Bishop Hill No significant trends in rainfall extremes
Sep 30, 2015 Climate: WG2 ;)
In January 2013, in the wake of some heavy rainfall, the Met Office published a non-peer-reviewed briefing paper that found that extreme rainfall might have increased. This was reported by Roger Harrabin here, with readers informed that "Professor Julia Slingo, chief scientist at the Met Office, said the preliminary analysis needed further research but was potentially significant."
Today, the International Journal of Climatology has published just what Prof Slingo was a looking for. However, the paper concerned has a more mixed message than I think she might have wanted:
The England and Wales precipitation (EWP) dataset is a homogeneous time series of daily accumulations from 1931 to 2014, composed from rain gauge observations spanning the region. The daily regional-average precipitation statistics are shown to be well described by a Weibull distribution, which is used to define extremes in terms of percentiles. Computed trends in annual and seasonal precipitation are sensitive to the period chosen, due to large variability on interannual and decadal timescales. Atmospheric circulation patterns associated with seasonal precipitation variability are identified. These patterns project onto known leading modes of variability, all of which involve displacements of the jet stream and storm-track over the eastern Atlantic. The intensity of daily precipitation for each calendar season is investigated by partitioning all observations into eight intensity categories contributing equally to the total precipitation in the dataset. Contrary to previous results based on shorter periods, no significant trends of the most intense categories are found between 1931 and 2014. The regional-average precipitation is found to share statistical properties common to the majority of individual stations across England and Wales used in previous studies. Statistics of the EWP data are examined for multi-day accumulations up to 10 days, which are more relevant for river flooding. Four recent years (2000, 2007, 2008 and 2012) have a greater number of extreme events in the 3- and 5-day accumulations than any previous year in the record. It is the duration of precipitation events in these years that is remarkable, rather than the magnitude of the daily accumulations.
I haven't read the paper, and so I wonder what other multi-day accumulations they tested and whether they assessed the possibility that these results might have happened by chance. I'm not sure I'm alarmed by any of this.
Reader Comments (28)
At this point I can not see anything wrong with Slingo's comment however there is not enough information given for us to draw a conclusion. This is all about GB and so has no relevance to climate anyway?
Ah, but, Dung, you are missing the obvious: when something happens in the UK that could be “potentially significant”, its significance is that it can lead to the catastrophic consequences of human-caused climate change, and it must soon affect the whole world! However, should you note that it has been a bit chilly this year, that is mere weather, and, anyway, it’s too local to be of significance. You really don’t know much about spin, do you?
Hmmmm...
"multi-day accumulation".
I do hope that under any statistical analysis they remembered to adjust their estimates for variance to account for the serial correlation of rainy days.
In other words, that they didn't make the mistake of underestimating the potential standard deviation of their statistic of choice (which appears to be the sample mean of number of days). Doing so might lead them to conclude that a measured amount of actual rainfall over a 5 day period is more unlikely than it actually is.
I've just been blogging about the lack of news stories on global warming today.
It's probably no more significant than the flooding, and purely coincidence, but it does occur to me that with the climate extremist journos yanked off climate to do some real work on the VW affair, it is a good day to bury bad news on the climate.
... and there's the steel mill closing due to carbon taxes
... and there's the news "MASSIVE GLOBAL COOLING process discovered “could explain the recent disagreements” between models and reality."
So it doesn't rain more often and there isn't more rain... it's closer together? So CO2 makes clouds more sociable? Are they scared to hang out on their own now?
Re: "It is the duration of precipitation events in these years that is remarkable, rather than the magnitude of the daily accumulations." Sounds to me like they kept looking until they found something "significant".
Obligatory XKCD Cartoon, "Significant" - http://xkcd.com/882/
English extremes are still less than the Scottish average.
Seems to be a lot of data diving going on, If the flooding issue is rain events of up to 10 days duration presumably there is a lower bound. Also, given the difference in rainfall amounts between west and east, north and south, high ground and low ground I expect there to be a far more refined hypothesis under investigation. Otherwise this smacks of dustbowl empiricism or headline hunting.
I'm always suspicious when the headline or abstract claims that the level of something was greater "than any previous year in the record" without also stating the date range covered by the record and the amount by which the previous "record" was exceeded.
It is impressive that they can examine records for the Jet Stream going back to 1931. Electron microscopy has revolutionised dendrochronology, to give the age of the tree that dropped an apple on Newton's head.
Re: Pete Austin
That xkcd cartoon ( http://xkcd.com/882/ ) could be re-written as:
"We found hothing significant for 1 days accumulation"
"We found hothing significant for 2 days accumulation"
"We found it rained more in 2000, 2007, 2008 and 2012 for 3 days accumulation"
"We found hothing significant for 4 days accumulation"
"We found it rained more in 2000, 2007, 2008 and 2012 for 5 days accumulation"
"We found hothing significant for 6 days accumulation"
"We found hothing significant for 7 days accumulation"
"We found hothing significant for 8 days accumulation"
"We found hothing significant for 9 days accumulation"
"We found hothing significant for 10 days accumulation"
It is potentially significant that nothing significant was found. There is a statistically high confidence level in this possible conclusion, when compared to a wide range of probability indices.
Translation. We had a guess, and can't prove we were wrong. But nobody else can either.
RE Geckko.
What, you mean like when they stated that chance of a severe UK winter was 1 in 20 and that severe winters from year to year were independent? So after three consecutive severe UK winters around 2010, based on their statements, I now know I must have lived through a 1 in 8,000 chance event. Amazing, something that probably has only happened once in 8,000 years.
Or not.
Winter of 1929/30 still wettest on record, except it isn't if the Wet office say so. Last time I looked the rainfall for England, Wales, & the UK has been a flat-liner since records began!
1931 to 2014? 83 years? Trends?
The major influence on our weather is the Atlantic. Even the Met office now recognises the AMO, (Atlantic Multidecadal Oscillation) and that it oscillates on approximately 60 year cycle. So what can possibly be derived from data spanning such a relatively short timescale? Never mind the numerous other cycles that are known to contribute, El Nino being the latest:-)?
We know nowt! Maybe in another millennia we might start to scratch an itch.
I consider myself severely reprimanded however:
Slingo only said "the preliminary analysis needed further research but was potentially significant."
I would add that Penfold is getting far to big for his boots!
Exactly, TerryS.
Statistical data mining of this sort is not allowed in clinical drug trials for good reasons.
But climate-science actively encourages behavior that befits a sports commentator looking for a record to report.
Seek, and yea shall find.
A hypothetical example:
"That is the record fastest 50 scored by an Australian number four batsman on the second day of the third test match in an Ashes series played in England during the month of July this century on a Friday."
Dung, in climate science, significant errors are those which have a downward trending, abrupt negativity on salaries. No climate scientist ever admits to significant errors, when computers can be blamed for input data processing inconsistencies.
Previous Met Office analyses, finding that extreme rainfall has increased in the UK, have usually started in 1960. However, the 1960-90 period is well known as a flood dry one compared to previous decades, making trends meaningless.
https://notalotofpeopleknowthat.wordpress.com/2015/09/24/state-of-the-uk-climate-2014/
As I have shown many times, when you go back to 1930, the trend disappears. For instance:
https://notalotofpeopleknowthat.wordpress.com/2014/01/24/corinne/
michael hart
Thank you very much for reminding me of a limerick which encapsulates the cricket commentator's urge for finding and reporting records, non-records, near-records:
"There once was a bowler named Grover
Who bowled sixteen wides in one over.
It had never been done
By a clergyman's son
On a Thursday in August at Dover."
TerryS and michael hart, in many fields of human behaviour etc, the constant source of data is the 'swedish twins survey'. I do not know how old all those twins are now, but their data has been serially abused, by professionals looking to prove their theories.
Judging by marks made in walls and on bridges, the highest flood recorded on the upper Thames was in 1894. Nobody alive today can remember it, so it is officially forgotten.
1947 was a very bad year too. It was a long hard winter with huge accumulations of snow. Then it warmed and rained, and all the winters precipitation hit the rivers in 48 hours. So it was massive flooding NOT caused by heavy rain. Terribly inconvenient for climastastrophogical schience
If Slingo had twins, what would be the probability BOTH aree hired by the MET UK ??
Having read this posting and the comments about Carney's speech, here's another green conundrum this time concerning the move from petrol to diesel to cut CO2 emissions.
"The RAC said diesel demand had risen 76 percent over the past 20 years, compared with a 46 percent decrease for petrol.
This mismatch has for years put pressure on European refineries' profitability. Capacity of more than 2 million barrels per day (bpd) has closed since 2009, including the Coryton, Teesside and Milford Haven refineries in the UK.
These changes tripled Britain's net imports of petroleum products last year."
From http://uk.reuters.com/article/2015/09/16/uk-britain-diesel-supply-idUKKCN0RG2EY20150916
Now laid bare by the Volkswagon fiasco no wonder the UK's deficit won't go away, green malfeasance in all directions.
Is Mann Made Global Warming (tm) climate science the phlogisten of the 21st century?
Mailman
mailman, phlogiston theory was good, until proved wrong. Mann's proof was flawed, and never shown right.
Subsequent evidence has however demonstrated that since Mann's Hockey Stick first reared it's non-existing blade, the excess heat in the atmosphere, appears to have been sucked out. Where it is hiding, nobody knows. But Mann is phlogging his non-combustible dead stick through the legal system.
We've already been through this. There is an identified 60 year cycle of wet weather in the UK caused apparently by natural warming of the Atlantic, ie the AMO. The met office surely know this. So of course going back to the 30's removes any trend. What is the bleeding point of the paper - just to display yet more ignorance of basic stats? Global warming could only be good for the UK - that's the inconvenient fact.
Do rainfall records only go back to 1931?