17 Thoughts on “WUWT mentions the reanalysis

  1. Where Anthony Watts gets it wrong yet again and pastes Australian data into the post.
    And you get a hearty endorsment from Ken Ring.

  2. Richard C (NZ) on November 3, 2014 at 4:50 pm said:

    Steve McIntyre commented at BH. I don’t think he’d read the paper and he’s asking for code which RS93 never supplied, neither SI. Why then de Freitas et al for which there is not necessarily code as RS93? The method is there to be implemented in whatever way, not necessarily code. I note Spencer and Braswell (2014) used both Fortran and Excel to get the same results.

    He’s also asking for data on an ftp server. Given it’s readily available in CliFlo for which I gave him a link, I would have thought that’s unnecessary. Whether he goes from “semi-interested” and “insufficiently interested to try” to interested enough to replicate remains to be seen. I hope i haven’t antagonized him but what he’s asking is superfluous to my mind, although granted he operates in a different and loftier sphere than I do. Perhaps I should stop helping in that respect.

    http://www.bishop-hill.net/blog/2014/10/31/new-zealands-temperature-record.html

  3. Still, that’s no fault of the paper.

  4. Richard C (NZ) on November 3, 2014 at 5:32 pm said:

    I’ve thought about coding RS93 and PM-95 (ACORN) but I’m not skilled in any modern scripting and probably too late now to pick up from donkey’s years ago. Don’t think I have the stats nouse anyway which is why I backed out early on when it became obvious that the rigour of RS93 was required. It would be very useful though to have these breakpoint analyses techniques coded in software to run against other series e.g. PM-95 against 7SS and RS93 against ACORN. BEST’s Scalpel is another one. Few of us are developers or fluent with stats but most can be users when an easy tool is at hand for rudimentary demonstration.

    RS93 looks a bit daunting at first but I’m wondering how the process can be eased by modern IT techniques rather than low level coding e.g diagrams that automatically generate code (think Bizagi for business) or modular pseudo code, pseudo English, or something. Not sure of the appropriate jargon nowadays, let alone higher level tools. I’m thinking in terms of making the stats simpler so that more people understand the breakpoint analysis process. Huub Bakker swears by VB6 but I’m wondering if there’s a way less time consuming and laborious for which less specific skill is required..

    BOM apparently has code for PM-95 but to my knowledge it hasn’t been released, probably because they don’t want it used in the above manner. That would be really worth picking up and its release was promised. I thought of this when challenged at JoNova to run PMOD (solar, TSI) through breakpoint analysis because I raised the idea when discussing a PMOD breakpoint that I thought was readily identifiable by eye here:

    http://www.woodfortrees.org/graph/pmod/from:2005/to:2008

    In other words, these tools would not be restricted to temperature series as for example EMD is not restricted to say radio signal analysis and can be just as easily used on climate series for which it is very useful. I much prefer off-the-shelf tools simply because of time constraints and, how shall I put this, my amateurism.

    There’s Quartile Matching (not Percentile Matching as in PM-95) breakpoint analysis software, RHtestsV4. It’s an implementation for North America, User manual here:

    RHtestsV4
    http://etccdi.pacificclimate.org/RHtest/RHtestsV4_UserManual_20July2013.pdf

    From this thread:
    http://jennifermarohasy.com/2014/09/rutherglen-still-looking-for-answers/#comment-564294

    Possibly no substitute for skill, knowledge and hard work but I’m inclined to think there’s a pathway that hasn’t been exploited yet.

  5. Steve McIntyre, Steven Mosher and all those guys are all “R” programmers, (which I don’t know) so it would actually be really helpful if someone got some R code and data out in the public domain

  6. Richard C (NZ) on November 3, 2014 at 8:42 pm said:

    Yes, I see this in RHtestsV4 users manual

    “This simple manual is to provide a quick reference to the usage of the functions included
    in the RHtestsV4 package (and also to the usage of the equivalent FORTRAN functions,
    which are available by sending a request in English to [email protected]). Users
    are assumed to have the general knowledge of R (how to start and end an R session and
    how to call an R function).”

    Plenty of tutorials e.g Getting Started with the R Data Analysis Package
    http://heather.cs.ucdavis.edu/~matloff/r.html

    But I’ve yet to discover BOM’s code i.e. whether it’s R or something else. And if BOM has a bespoke package then is it really necessary to know R anyway except how to start it?

    The key is the stats though. I’ve never coded anything like that. My experience, years ago now, was information crunching in business situations – RDBMS in COBOL with embedded SQL at most academically, then just ACCESS RDBMS commercially. I did do a lot of systems analysis diagramming manually because of my drafting background before the automated tools came in which I’ve used a bit since but not commercially.

    I think modular analysis by diagram is more important than code to get to grips with a system. or technique in this case, in terms of ease of communication to non-stats, non-coders. There’s diagram standards too.

    I had just assumed Bob did all the work in Excel even though I knew he was a programmer. I didn’t think of code until people started asking for it. Even if there was code I’m not convinced code needs to be produced for replication in this particular situation. The method is in R&S (1993) so anyone can do what de Freitas et al did but by a number of alternative techniques as they wish.

    As I see it, the nature of replication is the corroboration, or otherwise, of parallel alternative techniques. From the method, the alternatives should return the same results as per Spencer & Braswell upthread i.e. different alternatives can be implemented within the same paper or different papers – there’s no limit to technique alternatives and no stipulated code specification.

  7. Richard C (NZ) on November 3, 2014 at 9:35 pm said:

    While I’m at it, I’ve laid out the ACORN-SAT methods at JN here (#38.2):

    http://joannenova.com.au/2014/08/hiding-something-bom-throws-out-bourkes-hot-historic-data-changes-long-cooling-trend-to-warming/#comment-1554309

    Following on in the next comment (#36.2.1) is a percentile-matching (PM-95 for breakpoint analysis) definition and tutorial. We are dealing with “parameter estimates”, with no guarantee that an estimate will be unique.

    Also discussed here:
    http://jennifermarohasy.com/2014/09/so-much-conversation-so-little-evidence/#comment-564149

  8. You know Huub Bakker? I worked for him for a while, a good man. Well, two levels under him, to be fair.
    I think it may be a different Huub, actually.

  9. Actually, I quite appreciated Ken’s post, did you read it, Simon? About alleged changes in pressure levels around NZ due to climate change.

  10. Richard C (NZ) on November 4, 2014 at 8:59 am said:

    >”You know Huub Bakker?”

    I don’t know him but he used to frequent this blog. Commented a bit recently at JoNova in the solar model threads. That was where I saw he uses VB6 for teaching:

    Dr Huub Bakker – Senior Lecturer – Massey University
    http://www.massey.ac.nz/massey/expertise/profile.cfm?stref=944130

    Same Huub?

  11. Richard C (NZ) on November 4, 2014 at 9:04 am said:

    I read it too. Can’t say I know much about Ken Ring, controversial I gather, but his post on historical NZ pressure was interesting.

  12. Richard C (NZ) on November 4, 2014 at 9:15 am said:

    I can’t see much to be gained by applying the same breakpoint analysis technique to a series by multiple parties for replication. It’s been done by de Freitas et al and someone can replicate it if they want to verify. OK it needs to be done but what does that really prove?.

    It’s the different methods that would really be revealing on the same series – RS93 vs PM-95 vs Scalpel for example. NIWA vs NZCSC vs BOM vs BEST. We’ve got NIWA vs NZCSC now and that’s a real eyeopener but I think there’s many more surprises in store.

    Then there’s GISS. What’s their method?

  13. Richard C (NZ) on November 4, 2014 at 10:23 am said:

    >”NIWA/NZCSC vs BOM vs BEST [vs GISS]”

    Apart from breakpoint techniques, the respective homogenization methods are vastly different too. But there are common breakpoints. Then there’s non-site change break adjustments by the others extra to those in NIWA/NZCSC – messy.

    In Anthony’s post he featured Rutherglen Research (for some reason) but BEST has a much shorter series for Rutherglen Research. Their earlier Rutherglen data is not homogenized to Rutherglen Research, they go to Rutherglen Post Office.

    RUTHERGLEN RESEARCH
    BEST Mean http://berkeleyearth.lbl.gov/stations/151882
    BOM Min https://wattsupwiththat.files.wordpress.com/2014/11/rutherglen1.png?w=720

    Earliest Observations: 1965 vs 1912ish

    BEST makes an adjustment at the 1980 Record Gap but as far as I know BOM doesn’t. In the immediate vicinity of Rutherglen, BOM has a continuous series, BEST has a gap 1921 – 1965.

    RUTHERGLEN POST OFFICE
    http://berkeleyearth.lbl.gov/stations/3816

    Earliest Observation: Jan 1903
    Most Recent Observation: Nov 1921

    The NZ 7SS is clean by comparison to Rutherglen, BOM vs BEST.

  14. Richard C (NZ) on November 4, 2014 at 10:49 am said:

    BEST runs Albert Park and Auckland Aero in parallel:

    AUCKLAND, ALBERT PARK http://berkeleyearth.lbl.gov/stations/157062
    AUCKLAND AERO AWS http://berkeleyearth.lbl.gov/stations/157061

    NIWA/NZCSC drops Albert Park when Aero starts, homogenizing Albert Park to Aero.

    BEST Albert Park is above NZCSC Auckland but below NIWA Auckland. If BEST were to adjust for UHI/sheltering, their Albert Park series would corroborate de Freitas et al (2014)/’Statistical Audit’ Auckland but nowhere near NIWA Auckland.

  15. No, different, sorry.

  16. Richard C (NZ) on November 4, 2014 at 1:32 pm said:

    Hey Andy, Huub Bakker’s wind turbine/human effects work:

    Books

    Dickinson, P. (2010). The sounds from wind turbines. Theory, practice, assumptions, and reality.. In B. Rapley, & H. Bakker (Eds.) Sound, Noise, Flicker and the Human Perception of Wind Farm Activity. New Zealand: Atkinson and Rapley Consulting
    [Chapter]Edited by: Bakker, H.

    Bakker, HH., & Rapley, BI. (2010). Sound Characteristics of Multiple Wind Turbines. In BI. Rapley, & HHC. Bakker (Eds.) Sound, Noise, Flicker and the Human Perception of Wind Farm Activity. : Atkinson and Rapley Consulting
    [Chapter]Authored by: Bakker, H.Edited by: Bakker, H.
    Read Abstract: abstract icon

    Rapley, BI., & Bakker, HH. (Eds.) (2010). Sound, Noise, Flicker and the Human Perception of Wind Farm Activity. Palmerston North: Atkinson & Rapley Consulting Ltd (Palmerston North, New Zealand) in association with Noise Measurement Services Pty Ltd (NMS) (Brisbane, Australia)
    [Edited Book]Authored by: Bakker, H.Edited by: Bakker, H.
    Read Abstract: abstract icon

    Bakker, HH., Bennett, D., Rapley, BI., & Thorne, R.Seismic Effects on Residents from Wind Turbines. In BI. Rapley, & HHC. Bakker (Eds.) Sound, Noise, Flicker and the Human Perception of Wind Farm Activity.
    [Chapter]Authored by: Bakker, H.Edited by: Bakker, H.
    Read Abstract: abstract icon

    http://www.massey.ac.nz/massey/expertise/profile.cfm?stref=944130

  17. Richard C (NZ) on November 4, 2014 at 8:32 pm said:

    ‘Evaluating the impact of wind turbine noise on health related quality of life’

    Daniel Shepherd, David McBride1, David Welch2, Kim N. Dirks2, Erin M. Hill (2011)
    Department of Psychology, School of Public Health, Auckland University of Technology, Auckland,

    1Department of Preventive and Social
    Medicine, University of Otago, Dunedin,
    2School of Population Health, The University of Auckland, Auckland, New Zealand

    Abstract
    We report a cross-sectional study comparing the health-related quality of life (HRQOL) of individuals residing in the proximity of a wind farm to those residing in a demographically matched area sufficiently displaced from wind turbines. The study employed a nonequivalent comparison group posttest-only design. Self-administered questionnaires, which included the brief version of the World Health Organization quality of life scale, were delivered to residents in two adjacent areas in semirural New Zealand. Participants were also asked to identify annoying noises, indicate their degree of noise sensitivity, and rate amenity. Statistically significant differences were noted in some HRQOL domain scores, with residents living within 2 km of a turbine installation reporting lower overall quality of life, physical quality of life, and environmental quality of life. Those exposed to turbine noise also reported significantly lower sleep quality, and rated their environment as less restful. Our data suggest that wind farm noise can negatively impact facets of HRQOL.

    References

    21. Bakker HH, Rapley BI. Sound characteristics of multiple wind turbines.
    In: Rapley BI, Bakker HH, editors. Sound, noise, flicker and the human
    perception of wind farm activity. Palmerston North, New Zealand:
    Atkinson and Rapley; 2010.

    https://www.michigan.gov/documents/energy/NAH_2011_419860_7.pdf

    Hmm, Michigan Govt documents.

Leave a Reply

Your email address will not be published. Required fields are marked *

Post Navigation