12z Model Comparisons Results

[Thundery wintry showers]

11 August

1005mb low in the North Sea, and 1030mb high out to the west and a pronounced northerly flow over the British Isles.

GFS:

144: Sadly unavailable.

120: Decent, but low a bit too far west, entrenching into NE England, with more of a cyclonic northerly regime than actually happened, 7/10.

96: Accurate, low just a little too far west, 8/10.

In view of the fact that T+144 was missing I give this 7/10.

UKMO:

144: Poor (just 3/10), high pressure well in charge, situated over western Ireland and no low pressure in sight, the only thing it got right was the fact we had northerlies.

120: Reasonable, similar to the GFS T+120, with 1010mb low entrenching into NE England, 7/10.

96: Very accurate indeed, a solid 9/10 for this output.

Overall 7/10.

ECMWF:

144: Low pressure too far north, giving more of an anticyclonic north-westerly flow, though not as bad as the UKMO. 4/10.

120: Decent, but the northerly flow was somewhat overdone, strong to gale force winds suggested, due to high pressure being too far east, so only 6/10.

96: Very accurate (9/10)

Overall 7/10.

NOGAPS:

144: Low pressure considerably too far north, Azores ridge into the south, westerlies, 3/10.

120: High pressure too close, with anticyclonic NW winds, similar to UKMO T+144 and only 3/10.

No T+96, for whatever reason, NOGAPS have not been churning out the T+96 outputs properly to WZ. I can only go by the T+144 and T+120, so 3/10 overall.

JMA:

144: Low to east somewhat underdone, but it had us in northerlies and the high was correctly placed, 6/10.

120: Similar but with the high too far east and an anticyclonic northerly, low even more underdone so only 4/10.

96: Low underdone and too far north, with NNW winds over Britain, high pressure in the west, 6/10.

Overall only 5/10.

GEM:

144: Decent, the low only a little too far north, giving a cyclonic NW/N flow, though the high to the west was all wrong, based predominantly on a ridge from Greenland. 6/10.

120: Poor, 1025mb high to the SW and anticyclonic WNW winds, 3/10.

96: Decent, low underdone and a bit too far north, but we were correctly in northerlies, 7/10.

Overall 5/10.

The major three models- GFS, UKMO and ECMWF- all performed reasonably at T+120 and extremely well at T+96, but the NOGAPS, JMA and GEM all had problems, generally positioning the high too far east.

[Thundery wintry showers]

13 August

Unfortunately I was out on Saturday night so didn't get data for 12 August. 13 August had a 1005mb low off SE England and a cyclonic north/north-easterly regime over Britain. A toppling ridge of high pressure at 1020mb was evident off NW Scotland, with 1030mb high to the west.

GFS:

144: Not bad, there was an overdeepened Icelandic Low resulting in a more pronounced westerly flow to the N of Britain, but it had a 1005mb low in the North Sea, and north-easterlies over Britain. I give this 7/10.

120: General pattern correct- toppling ridge picked out- but low to east underdone. A number of niggly errors also bring this down to 7/10.

96: Very accurate, a good 9/10.

Overall, a solid 8/10.

UKMO:

144: Low marginally too far north, otherwise an outstandingly good output, especially for a UKMO T+144, and scores 9/10.

120: Low too far to the north-west, giving northerlies rather than a north-easterly flow for the east, still good but niggly flaws bring it to 7/10.

96: Only a 6/10- the main low was positioned off NE Scotland and we were shown in NNW winds.

Overall 7/10.

ECMWF:

144: Poor, only 4/10, with low well away to the east, secondary lows to the N, northerlies swinging to north-westerlies and high pressure too far east.

120: Better, but low pressure still too far east, the low also too far east, and a straight, fairly anticyclonic northerly regime, 5/10.

96: A reasonable 8/10, with the low marginally too far north, but on the whole a good output.

An overall score of 6/10.

NOGAPS:

144: Ridge of high pressure, 1020mb, over Britain, with anticyclonic NW winds, 4/10. Low too far to the SE.

120: Almost good, but had 1010mb secondary low over western England and a messy looking northerly regime, 5/10.

96: Still no NOGAPS for T+96.

Overall 5/10.

JMA:

144: Low pressure too far east, but Britain still in northerlies, 6/10.

120: Still had the low too far east, and like NOGAPS, had a secondary low over western Britain, though a generally northerly regime, 6/10.

96: Low too far to the NE, with straight, strong northerlies, but again general pattern okay, 6/10.

Overall score 6/10, steady if unspectacular.

GEM:

144: High pressure anchored to the NW of Britain with strong north-easterlies, but it had the right pattern with a low at 1005mb off SE England. Overall a 6/10.

120: Low a little too far north but the general pattern of northerlies correct, 7/10.

96: Reasonable, I give this an 8/10, the low pressure was only slightly underdone and was in approximately the right place.

Overall 7/10.

The UKMO was outstanding at T+144 but went off the rails a little at T+120 and T+96. The ECMWF was poor at T+144 and T+120 but got it right at T+96. The GFS was the most consistent in showing the correct outcome, although the GEM was reasonable as well.

[J10]

Hi Ian

Don't know if this helps but this is the actual synoptic chart on the 12th august at 12 noon.

[Thundery wintry showers]

Thanks a lot- I should therefore be able to update it, but I need to get to bed shortly- I'll update 12 and 14 August at around 10:30pm tomorrow.

[Thundery wintry showers]

I'm afraid I have a lot of project work on at the moment for 30 August, so I won't be able to continue updating these summaries on the N-W forum on a regular basis- the backlog is getting too much, for example.

I won't be giving it up entirely for this coming fortnight though- it is my intention to continue updating the Excel file with daily marks for the models, it's the daily summaries that take up the most time.

Apologies for this.

[Thundery wintry showers]

Just saying, I haven't given up on these, it's just that at present I have a large project- it is intended that these will resume after 30 August (including an update of the scores, albeit no written summaries, to around 22 August)

[johnholmes]

tks Ian it is a very big commtiment so I do hope you can keep it going.

John

[The PIT]

Pity you're busy would have been interesting to see how the GFS has faied in the last week or two.

[Thundery wintry showers]

File currently updated to 15th August for the time being.

Notable points with regards the northerly spell of 10-15 August:

I was impressed by GFS's handling of the northerly spell, it seems to perform well when handling a mid-Atlantic block. However, on the 15th, when a small trough became embedded in the flow, it was unable to pick up the trough and so scored poorly then.

UKMO generally did okay during the period but was dragged down by a dogged persistence in keeping high pressure too far east and south, especially at T+144. The ECMWF suffered from similar problems, but to a lesser extent. Both the UKMO and ECM were better than GFS at handling the trough on the 15th, though.

The NOGAPS was generally poor with the pattern often somewhat wrong, though it scored well for the 15th. JMA and GEM were okay, they usually got the general pattern right but were often wrong in where they positioned the low to the east.

The file will be updated to 25 August (I have enough data to update up to then) during the next few days, by which time the written summaries will be restarting. It is my intention that in the update to 25 August, the analysis of which models are performing best for a given synoptic pattern will also be updated to 25 August. On the evidence of the above, the GFS's previous tendency to score well under blocking to the west is likely to be continued.

model_comparisons.doc

model_comparisons.xls

[J10]

Thanks for doing these Ian, I can vouch for the amount of work required to keep them going.

[Thundery wintry showers]

1st September 2006

High at 1020mb over Azores with ridge into Europe, general 1008mb low to the north but 992mb low off north-west Scotland, with strong WSW winds. Only T+120 and T+96 available. I've changed the format again, partly to ease my own workload but also hopefully in a way that will make it easier and quicker for others to digest while not sacrificing the level of information.

The GFS correctly positioned the 992mb low at both T+120 and T+96, but at T+120 it overdeepened the low, and at T+96 it underdid the low. I was quite impressed overall, and rate it at 8/10.

The UKMO was very poor at T+120- low way off to the east over Scandinavia and NW winds shown, but reasonable at T+96, and scores 6/10.

The ECMWF did reasonably at T+120, though like GFS slightly overdeepened the 992mb low. At T+96 it had the low a bit too far west, and less of a pronounced westerly flow than actually happened, so I give this 7/10.

Only T+120 was available for NOGAPS but it performed poorly at T+120, the low too far west.

JMA was okay- low overdeepened and a bit too far east at T+120, with no secondary lows shown in the Atlantic, but at T+96 it was good, and scores 7/10 overall.

GEM was disappointing, positioning the low too far east (essentially just off northern Scotland) at both timeframes; T+120 was worse than T+96. I only give this a 5/10.

Overall, the GFS was slightly better than the others at handling the synoptic situation, although ECMWF and JMA were also reasonable.

[Thundery wintry showers]

2nd September 2006

992mb low off W Scotland, 996mb secondary low approaching Britain from the west, with strong SW winds. 1020mb high off the west of Spain.

The GFS had real problems in dealing with this at T+144 and T+120. At T+144 it had the primary low north of Scotland, and cyclonic W winds and no secondary low to be seen. At T+120 it had the secondary low a good 300 miles too far west, only at T+96 was it close to being accurate (even then the secondary low was underdone) So only 5/10.

UKMO wasn't really any better: cyclonic NW winds shown at T+144, at T+120 had the lows a long way too far west, though T+96 was very good. A score of 5/10.

ECMWF followed GFS at T+144 with a low over N Scotland, but then it got the pattern spot on at T+120 and T+96- secondary low slightly too far east in both cases, but it scores a good 7/10.

NOGAPS did okay- the T+144 had low pressure to the NW and SW winds, but no secondary low; the T+120 was quite poor with a mess of lows to the north, but at T+96 it was accurate. I give this 6/10.

The JMA was similar to GFS and ECMWF at T+144 with a large low plonked over N Scotland; T+120 was then very good, but at T+96 it had both lows too far west. Overall 6/10.

The GEM was similar to GFS, ECMWF and JMA at T+144- large low over N Scotland- but at T+120 and T+96 had a low to the NE of Scotland, and only at T+96 did it get the secondary low, which it positioned too far west, so only 4/10.

Overall, the ECMWF was the only model to get it close to nailed-on at both T+120 and T+96, and so was the most accurate of the models. No models were good at T+144, if anything the NOGAPS was closest at that timeframe.

[Thundery wintry showers]

3rd September 2006

Lows at 996mb to the north and a westerly airflow. There was a shallow 1004mb secondary low off NE England. Elongated 1020mb high over Europe.

The GFS was reasonable- it had the general pattern of low pressure to the N and NE and westerly winds correct at all three timeframes. However, at both T+120 and T+96 it overdid the low pressure to the NE and failed to pick out our secondary low, so scores 7/10.

UKMO was decidedly poor, with a low to the NW at T+144 and SW winds, then at T+120 it had the low pressure correctly to the N and NE but positioned a secondary low to the W of Ireland giving SW winds, then at T+96 it had an overdeepened low off NE Scotland. Only 4/10.

ECMWF was also poor, with low pressure to the NW and SW winds at T+144, it had us in a ridge of high pressure at T+120. It was, however, very good at T+96. Overall 5/10.

NOGAPS had southern Britain under the influence of high pressure and a warm sector at both T+144 and T+96, but was quite accurate at T+120- very inconsistent, only 4/10.

JMA was quite accurate at T+144, but then at T+120 and T+96 it had a large low pressure centre off NW Scotland (way too far west) and WSW winds- so it had the westerly pattern right, but the low positioning completely wrong, so only 5/10.

GEM had us in south-westerlies (quite a long drag) at T+144 and with slack low pressure over us at T+120. Only at T+96 did it pick up the correct pattern of low pressure to the N and NE and westerlies, and it did not pick out the secondary low. Only 4/10.

A bit of a struggle, the GFS did better than the others, mainly because it was consistently okay; the others had particularly big problems at T+144.

[Thundery wintry showers]

model_comparisons.xlsmodel_comparisons.doc

Have had trouble keeping up with these very well due to moving and work commitments, but here is the latest updated file, including up to 25 August and also 1-3 September.

[Thundery wintry showers]

18 September

Quite a deep 990mb low off NW Scotland, and a south-westerly airflow, a ridge of high pressure clinging on in eastern areas. The day was sunny in the east but rain spread in from the west.

Chart shown, unfortunately, is for 18Z, but my analysis is interpolated from 12Z based on the movement of the depression.

The GFS was pretty good at T+144, with the main low a bit too far south, giving more of a southerly flow, but it was pretty close. However T+120 was worse, with a 992mb low stick over southern Ireland, but the T+96 was very good. I rate this at 7/10.

The UKMO grossly underdid the low pressure at T+144- even having a full-on anticyclone over Britain, it was better at T+120 with a westerly shown but the low to the NW still underdone. T+96 was then very good. However, I mark this down at 6/10, as T+144 was especially poor and as a whole, UKMO was not as good as GFS.

ECMWF was spot on at T+144, and also good at T+120- low positioned slightly too far west at T+120. Then at T+96 it was superb, so overall an excellent 9/10 for ECM.

NOGAPS was available only at T+120, and had a straight westerly flow over Britain and low to NW underdone.

JMA had the low too far east at T+144 with a cyclonic SSW flow, but not too bad. T+120 had the main low too far NW with a strong anticyclone just off SE England, and then at T+96 the low was way too far west, with slack westerlies over the British Isles. Overall only 6/10.

GEM was quite good at T+144 with low only marginally too far west and WSW winds- but at T+120 and T+96 the low was way out in the Atlantic, and west or even NW winds shown over most of Britain. Only a 4/10.

Overall, an excellent run from the ECMWF, with GFS best of the rest.

[Thundery wintry showers]

19 September

Unfortunately I don't have an image for this day, but it suffices to say that there was a brisk westerly flow over Britain, 988mb low north of Scotland, deep 984mb low out in the Atlantic (containing remnants of "Gordon") and a strong Euro/Bartlett High to the SE.

The GFS was ridiculous in its overdeepening of Gordon at T+144- at 968mb, with strong to gale force SW winds sweeping into the west, and slack westerlies for the east, though it had everything positioned close to correctly. T+120 was poor, with a deep low over NW Scotland and cyclonic westerlies, and the low in the Atlantic too far west. T+96 was okay, but the high to the south was overdone. Overall only 6/10.

UKMO was reasonable at T+144, but like GFS vastly overdid the low to the west, to 965mb. T+120 was poor, with a low off NW Scotland and Gordon way out in the Atlantic, cyclonic SW winds shown. At T+96 it was quite good, though it underdid the low to the west a bit. Overall, 6/10.

The ECMWF was good at T+144, but at T+120 it had a very deep low off NW Scotland and cyclonic WSW gales- similar to GFS at T+120- and the Gordon low way out to the west. However the T+96 was pretty good. Overall this scores 7/10.

NOGAPS available only at T+144 and T+120. At T+144 it was especially poor with slack low over Britain, at T+120 it got the westerlies right, but underdid the low to the NW- only 4/10.

Quite a poor set of outputs from the JMA- it had a large anticyclone over the south at T+144, and at T+120 and T+96 gave us a SW flow, with the Gordon low underdone, and also the low to the N underdone, so only 4/10.

GEM was way off at all three timeframes. At T+144 and also T+96 it had a low off NW Scotland, a cyclonic SW flow and no sign of Gordon; at T+120 it was similar but with the low off NW Scotland almost over the top of Britain. The low actually ended up well to the north of Scotland, so only a 2/10.

Overall, some dodgy T+120 outputs, but in general ECMWF, GFS and UKMO were along the right lines.

[Thundery wintry showers]

20 September

970mb low to the west of Scotland, containing the remnants of Hurricane Gordon. A strong SSW flow over the British Isles, with a lot of cloud and warmth generally.

Only T+144 and T+120 available this time.

GFS at T+144 was remarkably good- low correctly positioned and of correct intensity. At T+120 it was spot on. In both cases strong SSWs were shown over Britain. A very good run from the GFS- I award it 9/10.

UKMO was so-so at T+144, with the ex-Gordon low too far west, and a moderate southerly shown- high pressure too close to the SE (seems to be a common trait of the UKMO). T+120 had the reverse problem with the low too far east and elongated with a strong SSW flow over Britain, though it was pretty close to the mark overall. A score of 7/10.

ECMWF was poor at T+144- a low over N Scotland and westerlies. However T+120 was very good indeed. I award this only 6/10 though, because of the poor T+144.

NOGAPS available only at T+144, which had the low too far to the NE, WSW winds over Britain.

JMA had the low too far east at T+144 with cyclonic SW gales over Britain- it wasn't much better at T+120 with the low still too far east and strong southerly gales, only 4/10.

GEM was very poor at both timeframes- elongated low to the N and NW at T+144 with WSW winds, then at T+120 it had us in a slack westerly. Only 3/10.

Overall, the GFS stood out as being particularly impressive.

[Thundery wintry showers]

21 September

A very warm southerly flow sucked up from very low latitudes. For most, the day was sunny and dry, but there was some rain in the far N & W. Mid-Atlantic low (remnants of Gordon) at 980mb, secondary low off SW England at 988mb.

T+120 outputs were not available.

GFS had the ex-Gordon low in more or less the right place at both T+144 and T+96 but failed to pick out the secondary low. As a result it had a cyclonic showery-looking SW flow over us in both cases, and at T+96 the low was elongated to the east giving us more of a westerly flow. Only a 5/10.

UKMO was poor at T+144 showing us in a showery westerly flow and low pressure to the N, but at T+96 it was pretty good- it even had the secondary low in, so I mark this up to 7/10.

ECMWF was also poor at T+144 with a cyclonic SW flow and some showers, but at T+96 was very good, so also scores a 7/10.

NOGAPS available only at T+144- and it was very poor, with cyclonic westerlies over Britain.

JMA was quite poor at T+144- the low too far east, with W winds pushing into the west, but southerlies in the east. T+96 was again good though, so this scores 7/10.

GEM was poor at T+144- more so than UKMO and ECMWF and JMA- with a westerly over Britain, and it looked nothing like the right pattern. T+96 had us in a cyclonic SSW flow- along the right lines, but lacking detail. I give this a 5/10.

Overall, decent runs by the UKMO, ECMWF and JMA. GFS was a bit disappointing.

[Dawlish]

21 September

A very warm southerly flow sucked up from very low latitudes. For most, the day was sunny and dry, but there was some rain in the far N & W. Mid-Atlantic low (remnants of Gordon) at 980mb, secondary low off SW England at 988mb.

T+120 outputs were not available.

GFS had the ex-Gordon low in more or less the right place at both T+144 and T+96 but failed to pick out the secondary low. As a result it had a cyclonic showery-looking SW flow over us in both cases, and at T+96 the low was elongated to the east giving us more of a westerly flow. Only a 5/10.

UKMO was poor at T+144 showing us in a showery westerly flow and low pressure to the N, but at T+96 it was pretty good- it even had the secondary low in, so I mark this up to 7/10.

ECMWF was also poor at T+144 with a cyclonic SW flow and some showers, but at T+96 was very good, so also scores a 7/10.

NOGAPS available only at T+144- and it was very poor, with cyclonic westerlies over Britain.

JMA was quite poor at T+144- the low too far east, with W winds pushing into the west, but southerlies in the east. T+96 was again good though, so this scores 7/10.

GEM was poor at T+144- more so than UKMO and ECMWF and JMA- with a westerly over Britain, and it looked nothing like the right pattern. T+96 had us in a cyclonic SSW flow- along the right lines, but lacking detail. I give this a 5/10.

Overall, decent runs by the UKMO, ECMWF and JMA. GFS was a bit disappointing.

Well done for keeping this up TWS. I do find it interesting, even though I don't use any other models, apart from the gfs, at present (bear of very little brain and all that!). Do you have a summary of what you think is the best model so far, based on your data collection and judgements? I think I've asked before and you led me to it, but I've forgotten what you said (brain, bear, same kind of thing and all that).

Thanks, Paul

[Thundery wintry showers]

22 September

Southerlies over Britain, 984 and 988mb lows to the west of Britain. T+144 outputs not available.

The GFS was pretty good- at T+120 the low was slightly underdone to the west, giving slack southerlies over Britain, but not a bad effort, and T+96 was spot on, so the run scores 8/10 overall.

UKMO wasn't as good as GFS- at T+120 the low pressure was to the N of western Scotland giving strong WSW winds, while at T+96 it was similar to the GFS's T+120- an underdone low to the west and slack southerlies. Only 5/10.

ECMWF was also disappointing- strong to gale force SSWs shown at T+120, with overdone low off NW Scotland. A similar pattern was still there at T+96, though it was closer to being correct- lowest pressure centred too far north. Overall 6/10.

NOGAPS available only at T+120- and quite poor, low way too far east, almost over the top of Britain.

JMA was very good at T+120- little to fault- but at T+96 fell into the ECMWF trap of having lowest pressure too far north and more of a cyclonic S/SW influence. Overall 7/10.

GEM had lowest pressure N of Scotland and westerlies at T+120, and cyclonic SWs at T+96, low still too far east. Only a 4/10.

Overall, a good output by GFS, and reasonable by JMA, other models struggled.

Re. Paul's comment- in recent months the ECMWF has been scoring highest, with GFS a very close second, and the UKMO third.

[Thundery wintry showers]

23 September

1004mb low off NE Scotland, 996mb low to the west, 1024mb Euro High, slack SSE winds over Britain.

GFS was superb for 23 September. At T+144 and T+120 it had the SSE flow in place- slight errors in positioning of low NE of Scotland, but 996mb low always correctly placed to west, and 1024mb high to east. At T+96, also very good. A score of 9/10.

UKMO good at T+144- lows to west underdone, but the southerly flow and high to east intact, and very good at T+120. T+96 was then a letdown- low not only overdeepened but a good 300 miles to the SW of where it should have been, with slack pressure over Britain, southerlies only for the SW. Overall 7/10.

ECMWF was poor at T+144, with a strong cyclonic WSW flow shown over Britain. It was very good at T+120 and spot on at T+96, though, so scores 7/10.

NOGAPS was reasonable- lowest pressure a bit too far north at T+144 with slack southerlies shown; low a bit too far west at T+120 but with Britain correctly in slack SSE winds. Overall a 7/10.

JMA was very good at T+144- little to fault in the output- but T+120 had the low pressure centred well down to the SW, with easterly winds covering most of Britain. T+96 was worse, with a Scandinavian High extending towards Scotland, and ESE winds, so overall only 4/10 despite the good T+144 output.

GEM had slack westerlies at T+144- nothing resembling the correct output. No low pressure to the west; the only low pressure nearby was centred north of Scotland. T+120 was better with a low to the SW- lacking detail, but showed moderate SSE winds over britain. Low was too far east at T+96, however- almost on top of Britain- with cyclonic southerlies in the eastern half of the British Isles. Overall 5/10.

The GFS was the most accurate of the models for the second day in a row.

[Thundery wintry showers]

24 September

Low at 1000mb off SW England and slack easterlies in the north, slack southerlies in the south.

GFS disappointing at T+144- lowest pressure too far NW, with southerlies across Britain. T+120 was very good though- low just 100 miles too far south, with slack ESE winds for most, southerlies in south, then a very accurate T+96. Overall a rating of 7/10.

UKMO at T+144 had high pressure dominating over Scandinavia and slack ESE winds for all, low pressure well away to the SW. T+120 had high pressure dominating over N Scotland- again low pressure way too far south. T+96 was reasonably accurate though, so the run scores 5/10.

ECMWF was similar to GFS at T+144- low too far north, though less so than for GFS, with southerlies. T+120 was very accurate but T+96 had the low too far SE with easterlies for most, so also 7/10.

NOGAPS available at T+144 and T+96- T+144 had southerlies and an elongated high to the E dominating our weather, and at T+96 SE winds and low pressure too far away to the SW. Only 4/10.

JMA was decent at all three timeframes- T+144 had the low about 200 miles too far SW, T+120 was accurate, but at T+96, similar to ECMWF, had the low too far east- error more pronounced with moderate easterlies for all. Another 7/10 run.

GEM was quite accurate at T+144 and T+96, though at T+96 low was too far east, giving easterlies for all. At T+120 it was well out with a slack low over Britain- overall 6/10.

[Dawlish]

22 September

Re. Paul's comment- in recent months the ECMWF has been scoring highest, with GFS a very close second, and the UKMO third.

Thanks TWS. do you keep any running comparisons, or scores, that you can attach? (Sorry - just found them (scrolling back first really helps! B) B) B) ). Blooming well done!

Paul

Edited September 29, 2006 by Dawlish

[Thundery wintry showers]

25 September

I don't have the chart for 12Z, but there was a 1010mb low over central and northern Britain, slack westerlies in the south, and an advancing ridge from the Azores High.

GFS was decent at both T+144 and T+120, but had the slack low a bit too far to the SW, giving slack southerly winds for most. At T+96 it had the low to the south-east of Britain giving easterlies. I only award this a 6/10.

UKMO had no slack low at T+144- rather we were on the eastern flank of a large low with strong southerlies. T+120 had us in a slack SW flow with high pressure in charge over the SE, while T+96 had strong low pressure over Britain and strong westerlies over the south. A real mish-mash that only scores 3/10.

ECMWF had a slack high pressure regime over us at both T+144 and T+120- low presumably underdone- but it was very accurate at T+96, by far the best output of the "big three". However because of the mediocre first two outputs, this scores 7/10.

NOGAPS was decent at both T+144 and T+96 with the low a bit too far west giving southerlies, much like the GFS at T+144 and T+120. However, at T+120 it had us in a SE'ly flow and with low pressure well away to the SW. This scores a 6/10.

JMA had it wrong at all three timeframes- T+144 had low pressure to the SW and SE winds, while T+120 and T+96 had lowest pressure to the SE and ENE winds, so only 3/10.

GEM had a slack high over us, slack N winds in the east at T+144, at T+120 and T+96 it was okay, but had the slack low pressure too far south, giving easterlies for most. 5/10.

Overall, nothing particularly great from the models; ECMWF was the nearest.

[Thundery wintry showers]

26 September

1020mb high to the south, slack westerlies for much of Britain.

GFS had slack high pressure to the east at T+144 and southerlies, and slack high pressure over Britain at T+120, but was then very accurate at T+96. I award this a 5/10 as the T+144 and T+120 did not resemble the correct outcome.

UKMO was poor for the second day in a row. T+144 and T+96 both had high pressure right over Britain, while T+120 had a very deep low over Scotland and strong westerlies further south. Overall 3/10.

ECMWF was also poor, showing a high over Britain at T+144, a strong low off NW Scotland with strong cyclonic SW winds for most at T+120, and then a moderately deep low just north of Scotland at T+96 with strong westerlies over Britain. Overall also just 3/10- no outputs resembled the correct outcome.

NOGAPS had high pressure to the East and southerlies at T+144, a deep low to the west and cyclonic southerlies at T+120, and then a Scandinavian High and slack easterlies at T+96. A very poor set of outputs, even worse than UKMO and ECMWF, and scores 2/10.

JMA was also nowhere near correct at T+144 (high to N and W, NNE winds) and T+120 (high over Britain). However, it was very accurate at T+96, so scores 5/10 overall.

GEM was poor at all three timeframes- high pressure to the NE on all three occasions with SE winds shown over the British Isles, more southerly in the case of T+120, and low pressure to the W and SW. Only a 2/10, it was consistent with its outcome but it was the wrong one.

UKMO and ECMWF were uncharacteristically poor, and while GFS and JMA weren't great, they were still the most accurate models.