Wednesday
Sep112013
by 
Bishop Hill
Bishop Hill Time series analysis for experts
Sep 11, 2013 Climate: Statistics One regular criticism that is made of climatologists is that their statistical analyses tend not to make use of up-to-date methods, particularly as regards time series analysis. This course in Germany therefore looks to be useful:
Advanced Course in Climate Time Series Analysis, Hannover, Germany, 20 to 24 January 2014
I'm not sure how advanced though:
- Level: academic (PhD students and postdocs), industry (researchers and analysts)
- Audience: Climatologists, Geographers, Geologists, Hydrologists, Meteorologists, Physicists, Risk analysts, Statisticians
- Basic knowledge in statistics is required (e.g., you should know what "standard deviation" means).
Reader Comments (22)
Ability to use Excel a real bonus.
They are going to wonder why the course is not very well attended ...
Bit cold and damp up there at that time of year, no skiing either...
Standard deviation will be the positive offset of model output over observations adopted by 97% of climatologists. This allows for the heat that has not yet been found / will emerge in the future.
Will Phil be there?
Maybe, just for the climatologists the course will have a special section on data integrity and sampling techniques:
Surface station closures in 1990 - Ross McKittrick's graph
...Maybe, just for the climatologists the course will have a special section on data integrity and sampling techniques:...
Indeed. The OP missed out the rest of the requirements:
You should know what "standard deviation" means
You should be able to add 2 and 3 to get 5
Climatologists should be able to add 2 and 3 to get 6. Or 4, as necessary.
Dear all,
thanks for posting. --- Before the partcipants come, they should know at least "what 'standard deviation' means". After the course, they will know more. This five-days course is based on my book (Climate Time Series Anlysis, Springer, 474 pp.), a sample of which is at www.manfredmudelsee.com/book. My company and this course is devoted to excellence in research and does not discriminate against persons on the basis of ethnicity, family status, gender, medical condition, nationality or religion, even not of "climate religion".
Manfred Mudelsee
Manfred: good post!
When I am teaching Geostatistics courses to geologists and geophysicists I always begin with a brief basic
statistics primer and tell the attendees that as long as they understand mean and standard deviation they should be ok with the rest of the course. Like your advert, the attendees on my course will be at least graduates in an earth science, many with industry experience and/or PhD's, just not specialising in geostatistics (which is why they are attending a course). So I am with you.
Do the uk quackademia offer anything of the like?
Peferably general tsa .. No reason to make climate
Specific
Iapogus (Sep 11, 2013 at 1:43 PM):
Interesting graph, and one that tends to agree with my hypothesis expressed a while back as to the flat-lining in temperatures coincided with the end of relocation of observation stations.
Do you think I could get a grant to further my research?
(p.s. what is standard deviation? Is it something to do with pink fluffy handcuffs? If it includes limp celery, count me in!)
@Manfred Mudelsee.
I think that a course based on your book would be interesting. I certainly think that the links between signal processing and statistics should be emphasized ( I ground my teeth on biomedical signal processing) and this would be a great selling point for your course.
I have to say that I think one would have to know a little more than what a standard deviation is to get the most out of your course!
ps: I see you use f for complex frequency, as does Bendat and Piersol. I always have to reimagine each equation with "omega" in place of f!
Will the course include the difference between normally and non-normally distributed data? This seems to be a deep mystery to climate scientists who happily assume normality in cases where it quite obviously does not apply.
'One regular criticism that is made of climatologists is that their statistical analyses tend not to make use of up-to-date methods,' To be fair you cannot get more 'up to date ' then inventing your own statistical methods. However the fact that this was only needed because the current ones could not supply the 'result ' they required , does take the shine of that.
Hi RC Saumarez,
if I may cite from the book (p. 181f): "A word on the notation: The literature has developed a rich variety of different notations (factors 2, frequency versus angular velocity, etc.), and our is just one option." I am sure you will agree with me that consistency in usage is more important that choice of notation.
Manfred Mudelsee
Hi tty,
if I may cite again (book's back cover), this time on non-normal distributional shape and other data-analytical challenges:
"Climate is a paradigm of a complex system. Analysing climate data is an exciting challenge, which is increased by non-normal distributional shape, serial dependence, uneven spacing and timescale uncertainties. This book presents bootstrap resampling as a computing-intensive method able to meet the challenge. It shows the bootstrap to perform reliably in the most important statistical estimation techniques: regression, spectral analysis, extreme values and correlation.
This book is written for climatologists and applied statisticians. It explains step by step the bootstrap algorithms (including novel adaptions) and methods for confidence interval construction. It tests the accuracy of the algorithms by means of Monte Carlo experiments. It analyses a large array of climate time series, giving a detailed account on the data and the associated climatological questions."
Manfred Mudelsee
And right on cue:
"Digital Signal Processing analysis of global temperature data time series suggests global cooling ahead"
Hi Manfred,
This wan't meant as a criticism. Simply as an engineer, I have grown up with omega and I always do a double take whwn I see f. You can call any variable anything you like, although I would suggest that pi remains pi!
"you should know what "standard deviation" means": for how else can you identify 'outlier science'?
In 1988 CSEs and O levels were combined which meant A Levels started to decline. Until recently many Russell Gp Universities wanted A Level Maths for Geology or Geography degrees. The pre 1988 A level maths meant that anyone at university could quite a high standard in stats over a term. The pre 1988 Maths A level covers about 60-70% of the present day Further Maths A level The problem is that in environmental science, geology and geography the lowest level of maths ability is very low indeed.
As Wegman has pointed out climate science needs the same standard of stats as in medicine and drugs research. Most pre 1988 statisticians entering medical/drug research would have completed either Pure and Applied Maths or undertaken Maths and Further Maths A levels . Pre 1988 standards in Maths is probably only found amongst those reading maths,physics, engineering at Cambridge or Imperial. Have heard of someone readings physics at Oxford without Further Maths A level.
Many of the problems in science and engineering are because standards of those entering university are 1-2 years below those who went up pre 1988. Someone who won a scholarship to Cambridge /Imperial/Manchester pre 1988 is probably equal to someone half way through the second year at university.
Some people are naturally good at stats and those who are not, a two day course could result in " A little knowledge is dangerous". I think stats is subject where unless one is undertaking the calculations all the time, one could become very rusty, very quickly.
@Charlie,
I agree with your lamentation about Maths A level. I did mine in 1965 - shows what an old F**t I am, and these were built on when I did engineering courses.
I was horrified to discover when I went back to my old School that calculus isn't taught until half way through the 2nd year of maths A level. I distinctly remember in the "advanced O levels" maths, there was always a volume of a solid of revolution problem.
Presumably, universities now have to do a good deal of remedial teaching.
As regards statistics - I do statistics occasionally and I always have a terrible fortnight's self education but at least if you have learnt it sometime, you can pick up the pieces.
RC Saumarez.
Teachers and education officials appear to have deliberately erased memory of previous high standard of UK maths education. Pupils were allowed to take maths O level 1 or 2 years early. There used to be 2 O Level syllabi , a standard one and a more advanced one which included calculus .There were also Additional A levels which could be taken at the end of the first year of A Levels. Additional O Level Maths is probably 60-70 % of modern Maths A level.
Most top schools , split A Levels into 2 sets, the top one ( destined for Oxbridge/IC/Manchester-scholarship) was taught to first year degree standard. By the mid 80s, most comprehensives complained they lacked the resources to teach Oxbridge ( and IC scholarship such as Royal Scholarships). What they meant was that they lack staff with good enough education to prepare pupils for scholarship exams.Those who took Maths O Level 1- 2 years early ( used to beAutumn O levels ) and destined for university scholarships at Ox/IC/Man had probably reached second year Maths at most universities.
Lack of scholarship training led to massive decline in standards. Basically IC covered a degree in 2 years and the 3rd was basically a masters. In maths, physics, engineering and physical chemistry enabled UK graduates at top universities to achieve very standards of mathematical and theoretical knowledge without the need for doing a masters and doctorates used to be completed in 3-4 degrees.Most other countries required a two masters and 4-5 year doctorates. The UK method saved on 2 yr masters and at least one of doctorate.
Those who took Maths O Level 1- 2 years early ( used to be Autumn O levels ) and destined for university scholarships at Ox/IC/Man had probably reached second year Maths at most universities.
Pre 1945 artillery and engineering officers were educated at Woolwich, the top 50% went to Cambridge and completed an engineering degree in 2 years. RN officers were also taught maths to high standard at Dartmouth. Many people went to polys to study to at least HND standards in engineering and science.
I cannot help wondering whether the problem the U of EA has had is because of the lack of top rate mathematicians/statisticians when they began their research: this also goes for M Mann.
Were far more members of the middle and upper classes more numerate and technically trained 50 years ago and is this partly why so many have fallen for AGW and in-effective wind turbines and similar energy producing systems?