Kz Vs Cs Sunshine

[Summer of 95]

Hi everyone,

I'm aware of the recent change of sunshine recording equipment that occurred at the start of this decade at many MO observation sites, and the difficulties involved comparing the sets of data from the different sensors. Most sources I've looked at suggest that CS reads slightly higher (10-15%) than KZ, and that the difference is more pronounced in intermittent sunshine. But according to the latest available estimates, what is the most accurate conversion formula?

I'm asking this because on the face of it (according to the MO data pages), very sunny (>200hrs) months have become much rarer just up the road at Shawbury ever since the change happened. Since late 2001 when the sensor changes, only two months- July 2006 and April 2007- have topped the 200 hour mark there, and they were both really exceptional once-in-a-blue moon months when averaged nationally. While prior to the change the vast majority of years managed at least one month over 200, even in many decidedly mediocre or even poor summers (often May managed it in these years, neither it nor June has since the sensor changed), since then years with no 200+ month have been in the majority-

No. of years with no month >200hrs sun at Shawbury:

1950-59: 3

1960-69: 2

1970-79: 3

1980-89: 4

1990-99: 2

2000-08: 6

No of years with 2 or more months at 200+:

1950-59: 1

1960-69: 2

1970-79: 3

1980-89: 3

1990-99: 5

2000-08: 0

Even if we take the discrepancy to be 10%, 220 hours CS was nothing unusual pre-2001 (reached 5 times for example in the 1960s, which supposedly had poor summers) while it's only been done once since the change, in July 2006, which interestingly was actually sunnier KZ than the CS July 1983, all months of 1976, and all months of 1959- though marginally duller than August 1995. This suggests to me that perhaps it's *unbroken* sunshine in summer, which produces AFAIK comparable KZ/CS readings, that's really becoming the rarity- which is certainly my perception since 1995 generally, and 2002ish especially.

Is this just a change of sensor thing, or does it genuinely reflect a change in weather?

Edited January 18, 2009 by Summer of 95

[Thundery wintry showers]

I think most of it is down to changes in sensor. Generally the CS sensors over-read, especially in conditions of strong but intermittent sunshine, but there are also circumstances when they can under-read, which I guess will mainly arise in winter. Thus we would expect a change to a KZ sensor to lead to lower sunshine totals overall, and especially in summer. This is reflected by the abrupt drop in spring and summer sunshine at the KZ sensor stations in Weather Log and the much smaller decline at stations which still use CS recorders.

The Met Office does have standard conversion formulas, as is documented in the Weather journal, but while they probably hold when averaged over a long time period, they aren't trustworthy for individual months, e.g. using them as a straight formula for conversion would have led to grossly inflated CS sunshine estimates for July 2006 which had many days with largely unbroken sunshine.

Recent summers have indeed had rather less sunshine than those of the 1990s, with 2007 and 2008 being especially poor over much of western Britain, but I don't think they've been any duller than the summers of the 1960s. My guess is that with a CS sensor Shawbury would probably have made 200+ hours in Mays 2004 and 2005. June 2003 would have run it close, and I think August 2003, June 2006 and May 2008 would've been close contenders in spite of the dull second halves of those months.

[Summer of 95]

I think most of it is down to changes in sensor. Generally the CS sensors over-read, especially in conditions of strong but intermittent sunshine, but there are also circumstances when they can under-read, which I guess will mainly arise in winter. Thus we would expect a change to a KZ sensor to lead to lower sunshine totals overall, and especially in summer. This is reflected by the abrupt drop in spring and summer sunshine at the KZ sensor stations in Weather Log and the much smaller decline at stations which still use CS recorders.

The Met Office does have standard conversion formulas, as is documented in the Weather journal, but while they probably hold when averaged over a long time period, they aren't trustworthy for individual months, e.g. using them as a straight formula for conversion would have led to grossly inflated CS sunshine estimates for July 2006 which had many days with largely unbroken sunshine.

Recent summers have indeed had rather less sunshine than those of the 1990s, with 2007 and 2008 being especially poor over much of western Britain, but I don't think they've been any duller than the summers of the 1960s. My guess is that with a CS sensor Shawbury would probably have made 200+ hours in Mays 2004 and 2005. June 2003 would have run it close, and I think August 2003, June 2006 and May 2008 would've been close contenders in spite of the dull second halves of those months.

June 2006 (195.1KZ) only just missed it, so I'd tend to agree with that one (it was predominantly a sunny, warm month spoilt only by a disappointing cool, cloudy spell about 17th-26th). It was similar sunshine-wise as I remember to June 1995 (235.0) and May 1990 (239.3).

August 2003- no, that was NOT a sunny month here with the only sunny days being 3rd-4th and 7th-9th; the rest of the month was largely cloudy except for intermittent sunny spells usually late in the day. I'd certainly think it duller than June 1992 (190. and probably also August 1990 (183.1)- I'd actually estimate it close to August 2001 (181.6).

May 2008- no again, the second half was terribly, terribly dull in the W Midlands area in particular (IIRC Birmingham was the dullest place in the country that month)- it probably got about 110 of those 149 hours in the first half. Working on the assumption that the first half would have produced similar readings but a CS would have recorded more in the second half, I reckon something in the 160s or low 170s would have shown up for the month.

As for 2005, May, June and August are all possible contenders. May certainly had several brilliantly clear, if cool, days; June a sunny but chilly start then a hot sunny thundery second half. August wasn't short on sunshine either- however it seldom got above about 22C despite several days with 8-10 hours sun. One of them at least I think would have made it- especially looking at 200+ hour totals for the unremarkeable July 1996 and the (as I recall) cold and wet June 1999.

This takes the number of years with no 200+ month down to 5 if we say 2005 wasn't on the list, and gives 2006 a "2 or more months" status- however this still leaves the 2000s seemingly lacking in very sunny months.

One CS total that really surprises me BTW is the meagre 149.9 for April 1987. Was that month yet another letdown in this area (it was shortly before I moved here)?

Edited January 19, 2009 by Summer of 95

[Thundery wintry showers]

April 1987 was one of those "Bartletty" type warm months that was warm and sunny over most of the country, especially east and south, but was dull in parts of the west and north-west. In Lancaster sunshine for April 1987 actually came out slightly below the average.

I can't remember the sunshine stats for Summer 2005 around Shrewsbury but I recall June and more especially August definitely being above average sunshine wise in the Tyne & Wear, and July near average but spoilt by an insanely dull last third.

[Terminal Moraine]

One CS total that really surprises me BTW is the meagre 149.9 for April 1987. Was that month yet another letdown in this area (it was shortly before I moved here)?

The first fortnight or so of April 1987 was generally unsettled but the second fortnight was warm and sunny.

Up to the 13th I recorded only 34.6hrs of sunshine but from the 14th onwards there were another 112.5hrs.

Incidentally the problem with CS sunshine recorders was as much to do with the observers interpretation of where the actual burn on the card ended as it was with intermittent sunshine causing burns to merge into a continuous scorch mark.