12z Model Comparisons Results

[Thundery wintry showers]

27 September

A more straightforward picture, 988mb low just off western Ireland and strong cyclonic southerly winds for the British Isles, weak 1016mb high to the east.

GFS was poor at T+144- high pressure over the SE and only slack south-westerlies shown, the deep low nowhere to be seen. T+120 was reasonable, the low slightly underdone but accurately placed, strong SW winds shown. T+96 was then poor- the low shown too far east, with a showery westerly over Britain. Overall 5/10.

UKMO was disappointing at T+144 with a pronounced westerly flow, the main low too far east. However, T+120 was very accurate. At T+96 the low to the west was slightly too far north and elongated giving a broad WSW flow over Britain. Slightly better than the GFS so I rate it at 6/10.

ECMWF wasn't too bad at T+144, the low displaced too far east giving us cyclonic W/SW winds, but it was quite close to being correct. However, T+120 and T+96 were very poor with elongated low pressure to the north and a straight flow of westerlies. Only 4/10, as the latter two outputs were nowhere near correct.

NOGAPS was okay at T+144- deep low a bit too far east giving westerlies, but close to being correct, but then T+120 and T+96 were poor- no low in sight, slack southerlies at T+120 and SWs at T+96. Overall 4/10.

JMA had a high right over Britain at T+144, but at T+120 and T+96 wasn't too far away- low to W underdone and placed a little too far north, giving a pronounced SW flow. Overall 6/10.

GEM consistently had a deep low to the west, but it was too far west at T+144 and T+96, with slack southerlies over Britain. However T+96 was better, the low only slightly too far west, and cyclonic southerlies, so it scores 6/10.

A rather poor set of runs again- UKMO, JMA and GEM having the highest scores. I will update the table of results and post them in the next couple of days.

[Thundery wintry showers]

28th September

Low at 976mb in the Atlantic with an elongated trough to the north of N Scotland, WSW winds.

GFS had the general pattern right at all three timeframes. T+96 was slightly in error as it had a secondary low over southern England, but otherwise all three timeframes were accurate with an elongated low to the north and WSW winds. I give it 8/10.

UKMO unavailable at T+144. At T+120 it was poor, with a deep low over NE Scotland and cyclonic WNW winds, and at T+96 had an overdeepened low off NW Scotland and strong westerly winds, so only scores 5/10.

ECMWF also had problems- it had a high over southern England at T+144 and anticyclonic W/NW winds, at T+120 it had a low to the NE and WNW winds, orientation of westerly flow wrong, but at T+96 it was reasonably accurate though had the deep 976mb low a good 300 miles too far east, giving a more cyclonic flow. Only 6/10.

NOGAPS poor at T+144- high over N Scotland, easterlies in the south, it was reasonable at T+120, but then at T+96 it had the deep low to the NE and a secondary low over eastern England. Only 5/10.

JMA was disappointing- a northerly flow shown at T+144, at T+120 it was closer with a slack westerly and low pressure to the north, but at T+96 had high pressure to the east and southerlies. Only 4/10.

GEM was also poor- high pressure and southerlies shown at T+144, at T+120 low pressure was in charge but too far south giving cyclonic southerlies. At T+96 it was better, with a low off N Scotland and westerlies, and the secondary low correctly placed, so it scores 5/10 because of the reasonable T+96.

The GFS was by far the most accurate at predicting the 28 September synoptic situation.

[Thundery wintry showers]

29 September 2006

I don't have the chart for 29 September at 12Z, but I can easily extrapolate based on the 00Z charts for 29 and 30 September. Essentially, a 980mb low out in the Atlantic, 1000mb troughs over north and west Scotland giving some cloud and showers, and a moderate south/south-westerly flow.

I also don't have any T+96 outputs for 29 September, but T+120 and T+144 are available.

GFS was quite disappointing at T+144 with elongated low pressure to the north and a straight westerly flow, but at T+120 it was closer, having a deep low to the W and a slack WSW flow; deep low too far west, but quite close to what actually happened, so it scores a 6/10.

UKMO was reasonable- it had the low to the west underdone and a bit too far north at both T+144 and T+120, but had a moderate SW flow and troughs over the north and west at both timeframes. I rate it at 7/10.

ECMWF was similar to GFS at T+144- strong high pressure to the south and more of an anticyclonic westerly flow, not particularly good. At T+120 it had the low to the W a bit too far east, but gave Britain SSW winds and a cyclonic/showery outlook, so it scores 6/10.

NOGAPS poor at T+144, high pressure in charge, slack westerlies. T+120 was better, but had a westerly flow, the Atlantic low too far north and grossly underdone. I rate this at only 4/10.

JMA disappointing at T+144 with a straight westerly shown. At T+120 it was better, the low correctly placed but another low overdone just north of Scotland, and high pressure too close to the SE, but with a moderate showery SW shown. I give this a 6/10.

GEM, after recent performances, was surprisingly good. In both cases the deep Atlantic low was a bit too far east, giving cyclonic SSW winds for all, and stronger winds than actually happened, but the general pattern was correct, so it scores 7/10.

Overall nothing outstanding, but UKMO and GEM were closest, showing something approximating to the correct outcome at both T+144 and also T+120.

[Thundery wintry showers]

30 September 2006

Little change from the previous day; 988mb low off SW England, and a stronger southerly flow of cyclonic curvature establishing over the British Isles.

No T+120 outputs, unfortunately. I didn't get to save the charts to my hard drive on 25 September.

GFS was superb at T+144- low slightly underdone, but shown propelling cyclonic southerly winds into Britain and correctly placed. However T+96 wasn't so good- it had lows to the north and west and a westerly flow veering southerly, so only a 6/10 because of the poorer T+96.

UKMO had us in SSW winds at T+144, but high pressure dominant to the SE and the low to the west massively underdone- it was only shown at 1005mb, so a rather poor output. However, the model was very accurate at T+96, so also scores 6/10.

ECMWF was reasonable at T+144, the low slightly too far east, giving very strong southerlies, but overall pattern was correct. T+96 was spot on, so the ECMWF scores 8/10.

NOGAPS wasn't bad. Low too far east at both timeframes, but southerlies shown over Britain, everything shifted a bit too far east, but the pattern was correct. I rate this at 7/10.

JMA was unavailable at T+144 so I can't give it a mark; T+96 was quite good though.

GEM had the low a bit too far west at T+144, giving lighter southerlies than actually happened, though on the whole a good output. At T+96 the low was too far east and underdone- indeed the chart resembled the actual outcome for 1 October, not 30 September. Overall 7/10.

The ECMWF was the most accurate this time around.

[Thundery wintry showers]

Results up until 30 September. Overall a change in assessment of how the models handle southerlies; until September, the general trend was for UKMO and ECMWF to be the most accurate and GFS to have problems, but in September the GFS handled the southerlies quite well, and UKMO did rather poorly.

Overall, the dominance of the GFS and ECMWF continues with UKMO third.

model_comparisons.doc

model_comparisons.xls

[Thundery wintry showers]

1 October 2006

996mb low over Ireland, SSW winds over most of the rest of Britain.

Just T+96 and T+120 available.

GFS was okay at both timeframes- low a bit too far west both times (more especially at T+120) giving a southerly flow for all. However, the general pattern was right, so 8/10.

UKMO was good at T+120, and also good at T+96, scoring a good 9/10.

ECMWF was also very accurate at both timeframes.

NOGAPS was also accurate at T+120, but at T+96 had the low over northern Scotland, giving westerlies. Overall a 7/10.

JMA was also good, though it tended to overdeepen the low, overall an 8/10, the low was correctly positioned at both timeframes.

GEM had the low significantly too far west at T+120 with general southerlies, but was accurate at T+96, I give this a 7/10.

Overall a very good set of runs.

[Thundery wintry showers]

2 October 2006

996mb low off NE Scotland, NW winds of cyclonic curvature over Britain.

GFS had the low too far west at all three timeframes. T+144 had southerlies over most of Britain, then T+120 and T+96 had a deep low over the centre of Britain, giving strong cyclonic westerlies in the south, easterlies for N Scotland. I only give this a 5/10.

UKMO wasn't bad at T+144, had the low a bit too far south giving cyclonic NW/N winds, then at T+120 and T+96 had the low about 100 miles too far west, but giving a cyclonic W/NW flow. I rate this at 7/10.

ECMWF had the low too far west at T+144 giving cyclonic westerlies, but it was very accurate at T+120 and T+96, almost perfect on both occasions, and scores 8/10.

NOGAPS had the low too far to the NE at T+144, giving moderate northerlies, but it was very accurate at T+120. At T+96 the low was positioned too far west, giving cyclonic westerlies. I give this a 7/10, it wasn't too bad.

JMA committed the same error at all three timeframes, positioning the low about 200 miles too far west, giving westerlies over Britain. Other than that, it had everything else correct, so scores 6/10.

GEM had the low way too far west at T+144, giving southerlies. At T+120 it had the low still too far west, but giving cyclonic WSW winds. At T+96 it was okay, similar to UKMO, low a bit too far west but giving cyclonic W/NW winds. Overall 5/10.

GFS had problems this time, ECMWF was the most accurate of the models, giving very accurate outputs at T+96 and T+120.

[Thundery wintry showers]

3 October 2006

Once again I missed 29 September, so T+120 and T+144 available.

Remarkably little change since 2 October, a 996mb low off NE Scotland, NNW winds covering the British Isles.

GFS still didn't have the pattern nailed down, with a deep low over Britain at T+144, and then to the east at T+120, giving north-easterlies, but at least it was fairly close. Overall, 6/10.

UKMO had us in a westerly at T+144, the low to the NE correctly placed but it had another low off NW Scotland, hence the westerlies. At T+120 it was accurate, so scores 7/10.

ECMWF kept positioning the low too far east. At T+144 it had Britain under high pressure, and at T+120 it had Britain on the western fringe of a northerly flow, high pressure in the west. In contrast to yesterday's run, the ECMWF for 3 October scores only 5/10.

NOGAPS was similar to GFS- it kept putting the low too far west, over the north-eastern quadrant of Britain. It was fairly close in other areas though, so I rate it at 6/10.

At T+144 the JMA was fairly accurate, low just 100 miles too far west and NW winds, but at T+120 went for a westerly flow over Britain, with another low off NW Scotland, so scores 6/10.

The GEM had the low too far west at T+144 giving westerlies, but was very accurate at T+120- if anything the low was a little to the east of where it turned out. I give this a 7/10.

Overall the models still had problems, with ECMWF having the most trouble; UKMO and GEM were the only two to provide very accurate T+120s.

[Thundery wintry showers]

4 October 2006

WNW winds over Britain, slack low pressure to the north, 1028mb high off the west of Spain.

I don't have the outputs for T+120.

GFS: T+144 was quite good although the slack lows were a bit too far south, situated over northern Scotland, high roughly accurately placed. However at T+96 the slack lows to the north were missing, leaving the deep Scandinavian low and northerlies. A score of 6/10.

UKMO was very good at T+144, the low out in the Atlantic bringing a change to more wet/windy weather on 6 October was underdone and a bit too far east, but everything else was good, WNW winds shown. However at T+96 it committed the same error as the GFS, giving us northerlies. 6/10.

ECMWF was much as GFS- good at T+144 but with lows too far south, at T+96 it had northerlies over Britain, no weak low to the north. 6/10.

NOGAPS slightly worse- at T+144 it had a low right over Britain with slack winds, and at T+96 it had northerlies, much as the others. 5/10.

JMA had a deep low over Scotland at T+144, with westerlies in the south. At T+96, it had a NW flow, with some evidence of the slack low to the north, but also a secondary low to the south that failed to materialise. Overall 6/10.

GEM was quite good at T+144, with slack lows to the N and westerlies- as with UKMO, the low well out in the Atlantic positioned a bit too close. However it was out again at T+96, with northerlies. 6/10.

All models failed to pick up on the shallow low to the north at T+96, so we ended up with westerlies instead of the projected northerlies- watch out for this in future, it reminds me of those northerly downgrade situations.

[Thundery wintry showers]

5 October 2006

T+144 unavailable. A 992mb low off western Scotland and strong SW winds over Britain, 1028mb high just off the west of Spain.

GFS had the low a bit too far west at T+120, so although we were in SW winds they were slacker and the Atlantic systems further west than actually happened. At T+96 it was pretty good, the low still about 100 miles too far west, but with strong SW winds over Britain. I award this 7/10.

UKMO was similar to GFS at T+120, with the low about 200 miles too far west and moderate SWs. However, at T+96 it had the low way out in the western Atlantic with slack WNW winds over Britain and high pressure to the south with low pressure to the NE. Only 5/10.

ECMWF also had the progression of the Atlantic low underdone, positioning it too far west with Britain under a ridge of high pressure at both T+120 and T+96, though with the SWs starting to push into the west at T+96. Also only 5/10.

NOGAPS was very poor at T+120 with slack NW winds and no sign of that Atlantic low, a high to our NW in fact. However at T+96 it was quite accurate, if anything the low about 100 miles too far east, with strong SSW winds. I give this 6/10.

JMA was quite good. At T+120 the low was underdone and a bit too far south giving SSW winds, of slack nature, over the British Isles, but it wasn't bad; then at T+96 it was very accurate. Like the GFS this scores 7/10.

GEM had the low too far south at T+120, with southerly winds over Britain. At T+96 it had the low too far west, with a slack ridge of high pressure over Britain, south-westerlies coming into the west, so scores 5/10.

Overall the GFS and JMA- the two most progressive models- were pretty good at timing the arrival of the low pressure, but the Euro models underestimated the speed of approach and had the low too far out in the Atlantic.

[Thundery wintry showers]

6 October 2006

All three timeframes available. A very progressive looking chart, 988mb low over N Scotland and strong westerlies. With ex-tropical air in the circulation it was quite warm.

The GFS was very impressive indeed. Although there were minor quibbles about low orientation and central pressure, the low was positioned correctly at all three timeframes, the pattern in the Atlantic was also spot on, and strong westerlies were shown over Britain. Not absolutely perfect but certainly a good 9/10.

UKMO poor at T+144- slack low off East Anglia, high pressure to the SW and northerly winds. T+120 was getting onto the right track, but much as yesterday, underestimated the speed of the low- it was still out in the Atlantic giving a warm sector and SSW winds and overdone high pressure to the south. T+96 was quite accurate though, the low's orientation a bit elongated but otherwise a very good output. However this only scores a 5/10, as both T+144 and T+120 were poor.

ECMWF was also good at all three timeframes, though not quite as good as the GFS. T+144 had the low significantly underdone (only 1005mb central pressure), and at T+120 and T+96 it was elongated giving a long-draw westerly rather than a cyclonic one. However the low was correctly placed at all three timeframes. I give this 8/10.

NOGAPS had a low to the East at T+144 giving northerlies. At T+120 it had a low off NE Scotland, giving NW winds. At T+96 it had a low over Shetland giving cyclonic WNW winds. I can only guess that it was a case of the main low being shifted too far east, as there was no evidence of a further low out in the Atlantic. In any case, only a 4/10, as only T+96 was close to being correct.

JMA had the low too far east at T+144 with northerlies for all. At T+120 it had the low too far east again, but only by about 200 miles, so northerlies for the west and westerlies for the east. Reasonable at T+96 but with the low about 200 miles too far west this time, giving WSW winds. Overall a 5/10.

Low over southern Britain at T+144 on the GEM with easterlies elsewhere. At T+120, low underdone and way out in the Atlantic with a warm sector and WSW winds over Britain. At T+96 it was close to being accurate though the low was about 200 miles too far west, giving WSW winds. Only 4/10, as the T+144 and T+120 were worse than the JMA.

Overall, it seems that all models bar GFS and ECMWF had big problems with placing the low correctly, but on the other hand, the GFS and to a lesser extent ECMWF were outstandingly good.

[Thundery wintry showers]

7 October 2006

A strong anticyclonic WNW flow over the British Isles, 1020mb high to the south, 984mb Scandinavian Low, 980mb low SW of Iceland.

GFS was okay at T+144 with the high to the south overdone and the Icelandic Low too far north giving higher pressure over Britain, but an anticyclonic westerly flow. T+120 was poor with a showery NNW flow and both lows too far west. T+96 was then accurate. Overall a 6/10.

UKMO's best output, unusually, was T+144 with the WNW flow slacker than it should have been because the lows to the north were underdone, but they were correctly placed as was the high to the south. However T+120 had the two lows too far west, and the same was true at T+96, with high pressure to the SW and NW winds. Overall I give this a 6/10 as well.

ECMWF had us under NW winds at T+144 with the two lows too far west. At T+120 the lows were overdeepened but correctly placed, with Britain under a strong W flow and the high to the south too far south. At T+96 it was accurate. Overall slightly better than GFS and UKMO so scores 7/10.

NOGAPS kept having the Scandinavian Low too far to the NW, and the Atlantic low roughly correctly placed. At T+144 and T+120 high pressure was too dominant over southern Britain with cyclonic westerly winds in the north, then at T+96 it was a very progressive picture, elongated low pressure to the north and strong WSW winds. I give this 6/10.

JMA was poor at T+144 with high pressure right over Britain. At T+120 it was very accurate, but then at T+96 it had the Scandinavian Low way too far west, giving a cyclonic NW flow, so only 5/10 for this one.

GEM consistently overdid the Atlantic low and failed to pick up on the Scandinavian Low. At T+144 and T+120 there was a strong WSW flow over Britain around the Atlantic low, and at T+96 an elongated low to the north giving westerlies. It had westerlies at all three timeframes but low positioning was always wrong, thus 5/10.

Nothing great from the models for 7 October, ECMWF being marginally the most accurate.

[Thundery wintry showers]

8 October

1020mb high off south-east England. 976mb low just SW of Iceland. A secondary low at 1000mb suggested to the west of Ireland, with south/SSW winds.

GFS had everything a bit too far west at T+144, so HP to the south, westerlies, low well to the W of Iceland. T+120 was also dodgy with SW winds over Britain and an elongated low to the north, though the high was correctly placed and southerlies covered with south, but T+96 was spot on. I rate this at 7/10 as it wasn't too bad at T+144 and T+120 and good at T+96.

UKMO had the low correctly placed at T+144 but underdone, high pressure to the south (i.e. a good 300 miles too far west) and westerlies, overall about the same as GFS T+144. T+120 and T+96 were both reasonable, but had the Icelandic Low too far west and high pressure too far NW, giving SE England high pressure and other areas an anticyclonic S/SW flow. I also give this 7/10.

ECMWF had a strong W flow at T+144 with low pressure to the N and NE. It was good at T+120 and T+96, but elongated the Icelandic Low so gave us W/SW winds mainly, similar to the GFS T+120 but better in that lowest pressure was centred south of Iceland. Another 7/10, as T+120 was better than the GFS, but at T+96 the GFS was better.

NOGAPS had the low too far east at T+144 giving strong SSW winds, HP to the south (too far west). T+120 and T+96 both had the Icelandic Low too far west, giving anticyclonic SW winds. Overall a 6/10.

JMA had high pressure too dominant at T+144 and more especially T+120 and the low underdone and too far west- indeed at T+120 high pressure was right over Britain. At T+96 it was spot on, but because of the poor T+144 and T+120, scores only 6/10.

GEM was dodgy at T+144 and T+120, the Icelandic Low elongated and giving a strong SW flow over the British Isles, then T+96 had the low positioned correctly, but a large low to the NE overdone, giving northerlies into northern Scotland, so only a 5/10.

Fairly consistent if unspectacular outputs from the models; a tendency to elongate the Icelandic Low giving SW winds rather than S winds, and to position the Euro High too far north and west particularly in the case of JMA.

[johnholmes]

to Ian

Many thanks for continuing with this Ian. Its fascinating reading the results.

As you know I'm unable to access Excel documents on the web. Would you be knid enough please to send me a copy in Word for me?

many thanks

John

[Stuart]

Thanks for that update

[Thundery wintry showers]

9 October

No chart available, but pressure was low at 980-985mb west of Ireland and we had a slack W flow over Britain and high pressure at 1024 to the S and E; a front just having moved through.

GFS is another 7/10- the T+144 output was only okay, a long draw of WSW winds and elongated LP to the north, high correctly placed to the east. However at T+120 and more especially T+96 a ridge of high pressure to the south was shown, with strong HP remaining to the east, and lows in the correct places.

UKMO wasn't bad at T+144, though as is usual for UKMO, high pressure was overdone to the east, giving more of a SW flow and high pressure over the south. T+120 had the general pattern in place but a low over the British Isles giving slack winds. T+96 was accurate. Also 7/10.

ECMWF had a cyclonic westerly at T+144 with low pressure in charge. At T+120 it reverted to the long-draw WSW flow shown by GFS T+144. At T+96 it was pretty accurate. Overall only 6/10, as it seemed further out than GFS and UKMO.

NOGAPS was poor at T+144, showing southerlies. At T+120 it was better, more of a cyclonic output than actually happened but with WSW winds and HP to the south and east. At T+96 it was pretty accurate. Also a 7/10.

JMA had southerlies at T+144, an elongated WSW-type flow at T+120 with the lows and highs in the right places, then was very accurate at T+96. It scores 7/10 as well.

GEM was poorer, at T+144 it had low pressure close by, and northerlies spreading into N Scotland, southerlies elsewhere. Cyclonic SW regime at T+120, better at T+96 but showing an elongated WSW flow. Only 5/10.

Overall, fairly consistent set of outputs, generally poor at T+144 but good by T+96. The GEM was rather behind the others in terms of accuracy.

[johnholmes]

to Ian

Many thanks for continuing with this Ian. Its fascinating reading the results.

As you know I'm unable to access Excel documents on the web. Would you be knid enough please to send me a copy in Word for me?

many thanks

John

[Thundery wintry showers]

I sent an email yesterday to the address you gave me the last time- has it not got through?

[johnholmes]

I sent an email yesterday to the address you gave me the last time- has it not got through?

hi Ian

sorry you are one I have not given my new e mail to, will send a pm

thanks

John

[Thundery wintry showers]

10 October

Large Scandinavian/Euro High at 1024mb and 988mb low out in the mid Atlantic. We had light southerlies over the British Isles.

GFS had a dodgy run, with high pressure on top of Britain at both T+144 and T+120, but it latched onto the correct pattern at T+96. Overall the T+144 and T+120 were a bit of a mess in terms of how everything was positioned, but the outcome over Britain wasn't too far off (high too far west, essentially). I give it 6/10.

UKMO was worse than GFS at T+144 having the high pressure to the SW of Britain, and anticyclonic WNW winds. However overall I thought it was slightly better than GFS because T+120 and T+96 were both very accurate, and so it scores 7/10.

ECMWF had high pressure too far east at T+144, low pressure off NW Scotland and SW winds, but T+120 and T+96 were very accurate. This also scores a 7/10 because of the poor T+144 which was well out, but again I think it outscored GFS for latching onto the correct pattern 24 hours earlier.

NOGAPS was poor at T+144 with high pressure over the south, westerlies in the north. At T+120 it was reasonable; the high a bit too far west, giving anticyclonic conditions in the east, southerlies in the west; at T+96 it was very accurate. Overall a 6/10.

JMA was very good indeed- it had southerlies over Britain and a strong Euro High at all three timeframes, though at T+144 the low to the west was underdone, at around 1000mb intensity. T+96 was slightly out as it had a weak low to the SW that shouldn't have been there. Overall, though, an 8/10 for consistently getting it more or less right.

GEM was dodgy at T+144- southerlies, but low pressure over Ireland, with a cyclonic bias to the weather, high pressure way too far east. T+120 had high pressure over the SE and SW winds for most. At T+96 it was pretty accurate, so the run as a whole scores 6/10.

Again not a great deal to choose between the models. JMA was the only model to get the pattern correct at all three timeframes, UKMO and ECMWF fell into line by T+120, and the others fell into line by T+96.

[Thundery wintry showers]

11 October

972mb low to the SW of Iceland, a weak 1008mb low over Britain and moderate southerlies, WSW winds spreading into the west.

GFS was pretty good, though at T+144 and to a lesser extent T+96 had the continental high too far west, and a general southerly flow, the low pressure and fronts probably about 100-200 miles to the west of where they ended up. T+96 was good. Overall a reasonable 8/10.

UKMO's T+144 wasn't too bad with general southerlies, though it had a well-defined low to the SW and underdid the Atlantic low. At T+120 it had everything displaced too far east, and WSW winds. At T+96 it was very good, though still having an overdone low over the south. I give this 7/10.

ECMWF was dodgy at T+144 with a WSW wind, everything too far east, but then it was very accurate at T+120 and also at T+96 and so scores 8/10.

NOGAPS was poor at T+144, the Atlantic low way too east giving very strong cyclonic SW winds over Britain. T+120 was better but still had a whopping 980mb low off W Scotland giving very strong SW winds. At T+96 it went to the other extreme and grossly underdid low pressure, a general southerly flow and weak 1010mb low off W Ireland. Only a 5/10.

JMA was pretty good again- a southerly with westerlies moving into the west was shown at all three timeframes and T+144 and T+96 were both accurate. T+120 was less accurate with low pressure too far away to the NW, and high pressure close to the south. Overall an 8/10.

GEM was poor at T+144- similar to NOGAPS, 980mb low off W Scotland and gale force SW'lys. T+120 was like the NOGAPS at T+96, with low pressure too far west, a general SSE flow, and unlike the earlier GFS outputs didn't have everything else close to the correct outcome. T+96 was good though, so I give this 6/10.

Overall, GFS, ECMWF and JMA were very good, and UKMO wasn't far off either.

[Thundery wintry showers]

Unfortunately I had internet problems over the past few days, so there will be a significant gap in coverage; however, the dates up to 19 October (the time when we were supposed to be getting a northerly) will be coverable.

12 October

A 976mb low SW of Iceland, 1024mb high off SE England and moderate SW winds.

The GFS was poor at T+144 and T+120 with a slack low to the south and the low correctly placed to the NW, with high pressure to the east, giving slack S/SE winds over most of Britain. The low SW of Iceland was correct, other features somewhat wrong. However, T+96 was pretty much perfect. The run scores a 6/10.

UKMO was okay at T+144, the low SW of Iceland too far to the NE situated right over Iceland, but SW winds were correctly shown over Britain with a 1025mb high to the south. At T+120 it was quite poor though, with a strong Scandinavian High and a ridge over to Britain, giving slack easterlies. T+96 was excellent though. Overall a 7/10; I think it slightly outscored GFS overall.

ECMWF was dodgy at both T+144 and T+120 with a slack low over the British Isles and slack westerlies for most, but the low SW of Iceland and HP to the south were correctly identified. T+96 was very accurate. Overall a 6/10- it was about level on accuracy with the GFS.

NOGAPS was good at T+144- slack low over the south spoilt it a bit, but other features were accurate, general SW winds. However at T+120 it committed the same error as UKMO- Scandinavian High, slack easterlies. High pressure was on top of Britain at T+96, so overall only 5/10.

JMA was poor at T+144- similar to GFS, with low SW of Iceland correctly placed, but a slack low to the south and S/SE winds. T+120 was similar to UKMO's offering, with a Scandinavian High and slack easterly winds. T+96 was decent with high pressure too far north, covering SE England, SW winds for the north. Overall, only 5/10.

GEM was poor at T+144- similar to GFS/JMA but worse, with a deep low to the south. T+120 had SW winds over the north-west, and high pressure elsewhere, but then T+96 was very accurate, so a 6/10.

Overall UKMO shaded it, but none of the models were outstanding.

13 October

No T+96 available. There was a strong Scandinavian High at 1036mb, southerlies for Scotland, easterlies further south. Low SW of Iceland weakened, to 992mb.

GFS wasn't great at T+144 with a low to the south, but high pressure correctly shown over Scandinavia, with SE winds shown. At T+120 it was very accurate. I give it another 6/10.

UKMO had the high a bit too far north at T+144, giving a pronounced ENE flow across the whole of Britain, but at T+120 it was very accurate. I think T+144 was slightly better than that of GFS, so it scores 7/10 for the second day in succession.

ECMWF again similar to the previous day- poor at T+144 with a low over Britain, though HP correctly shown over Scandinavia, then accurate at T+120, 6/10.

NOGAPS was reasonable- at T+144 high a bit too far north giving easterlies, at T+120 quite accurate, though with a low too close to the west, giving southerlies for western Britain. I give this a 7/10 as both outputs were good, though not excellent.

JMA had the high a bit too far to the NE at both timeframes, but the general outcome was correct- southerlies in Scotland, easterlies further south, HP in charge, though a more pronounced airstream than actually happened. A score of 7/10.

GEM was reasonable at T+144 with an overdeepened low right over Iceland, but light southerlies over Britain and a Scandinavian High. However at T+120 it was notably poor, an Atlantic low off W Scotland and strong SSW winds for all, elongated HP to the east. Therefore only 5/10.

A reasonable set of outputs, apart from perhaps GEM which was well out at T+120.

[Thundery wintry showers]

14 October

Something seems to be up with the colours, but the chart shown is correct. 1000mb Icelandic Low, 1036mb high to the east, cyclonic/southerly winds. No T+120 available.

GFS was okay at T+144- high a bit too far east, giving general southerlies, then at T+96 very accurate. I give this run 8/10.

UKMO if anything had the high a bit too far west at T+144, giving easterlies for the south. At T+96 it was very accurate, so also scores 8/10.

ECMWF scores 9/10- it was very accurate at both T+144 and T+96, a very impressive output.

NOGAPS had the high way too far north, to the W of Scandinavia, at T+144, giving a pronounced easterly flow for all, but was quite accurate at T+96. I rate this at 6/10.

JMA had the high right over northern Scandinavia at T+144, giving a pronounced ESE flow over Britain, high a long way too far NE. At T+96 it was good though, so it scores 6/10 also.

GEM was erroneous at both timeframes, placing the high over Scandinavia and showing pronounced southerlies over the whole of Britain, low pressure close by west of Ireland; only 5/10.

Overall excellent outputs by the "big three", with ECMWF leading the way.

15 October

Large 1030mb high in the North Sea with light southerlies and an anticyclonic regime over Britain. This time I don't have the T+144s, but the T+120s and T+96s are available.

GFS was very poor indeed at T+120- my first thoughts were "what was it thinking of?" with a low over northern Scotland and moderate westerlies over Britain. At T+96 it was accurate, but because of the wayward T+120 I only award it 5/10.

UKMO scores 7/10, having positioned the high too far north at both timeframes, giving a pronounced easterly flow for the south, and slack southerlies in the north.

ECMWF was spot on at T+120, but at T+96 the high was about 100 miles too far north, giving easterlies in southern Britain, otherwise very accurate, so a score of 8/10.

NOGAPS was accurate at T+120- high maybe a little too far west, with easterlies for the south, but very close to being spot on. T+96 was a bit further off, the high too far west with easterlies for the south, and not centred in the North Sea. Overall, 7/10.

I can't rate JMA as there were Entire Grid Undefineds at T+120. At T+96, for the record, it was pretty accurate but had the high over Scandinavia, with strong southerlies for western Britain.

GEM had our main high too far east, over Scandinavia, with pronounced southerlies for all, at T+120. T+96 was better, but the high was too far west this time, giving easterlies for southern Britain. Not as good as most of the others, so a score of 6/10.

The UKMO and ECMWF seem to be handling this eastern blocking very well, GFS generally good but with a couple of wobbles, like the T+120 for 15 October which was well out. Of the others, NOGAPS seems to be performing quite well.

[Thundery wintry showers]

16 October

No chart available, but we had 1025mb high pressure to the east, and slack southerly winds, 990mb low west of Spain. A wave feature at 1010mb west of Ireland kept us in southerlies. A 1000mb Scandinavian Low was sweeping northerly winds close to northern Scotland but they never reached the British Isles. All three timeframes are available.

GFS had a 990mb low north of Scotland at T+144, with westerly winds. At T+120, weak 1020mb low over Scotland, highs to the west and east and northerly winds setting in. At T+96, however, it was very accurate. Only 5/10.

UKMO was reasonable at T+144 but missed the wave feature west of Ireland, giving slack winds and high pressure over Britain. T+120 was worse, with the Scandinavian Low placed to the north of Scotland instead, giving westerlies for the north. T+96 was accurate, similarly so to the GFS. This scores 7/10, as the T+144 was actually quite good.

ECMWF was dodgy at T+144 and T+120, showing Britain in a slack trough, highs to the W and E and lows over Scandinavia and west of Spain, but again it took until T+96 before the ECMWF became very accurate. A score of 6/10.

NOGAPS was remarkably accurate at T+144, the low west of Ireland slightly overdone giving moderate southerlies, but closer than any other model at T+144. T+120 had slack HP over Britain, highs to the W and E, Scandinavian Low and low west of Spain. No T+96 though. I give this a 7/10.

JMA was available only at T+144- shame, because it was actually quite accurate at T+144.

GEM was quite accurate at T+144, showing southerly winds and the lows and highs correctly placed. At T+120 it had the Spanish low underdone, showing a pronounced high over Britain. No T+96. I give this a 6/10.

No models were great- all seemed to have trouble with the wave feature off the west of Ireland, which ultimately scuppered the subsequent northerly. The northerly that got downgraded out of existence will crop up in the 17/18 October analysis.

17 October

The wave west of Ireland deepened to 990mb, and southerlies established over Britain. A 990mb low resided over northern Scandinavia, propelling a northerly airstream off northern Scotland.

GFS was poor at T+144, showing high pressure over Greenland and Iceland, and ENE winds over Britain, starting to drag in colder Arctic air. At T+120 it was reasonable, showing slack southerly winds developing over Britain, but underdoing the 990mb low to the west (hence the slack flow). T+96 was spot on. I give this a 7/10, despite the poor T+144, as it killed off the northerly incursion by T+120.

UKMO had a 1030mb high off NW Scotland at T+144, with slack easterlies, changing to northerlies in northern Scotland. T+120 was quite good but had the 990mb lows too far west, giving more of a SE flow over Britain. T+96 was quite accurate but had the 990mb low too far east, again giving us more of a SE flow. Only a 6/10.

ECMWF was poor at T+144, showing a very cold NE flow over Britain, and temperatures supporting snow showers over high ground in the north. T+120 and T+96 were very accurate though, and so the ECM scores a 7/10.

NOGAPS scores an 7/10, although it was unavailable at T+120. Both T+144 and T+96 correctly showed southerlies over Britain but with the wave feature over Ireland underdone, resulting in higher pressure and a slacker flow than actually happened.

JMA again low on availability- only T+96- and it was quite poor, the low to the west too far east giving cyclonic southerly winds.

GEM was way out at T+144, with HP over the south, lows to the north and westerlies covering a large part of Britain. No T+120. At T+96 it had low pressure over the south, and ENE winds over Britain. Only a 3/10, it never came around to the right scenario.

Overall the NOGAPS seems to have been doing surprisingly well recently. GFS and ECMWF were also quite good, following their northerly projections for T+144.

[Thundery wintry showers]

18 October

Synoptic situation: low at 990mb to the SW, also NE Scandinavian Low and Greenland High, southerlies over Britain, northerlies to the north of Scotland.

GFS was out at T+144, showing NE winds over Britain, the low to the SW too far south. At T+120 and T+96 it was better, but showed us in easterlies rather than southerlies, the low to the SW remaining too far south, and no evidence of the wave feature off Ireland, only 3/10.

UKMO, likewise, showed NE winds at T+144, but at T+120 and T+96 it latched onto more or less the correct pattern, showing S/SE winds over Britain and lowest pressure off the south-west, so a score of 7/10.

ECMWF was okay at T+144 and T+120, showing ESE winds over the British Isles, and southerlies pushing into the south, but at T+96 it was worse, showing easterlies and Arctic air beginning to spread south to Britain, so 5/10.

NOGAPS had low pressure over Britain at T+120, and then easterlies at T+96; T+144 unavailable. Overall a score of 5/10.

JMA was unavailable at T+144. At T+120 it had the correct pattern, with SSE winds over Britain, the lows to the west a bit too far east. T+96 was poor though, a low over N Scotland and westerlies. Also 5/10.

GEM had ENE winds over Britain at T+120 and T+96; T+144 unavailable, and scores just 3/10, it committed much the same errors as GFS.

Overall, the UKMO was the first to latch upon the correct pattern with consistency.