12z Model Comparisons Results

[Thundery wintry showers]

19 October

Subsequently, with dates from 20 October onwards missing due to my week of internet problems, the next updates will be for recent days, starting 27 October.

19 October was supposed to be the starting date for a potential northerly incursion, but ended up with a 985mb low over Ireland, and a southerly flow for the British Isles.

GFS was poor at T+144, with NE winds over Britain and some very cold air sweeping in. T+120 was better, but still with cold NE winds; lowest pressure way to the south of where it should have been. T+96 was then much better, with lowest pressure to the west, but also a low over northern Scotland, giving SW winds over Britain. I only award this a 4/10.

UKMO was moderately poor at T+144, the main low too far south (but less so than on the GFS) with easterly winds over the British Isles. T+120 was good, the low about 200 miles too far west, but with southerlies shown over Britain. T+96 was then very good. I give this a 7/10.

ECMWF was okay at T+144, with easterlies over Scotland and southerlies elsewhere; lowest pressure a bit too far south. T+120 was poor though, similar to GFS, with NE winds and some very cold Arctic air sweeping in. T+96, however, was very accurate. Overall a 5/10.

NOGAPS had lowest pressure right over Britain at T+144 and T+120, with cold NE winds for northern Scotland, milder in the south. T+144 in particular had a very cold outlook for the north and west, then T+96 had the low more accurately positioned, over western Britain. Overall a 5/10.

JMA was all over the place: at T+144 it had a slack low over Britain, easterlies over Scotland, and the lows to the SW way out in the Atlantic. At T+120 it had a slack low pressure complex over Britain, a low over the south and another one over NW Scotland. At T+96 it had a slack low over the south, lowest pressure well away to the SW and easterlies over Scotland. I only give this a 3/10; it seemed very poor.

GEM had moderately cold north-easterlies over Britain at T+144, and very cold northerlies at T+120. Only at T+96 was it close to being correct, with a deep low over the west, and southerly winds in the east. This was about as poor as GFS so scores 4/10.

The failed northerly has been covered; it would seem that the UKMO was the most "circumspect" of the models, correctly backing away from the northerly quite early on. GFS stuck consistently with the northerly until T+96, whereas ECMWF seemed unsure of it until T+96, when it too went for southerlies. Maybe a result worth watching for when we see northerlies hanging around the T+120-T+168 frame.

[Thundery wintry showers]

A large leap ahead to 26 October

Deep 980mb low over northern Scotland, very strong westerlies (gales in the north and northeast).

Frustratingly, the GFS charts don't seem to be updating properly into the Model comparisons charts on N-W- I remember them being far more accurate than I see from the archive from that site (e.g. the low is shown at 1020mb at T+120 when I'm certain that it was shown as far deeper than that)

UKMO was poor at T+144, with a slack low to the east and northerly winds over Britain. T+120 was good, with the low of correct intensity but a bit too far south, situated over S Scotland. At T+96 there was an elongated low off N Scotland and NW winds. Overall I give this 6/10.

ECMWF was poor at T+144, showing northerlies and slack lows in the North Sea. T+120 was okay, with the low over southern Britain, but not as good as UKMO at T+120. T+96 was similar to UKMO's T+96 with an elongated low to the north. Overall only 5/10.

NOGAPS had the low off SE England at T+144 with NE winds. At T+120 it had the low off SW England, giving southerlies for the south, easterlies for the north. At T+96 the low was shown over SE England, with easterlies for most. Only a 3/10.

JMA had an anticyclonic northerly regime at T+144. At T+120 the main low was too far east, giving a chilly NW flow. Only at T+96 was it accurate, with the low of the right intensity and positioning; overall 5/10.

GEM had northerlies over Britain at T+144, and again at T+120 although at T+120 there was evidence of a deep low in the North Sea (probably the main low too far east). At T+96 the low was too far south giving easterlies in the north; only 3/10.

The Excel file will be updated on 1st November.

Edit: Looking at the "model comparison" page, the GFS seems to be working okay now, but it wasn't for 26 October.

Edited October 28, 2006 by Thundery wintry showers

[Thundery wintry showers]

27 October

High at 1025mb over Europe, 990mb low south of Iceland, deep east-Scandinavian Low, slack south-westerlies covering the British Isles.

T+96's are missing.

GFS seemed to be working this time. It showed a very accurate projection at T+144, Euro High slightly too far north, covering the far south of England, but otherwise excellent. At T+120, though, the low south of Iceland was underdone, and a large high was shown to the NE, overall very low resemblance of the correct pattern. Hence, I only give this a 5/10.

UKMO was okay at T+144, a deep low over Scandinavia shown a little too far west, and likewise the Icelandic Low (underdoing the speed of depression movement) with Britain in westerlies and a Euro High. However, T+120 was accurate, the Icelandic Low slightly overdeepened, and a SW flow shown. I give this 7/10.

ECMWF wasn't up to its usual standard- at T+144 the Icelandic Low was too far to the NE, positioned to the east of Iceland, giving a pronounced WSW flow over Britain in a warm sector, and at T+120 the low was too far east, giving a strong WSW flow over Britain. In both cases the Icelandic Low was a good 300 miles too far east, so this scores only 5/10.

NOGAPS was poor at T+144 with a deep low over NE Scotland, and gale force NW winds over Britain. T+120 was much better, the Icelandic Low slightly too far east, and SW winds shown over Britain and a Euro High; another 5/10.

JMA was dodgy at T+144 with the Euro high too far north, giving an anticyclonic picture over the south, and slack westerlies in the north. T+120 was pretty accurate though, with the Icelandic Low overdeepened to 980mb giving stronger SW winds over Britain. I award this a 6/10.

GEM was poor at both timeframes. T+144 had a low over Scandinavia, no low to the NW and northerlies over Britain, and at T+120 a deep low was shown over NE Scotland and gale force W/NW winds over Britain. Only a 2/10.

UKMO seems to have been doing rather well recently, following a rather disappointing spell in which it was third after the ECMWF and GFS.

[Thundery wintry showers]

28 October

1000mb Icelandic Low, westerlies over Britain, a 1025mb high over France. No T+120s available.

GFS had the Icelandic Low too far west at both T+144 and T+96, but the general pattern over Britain was correct, WSW winds and a warm sector; high pressure a bit too far north. I give this 7/10.

UKMO also scores a 7/10, as it had the low too far east at T+144, giving more of a westerly flow and in a polar maritime sector, but the T+96 output was very accurate.

ECMWF was again poor at T+144- a Scandinavian Low was too dominant, giving WNW winds, and the Icelandic Low was too far west- it seemed to underdo the speed of the systems. T+96 was pretty good, similar quality to the two GFS outputs (low slightly too far west). A 6/10, I don't think it was quite as good as UKMO or GFS.

NOGAPS had us in WNW winds at T+144, with lows to the north, a good >300 miles too far east. T+96 was similar, a low over N Scotland connected to a larger low to the north, and NW winds. Only a 3/10, it had the westerly regime right, but that was pretty much it.

JMA was dodgy at T+144, the Icelandic Low instead situated to the north of Scotland, giving strong westerlies over the British Isles. T+96 was poor with the low too far SE and overdone, giving strong to gale force SW winds. Only a 4/10, though I think it was closer than the NOGAPS.

GEM was like ECMWF at T+144, with Britain between a Scandinavian Low and a low west of Iceland- slack winds over the north, westerlies over the south, if anything I think it was worse than ECMWF. T+96 was dodgy with the Icelandic Low situated to the north of Scotland, giving a cyclonic westerly for the north, and a warm sector SW'ly for the south. Also just 4/10.

Disappointment from NOGAPS, JMA and GEM, while the "big three" all did okay but not spectacularly.

[Thundery wintry showers]

29 October

No T+144s available; however availability of all three timeframes can be expected from 30 October onwards. There was a slack 1020mb high over Britain that brought mostly dry sunny weather and high temperatures.

GFS did well this time- the high marginally too far SW at T+120 giving WNW winds for Scotland, and high pressure elsewhere, then spot on at T+96. A good 8/10.

UKMO wasn't quite as good as GFS, positioning the high too far south at both timeframes, with high pressure over England & Wales and westerlies over Scotland. A score of 7/10.

ECMWF was similar to UKMO; slightly worse at T+120 but slightly better at T+96, in both cases with the high too far south, and another 7/10.

NOGAPS was remarkably poor- at T+120 it had a low over N Scotland giving NW winds, and at T+96 it had lows out in the North Sea, with northerlies for most. Only a 2/10.

JMA had northerlies at T+120, with a deep low in the North Sea. At T+96 it was better with high pressure over the west, but still northerlies in the east, so only 4/10.

GEM had our high over Europe at T+120, with south-westerlies over Britain and low pressure over Shetland; at T+96 it was similar to ECMWF/UKMO but worse with the high well to the south and westerlies for all, the high only just covering the south coast. A score of 4/10.

GFS was the most accurate this time though UKMO and ECMWF were close behind; the NOGAPS, JMA and GEM did very poorly though.

[Thundery wintry showers]

30 and 31 October coming up; 1 and 2 November with Excel/Word graphs to come tomorrow. I don't have time for them today as I'm doing the monthly forecast.

30 October

Low at 992mb to the south of Iceland, 1020mb Euro high and strong westerly winds.

GFS was very poor at T+144- no evidence of the Icelandic Low at all, and just a slack southerly flow. T+120 was much better, with the low slightly underdone but positioned correctly, and westerlies over Britain. T+96 was pretty much perfect. So despite the poor T+144 I award this 7/10.

UKMO picked out the low at T+144 but the orientation was completely wrong; low just off W Scotland and a SSW flow over Britain, westerlies penetrating into the west. T+120 was poor, with an over-strong Euro High and the Icelandic low way too far west, giving an anticyclonic SW regime. T+96 was better, with the Icelandic Low slightly underdone but positioned correctly, however it had a secondary low in the Atlantic that didn't materialise. Only a 5/10 for this one.

ECMWF was decent at T+144, although low pressure to the north was too elongated giving a very long run of westerlies. T+120 was poor, the Euro High overdone and low pressure vastly underdone, giving an anticyclonic SW flow. T+96 was reasonable, though as with UKMO there was evidence of a secondary low behind the Icelandic one that didn't materialise. A 6/10.

NOGAPS had low pressure too far west at all three timeframes, and southerly winds over the British Isles, the high to the SE also positioned too far north; only 3/10.

JMA was similar to NOGAPS- indeed at T+144 the low was so far west, that we had high pressure covering most of eastern Britain, with southerlies for all at T+120 and T+96. Only a 2/10, it never got near being correct.

GEM had southerlies at T+144 with low pressure too far west, but at T+120 it was accurate, with lows and highs in the right places. At T+96 the Icelandic Low was too far west, with SW winds the result. Overall a 5/10.

In general the GFS seemed to handle this one best, with the other models having problems; the NOGAPS and JMA were particularly poor.

31 October

The northerly blast over Britain commenced, with a low at 985mb to the east of Britain, out in the North Sea.

GFS had already picked out the evolution to the northerly by T+144, but had the low too far west, giving more of a NW flow across the country. At T+120 the low was over NE Scotland giving NW winds that were swinging around to the north. The low was still too far west at T+96, giving a NW flow- I think the problem seemed to be that the GFS underestimated the speed of movement of the low. Overall a 6/10.

UKMO had also picked out the evolution to the northerly at T+144, but with the low too far west, giving westerlies (and also massively underdone). However at T+120 the low was out in the North Sea, giving NW winds that were swinging around to the north, and then T+96 was pretty much perfect. Overall a 7/10.

ECMWF was nowhere near at T+144, with southerlies shown. T+120 was good, though the North Sea low was too far east so western Britain was in a slack northerly. T+96 was almost perfect. Overall 6/10.

NOGAPS was well out at T+144 with SW winds shown. The low was positioned to the north of Scotland at T+120, giving cyclonic westerlies, and at T+96 the low was about as far west as on the GFS, giving NW winds. Only a 4/10.

JMA was almost identical to NOGAPS, and so also scores 4/10.

GEM had an elongated low to the north at both T+144 and T+120, giving westerlies. At T+96 it had a low to the north of Scotland ready to provide a northerly behind it, but over Britain we still had westerlies. Also a 4/10.

GFS and UKMO were the first to pick out the northerly, but GFS tended to delay its onset by a day, so came off the worse of the two.

[Thundery wintry showers]

1 November

No chart available for posting up, but there was a strong high pressure at 1035mb over the west, with moderate northerlies down the eastern side bringing some east-coast rain showers.

GFS was extremely accurate at both T+144 and T+120, so obviously the delayed northerly for 31 October didn't upset the projections for 1 November. Surprisingly the only dodgy output was at T+96, with more of a north-easterly flow coming into northern Scotland, but otherwise accurate, so 8/10 overall.

UKMO was okay at T+144 though it had us in more of a NNW flow. T+120 and T+96 were excellent, so this also scores 8/10.

ECMWF also scores 8/10- it overdid the northerly at T+144 with high pressure too far to the NW, but T+120 and T+96 were spot on.

NOGAPS was dodgy at T+144, with a straight NW flow shown, but T+120 and T+96 were very accurate, so this scores 8/10 as well.

JMA was the first dodgy output, with a cyclonic NW'ly shown at T+144, an overdone northerly with high pressure too far NW at T+120, but an accurate output at T+96, overall 7/10.

GEM underdid the northerly at T+144, and overdid it at T+120 with high pressure too far west. T+96 was spot on though; overall 7/10.

A very good set of outputs from the models.

[Thundery wintry showers]

model_comparisons.xls

Updated to 1st November. A Word document will soon be arriving in John's email inbox.

[Thundery wintry showers]

2 November

The models should do well for 2 November as it was a pretty simple chart: 1035mb high right over western Britain and a rather chilly outlook.

GFS was very good although the high was slightly underdone at T+144- outlook still the same, but high at just 1025mb. T+120 and T+96 had 1035mb high, marginally too far north at T+120 with easterlies for southern Britain, and correctly placed at T+96. I give this 8/10, though it couldn't manage a 9 because of the errors in high intensity and positioning.

UKMO does score a 9/10 as it was very accurate at all three timeframes- even T+144- the only slight quibble being that the high was a bit more elongated N-S than should have been the case.

ECMWF was also very accurate at all three timeframes- again slight quibbles over elongation of the high N-S, but a solid 9/10.

NOGAPS was also very accurate at T+144 and T+96, but at T+120 it had the high about 100 miles too far west, with a chilly northerly for eastern England; overall a mark shaved off for that, so 8/10.

JMA was also accurate at all three timeframes, though at T+144 and also T+96 the high was situated too far north, giving easterlies for southern England; overall 8/10.

GEM also scores 8/10- it was unsure about the positioning of the high, but in all three cases it was thereabouts over western Britain.

As expected, the models were very good at identifying the anticyclonic outlook, with UKMO (which appears to have improved significantly over the past month) and ECMWF providing the most accurate outputs.

[Thundery wintry showers]

3 November

1030mb high over Britain with an extension out to the west.

GFS scores 9/10, as it was near-perfect at all three timeframes, just the slight issue at T+144 of the high being marginally overdone (1035mb)

UKMO was pretty much identical to the GFS, so also a 9/10.

Surprisingly ECMWF was marginally less good; high a bit too far north at T+144 with easterlies in the extreme south, and at T+120 there was relatively low pressure to the W that didn't materialise. However still good for an 8/10.

NOGAPS had us in northerlies at T+144, with a Greenland High and Scandinavian Low. However, at T+120 and T+96 it was accurate. Overall just a 7/10, because of the poor output at T+144 (I recall the NOGAPS was noted for sticking with a prolonged northerly)

JMA had a high correctly over us at all three timeframes, but pressure was low to the west at T+144 and T+120 (much as on the ECMWF T+120) rather than an elongation of the high. An 8/10.

GEM had the high too far west at T+120 with northerlies for eastern Britain, but T+144 and T+96 were both accurate, so this too scores 8/10.

All models were reasonable, with only NOGAPS going for a radically different scenario at any timeframe, with UKMO and GFS being the most spot-on.

[johnholmes]

model_comparisons.xls

Updated to 1st November. A Word document will soon be arriving in John's email inbox.

many thanks Ian, it is much appreciated. One day with a new pc - dream on!!

[Thundery wintry showers]

4 November

1035mb high over southern Britain and strong westerlies for Scotland.

The GFS was good at all three timeframes, though at T+144 and T+120 it had the high centred to the southeast, rather than south, of Britain. It scores 8/10.

The UKMO was almost identical to the GFS, positioning the high about 200 miles too far east at T+144 and T+120, but with the correct pattern over Britain, then accurate at T+96, so 8/10.

ECMWF also scores 8/10. At T+144 it was worse than UKMO and GFS, with a high in the North Sea and southerlies for many, but it latched onto the correct positioning of the high earlier, at T+120.

NOGAPS also had problems with the orientation of the high at T+144 and T+120, positioning it over East Anglia with an anticyclonic regime, but it was oriented correctly at T+96; in all three cases the pattern over Britain was correct, so another 8/10.

I only give the JMA 7/10. At T+144 the high was too far NE, giving SSE winds for most, and at T+120 the high was to the SE rather than S, though the pattern over Britain was right. At T+96 the high was a little too far north, which matters little away from northern Scotland, but with N Scotland under the high and not a westerly regime, it would have made quite some difference.

GEM is also at 7/10. Orientation problems at T+144, but T+120 was the worst of any output I've seen, with the high off SE England and a low coming into western Scotland with SW winds. However, T+96 was accurate.

Another good set of outputs, with little to choose between GFS, UKMO, ECMWF and NOGAPS.

[Thundery wintry showers]

5 November

I was out at a chess match on 1st November, so I don't have the T+96 outputs for this one. The high was sat over southern Britain with an anticyclonic westerly flow over the north.

GFS scores another 8/10: high slightly too far east at T+144 with the pattern over Britain correct, accurate at T+120.

UKMO was similar; high slightly to the east at T+144, but accurate at T+120, so 8/10.

ECMWF also pretty much identical to UKMO and GFS, so 8/10.

NOGAPS scores 7/10, as the alignment of the high was completely wrong at T+144 and it was too far north, but it was reasonable at T+120- high slightly too far east though.

The high was too far east on the JMA at T+144, giving southerlies, and it was too far north at T+120, though in both cases the anticyclonic outlook was correct. I think this one scores only 6/10 as the alignment was wrong at both timeframes.

GEM only scores 5/10- at T+144 it had Britain in a NW flow and the high well to the south, and at T+120 the high was just covering the extreme south with westerlies elsewhere.

Almost identical outputs from GFS, UKMO and ECMWF, but less impressive from the JMA and most especially GEM.

[Thundery wintry showers]

6 November

1025mb high over southern England, and an anticyclonic airflow from the WNW in the northern half of Britain. No T+120s available.

GFS scores another 8/10; the high was marginally too far south at T+144, but at T+96 it was spot on. In both cases, an anticyclonic WNW flow covered northern Britain.

UKMO also scores 8/10; the high was too far north at T+144 with westerlies only in Scotland, but it was perfect at T+96.

ECMWF only scores 7/10; the high was too far to the SE at both timeframes, with southerlies in the south-west, and W/SW winds in the north.

NOGAPS scores 7/10; it was similar to ECMWF, with the high too far east and south, particularly at T+96, giving W winds in the north and strong southerlies in the south-west.

JMA was very good- high slightly too far to the SE at T+144, accurate at T+96, and scores 8/10.

GEM is another 7/10- the T+144 was quite accurate, but at T+96 the high was over Europe, with an anticyclonic/south-westerly flow covering the British Isles.

Overall, the models are doing well in this anticyclonic spell, though I think it's purely because small differences in high positioning make far less difference to Britain's weather than differences in low positioning. GFS and UKMO and JMA were the most accurate.

[shuggee]

I've got a sneeky 'feeling' that it's the UKMO Fax that will have the accolade in the next analysis for this difficult week.

[Thundery wintry showers]

I missed a few days when in Glasgow last weekend, but I will be updating 7-10 (possibly as late as 11) November over the next few days, and then resuming up-to-date coverage by around 18 November.

The marks will probably be high for the earlier period, but I wouldn't be surprised to see some low ones for the more recent period...

[Thundery wintry showers]

Annoyingly, the data didn't download properly for 7-11 November, so the comparisons data will have to resume for 18 November.

Apologies for that.

[Thundery wintry showers]

18 November

975mb low to the north of Scotland, and a westerly flow covering the British Isles. Unfortunately for now only T+96 is available as I don't have the outputs for up to 13 November, but here are the results:

GFS appeared to be the closest of the three most high-performing models, with the low to the north accurately positioned, but a Euro High was shown, meaning the projected westerly winds were stronger than actually happened. Low to SW overdeepened; overall a 7/10.

UKMO's T+96 output was further from the mark, showing a low of correct intensity, but positioned over the north-east of Scotland, well to the south of where it should have been. A very strong WNW flow was shown, with a strong Euro High to the south. Low to SW overdeepened; overall just 6/10.

ECMWF was most accurate for lack of a strong Euro High and weak low to the SW, but it positioned the low to our north way too far north, giving Britain a slack westerly flow under a ridge of high pressure. The pattern over Britain was very different to what happened; overall also 6/10.

NOGAPS scores only 4/10, as a deep low was positioned in the North Sea giving northerlies, but other lows and highs were reasonably positioned.

JMA had a low over north-east Scotland giving NW winds, overall similar to, but slightly better than, the NOGAPS, so a score of 5/10.

GEM was similar, low over north-east Scotland, but too strong a high over Europe, resulting in strong NW winds, also low to SW somewhat overdone. However, it had Britain in westerlies, so a generous 5/10.

Given that these were T+96 outputs, it seems clear that the models are having problems at the moment.

[Thundery wintry showers]

19 November

Deep 965mb low south of Iceland. Although the day dawned frosty for many, a south-westerly wind picked up and brought cloud and rain in off the Atlantic.

T+120 and T+96 are available for this date.

GFS was very poor, at T+120 a 995mb low was situated over Ireland giving a long draw of very warm southerlies. At T+96 an 1000mb low was situated right over the British Isles. Only 3/10, it was never really close, I think T+120 was if anything the more accurate of the two outputs.

UKMO was no better than GFS. At T+120 a 980mb low was over Ireland with gale force southerlies for Britain, of very warm origin. At T+96 a 980mb low was over SW Ireland giving strong southerlies. Only a 3/10, it correctly showed an Atlantic low approaching us, but it was so far south and east that the outputs were completely different to what actually happened.

ECMWF was also poor at T+120, with a 1000mb low over SW England and very warm southerlies pushing into England & Wales although Scotland was in cooler westerlies. T+96 was remarkably accurate though, all the lows and highs in the correct places. I give this 6/10; the T+96 output was impressive especially relative to UKMO and GFS.

NOGAPS had a deep low over Ireland at T+120 with warm southerlies. T+96 had Britain in westerlies, but a 980mb low was positioned to the SW of Ireland with southerlies approaching into western Britain. Another 3/10.

JMA is another 3/10- indeed very similar to UKMO, with a low over western Ireland and southerlies at both timeframes.

GEM had a low in the English Channel at T+120, light winds in the north, and southerlies pushing into the south. T+96 was better, showing Britain in westerlies, but no sign of an Atlantic depression approaching except to the SSW of southern England. Only a 2/10; here I detected very little relationship to the actual pattern.

A remarkably poor set of outputs, consistent with what many members have been saying about T+72 being the limit of reliability at the moment. Only the ECMWF produced an output that closely resembled the actual outcome.

[Thundery wintry showers]

Some might be getting put off by lack of reliable updates recently, so many apologies.

Sometimes it's been down to less excusable reasons, but this time there is a genuine reason- my computer's graphics card has packed in, so I have very limited internet access, exclusively from my dad's computer.

I hope my computer will be up and running next week, but it seems that the system is unstable as well, so I can't guarantee anything.

[Thundery wintry showers]

These are back for the time being.

6 December

I only have the T+96 outputs for 6 December, it was a very zonal outlook indeed with a 975mb low off western Scandinavia, and westerlies over Britain, and a deep 970mb low approaching western Ireland.

GFS was quite good, though the Scandinavian Low was about 200 miles too far west giving strong westerlies over Britain, 970mb low correctly placed, overall a 8/10.

UKMO was less good, showing a broad NW flow over Britain, the Atlantic low a long way too far west, overall 6/10.

ECMWF also had the Atlantic low a little too far west, but much less so than UKMO, with westerlies over Britain and an accurately-placed Scandinavian Low, overall 8/10.

NOGAPS also scores 8/10; the Atlantic low was overdeepened (965mb) and winds over Britain were more WSW'ly, but otherwise it was spot on.

JMA had the Atlantic low a bit too far west, and the Scandinavian Low was underdone, a NW flow over Britain, only 6/10.

GEM scores 7/10; Scandinavian Low underdone, but the rest was pretty good, pressure too high over the British Isles.

A reasonable set of T+96 outputs, but UKMO and JMA were rather disappointing.

[Thundery wintry showers]

7 December

Another strongly zonal outlook; 965mb low over NW Scotland and very strong west-south-westerly winds.

T+120 and T+96 are available.

GFS was rather dubious at T+120, with a slack 980mb low over eastern Britain and very little wind, but T+96 was superb. I give this a 7/10 because it was let down by the T+120.

UKMO was okay, but at both timeframes the low was a good 200 miles too far west, with the strong showery WSW flow only just reaching the British Isles. The general pattern was right but a consistent error in low positioning; 7/10 overall.

ECMWF was the most accurate of the three main models, with the low accurately positioned at both timeframes- in particular T+96 was pretty much perfect. Overall a very good 9/10.

NOGAPS overdeepened the low at T+120 but had it correctly positioned; it was spot on at T+96. I give this 8/10.

JMA was poor at T+120, with the low over south-west England, southerlies in the south, slack winds in the north. T+96 was better, but like UKMO, the low was too far west, so only 6/10.

GEM had the main low too far west at both timeframes- and it had a secondary low just SW of Ireland at both timeframes. Only a 5/10; it was certainly behind the other models in accuracy.

Overall, a fairly respectable set of outputs, the ECMWF was by far the best this time.

[Thundery wintry showers]

8th December

975mb low off NE Scotland. A 985mb secondary low over the far south-east of England, NW winds generally over Britain but not especially cold ones, due to the short fetch around the depression.

GFS overdeepened the low over south-east England at T+144 and T+120, making it the dominant feature, cyclonic E winds over the north. T+96 was better, with the low over the south still marginally overdone, but with NW winds generally. Only a 5/10; the low should not have been turned into a huge cyclone, it looked from the T+144 and T+120 outputs as if there'd be severe gales in the south.

UKMO was decent at T+144, the two lows a bit too far west but of the correct intensity, with cyclonic westerlies for Britain. T+120 and T+96 had the lows slightly too far west and westerlies over Britain, and at T+120 a low over N Scotland was well overdeepened, suggesting gales. Only a 6/10, but I think it was slightly closer than GFS.

For the second day running, ECMWF was the most accurate of the "big three". The main low to the NE was accurately positioned at all three timeframes, but the low over the south was positioned over central S England and slightly overdeepened at T+144, then accurately placed at T+120 and T+96. Overall an 8/10.

NOGAPS was accurate at T+120, but at T+144 and T+96 it had a well-overdeepened low across the south, giving severe easterly gales in the north, so only a 5/10.

JMA had a well-overdeepened low off SW England at T+144, giving gale force southerlies. At T+120 and T+96 it was better, although both lows were positioned too far west, slightly less so than on UKMO. Overall 6/10.

GEM was poor at both T+144 and T+120, having an extortionately overdeepened area of low pressure to the N and NE, and westerly gales over Britain. T+96 was better, but the low over the south was overdone so western areas had northerlies. Only a 4/10.

The models are having some problems, although the outputs haven't been terrible. I don't know if ECMWF has regained prominence over GFS recently, but it certainly seems it on the evidence of the last two days.

[Thundery wintry showers]

9th December

A large 1040mb high over the Azores, 990mb North Scandinavia Low, and a rather vicious-looking 955mb low to the SW of Iceland.

GFS had the Scandinavian Low too far to the SW at T+144 and T+120, giving a cyclonic north to north-westerly flow over Britain, though T+120 was reasonably good. T+96 was accurate, so it scores 7/10.

UKMO was poor at T+144 with a strong westerly flow, and an elongated area of low pressure to our north, centres over Iceland and to the NNE of Scotland. However, it was very good at T+120 and T+96, so it also scores a 7/10.

Once again the ECMWF did better than both GFS and UKMO, showing a spot-on chart at both T+144 and T+120. If anything T+96 was the most dodgy of the three, with the zone of strong south-westerlies extending about 200 miles too far east. However, a good 8/10.

NOGAPS had the Scandinavian Low too far south at T+144 and T+120 with northerlies for Britain, and consequently the cold was rather overdone- chances of snow showers in the north. T+96 was accurate, overall a 6/10.

JMA was even more impressive than ECMWF, showing a north-westerly flow and all the lows and highs correctly positioned at all three timeframes. A score of 9/10.

GEM was also impressive at T+144 and T+96, but at T+120 it had lowest pressure too far south over Scandinavia, giving northerlies over Britain. Overall 7/10.

So while the ECMWF again outperformed UKMO and GFS, it was the JMA that produced the most accurate output for 9 December.

Does anyone have a copy of the model comparisons files that I used to have? When my computer's system registry packed in, Windows XP decided that Drive H (where I keep my files) was faulty, due to a system registry error, and forced me to run Chkdisk on startup to "repair" the files. The result was that it corrupted many of the files by embedding system registry data into them. I've lost both the model comparisons.doc and .xls files to this.

[Thundery wintry showers]

10 December

A ridiculously deep low at 930mb just WSW of Iceland, though not quite beating the record set in January 1993. Strong WSW winds over Britain, and the day itself was mild and damp for most. 1032mb high to the south and SE.

I downloaded the wrong charts for 6 December :blink: so only T+144 and T+120 are available for GFS and ECMWF and UKMO.

GFS was impressive this time, scoring 9/10. The lows were a bit too far east at T+120, but otherwise the pattern was spot on.

UKMO was poor at T+144, with a low just north of Scotland, and a gale force run of showery westerlies. T+120 was accurate, but it only scores 6/10 because of the poor T+144 which was well out.

ECMWF was good, though not quite as good as GFS; the lows were too far west at T+144, causing more of a SW flow to be established over Britain. T+120 was good, so it scores 8/10.

NOGAPS scores only 7/10. It was close at T+120, but at T+144 and also T+96 the lows were positioned over or just east of Iceland, so although Britain was correctly in a WSW flow, the polar maritime airmasses were a lot closer.

JMA was a reasonable 8/10; a slight secondary low at T+144 left Britain in more of a SW flow (similar to ECMWF), but there was little to fault with T+120 and T+96.

GEM was accurate at T+144, but at T+120 and T+96 the lows were positioned too far east (more or less over Iceland), giving a very cyclonic SW flow and polar maritime air moving in, so only 6/10 (it appeared worse than NOGAPS).

Overall I was quite impressed with the models this time. The GFS was most impressive, but ECMWF and JMA weren't far behind. UKMO's T+144 Achilles heel returned.