the T+96 though.

JMA: High pressure holding on too strongly at T+144, with an even worse output at T+120- high pressure over most of Britain, and the low way too far to the NW. Same at T+96.

Only a 3/10 for JMA- it kept high pressure in charge.

GEM: High pressure too dominant to the E at T+144, with the rain band too far west. Same at T+120. However, T+96 was excellent.

Overall 6/10, because of the excellent T+96.

The Euro models, and JMA/GEM seemed to hold onto the Euro high too strongly. This time around, GFS, with its more progressive outlook, was the most accurate of the models.

[Thundery wintry showers]

Wednesday 26th April

1025mb Azores High ridge over to southern England, with moderate W winds over Britain, 990mb lows over Iceland and to the east of Iceland also.

GFS: Had the Azores High too far west at T+144, and a weak low over N Scotland, so NW winds over Britain, but the general pattern was close. At T+120, had a 1005mb low over NE Scotland and too cyclonic an outlook, the Azores High too far south. At T+96, had a weak low over southern England for some reason, but everything else was spot-on.

Only 5/10 for this one, the general pattern looked good at T+144 and T+96, but T+120 was poor, and the specifics were generally wrong.

UKMO: Not bad at T+144- had a low to the North positioned slightly too far south, but the Azores High about right- the winds were a bit strong as a result. At T+120, 1000mb low N of Scotland, and the Azores High way too far south- similar to GFS's errors, but even worse. At T+96, it was again committing the same error- too strong a low to the north of Scotland, with showery WNW winds.

I can only give this 4/10- worse than GFS, certainly.

ECMWF: Had the two lows to the north too far west, but the Azores High was positioned correctly at T+144, and the general circulation was westerly, if a little more cyclonic than actually happened. Too cyclonic a flow at T+120 with the high too far south. T+96 was excellent.

I give this 7/10, certainly the best of the models so far.

NOGAPS: Available at T+120, which had low pressure too dominant to the N, and the high too far south. Slack WNW winds was the result. T+96 had the low to the N too dominant and overdeepened, so we were stuck with cyclonic WNW winds.

Only a 3/10 for NOGAPS.

JMA: High pressure still holding on in SE England on the T+144 run. Similar error at T+120, with a slack low also shown over NW Scotland that didn't materialise. Had a 995mb low over Iceland, and westerly winds over Britain and showery conditions, at T+96.

I can only give this 3/10 as well- it was consistent, but consistently wrong.

GEM: High pressure too far north at T+144, with the high covering southern England, and high pressure to the E too influential also, westerlies in the far north. High pressure right over Britain at T+120. Went for a cyclonic NW flow at T+96.

T+144 was vaguely close to the correct outcome, but the only timeframe to be so. I give GEM 3/10 as well.

A clear win for the ECMWF, which was the most accurate by a particularly large margin at T+96.