I have downloaded all my data directly from the NIWA database and I see that there are one or two corrections made this year. I have found in the climate debate that it pays to get as close as possible to the source. It was the station ID’s that I was particularly after as there was no way to tell which stations to include otherwise. Richard pointed me to the paper on the NIWA site and I am now following that.

I’m somewhat surprised that my initial picks are so different from NIWAs. I chose the oldest station as a base and then looked for any stations within a 10km radius, picking the ones that were closest and provided the necessary coverage over time. I know that some stations are more reliable than others but I don’t know how to find that from CliFlo; the descriptions as 1:Climat (Standard) etc don’t tell me much.

When I have something that looks reasonable (i.e. I’ve got all the data correct, located the possible step changes, have applied the FDM method correctly and don’t have any obvious errors) I thought I’d see if I could get it posted on the blogosphere somewhere. I don’t have the time, inclination nor pedigree to write it up as a proper paper and I’ve found that the blogosphere provides lots of comments, much of it useful. I will, of course make all my code available if that is the case.

I’m sticking with annual data at the moment but, from the comments on the ClimateAudit post (http://climateaudit.org/2010/08/19/the-first-difference-method/) there is utility in using the monthly data since it essentially gives you 12 estimates of annual differences albeit with more noise. Too much work at the moment though.

http://www.climateconversation.wordshine.co.nz/2010/08/why-cant-scientists-agree-on-global-warming/
or
http://tinyurl.com/28r3ce4

However, if you haven’t already I suggest you check out the paper the NIWA site refers to. Several different methods were used for calculating adjustments – average nearby station data was only one of them. And of course that would not be appropriate in every case.

Also, be a bit careful of Rchards data on the stations – check the NIWA database. I have found some difference between his data and the official NIWA data and suspect he has obtained some of his from different sites or sources (eg. monthly temperatures instead of annual, or old data which has since been removed).

What do you intend to do with your results? Any plan for inviting peer review? For publishing?

As a Professional Engineer with a PhD in Process Control, I have looked on with interest as questions are raised in New Zealand, Australia, the USA and elsewhere about the temperature record. I can see the necessity for making adjustments as well as the subjective decisions that need to be made along the way. What has surprised me has been the nature and magnitude of many of the adjustments as well as the uniform unwillingness of the various climate-monitoring organisations to publish the specific reasoning behind the individual adjustments. From where I stand I can see no legitimate reason for not publishing them and good reasons to do so. Surely not every single climate-monitoring organisation finds itself without documentation describing the adjustments!

Imagine my interest then when I read about the First Differences Method and its ability to objectively remove the effects of Time-of-Observation, site moves, equipment changes and other such step changes. I have implemented this on the seven centres noted in the Seven Station Series and can now see some of the issues in using this method. However, in light of the lack of an SOA there seems no other means of validating the SSS and the New Zealand Temperature Record. Certainly using an alternative method of creating a temperature series will provide a useful test.

I’m grateful to Richard for finding the station list for me and will attempt to replicate the SSS although I will also attempt to use as many of the neighbouring stations as possible since the FDM does not handle small numbers of stations very well.

As for doing this honestly, I’m doing it blindly at this stage, which amounts to the same thing!

About Ken. We’re old friends. I suggest you ask him:

1. Have you (finally) worked out your own SOA from NIWA’s raw data and what are they?
2. Have you read the post above, and have you noted the parts explaining how the methodology is inadequately described and not used by anyone?
3. Have you asked NIWA to describe the adjustments they made to our temp record and what did they answer?

And then welcome him back to the Climate Conversation. It’s nice to hear from him again.

NIWA provide a document titled The NIWA “Seven-Station” Temperature Series, dated 9 February 2010, which gives the station IDs (NIWA calls them Agents) for all the weather stations involved in the seven-station series. It turns out that about 32 individual weather stations contributed.

You can download the pdf from http://www.niwa.co.nz/__data/assets/pdf_file/0017/101834/7-Station_Temperature-Series.pdf
or http://tinyurl.com/254akhc

Curiously, I notice the link to this document at http://www.niwa.co.nz/news-and-publications/news/all/2009/nz-temp-record/seven-station-series-temperature-data
is shown under the heading “Schedule of adjustments” (which is what attracted me to it).

But it provides only the numerical adjustments — there are no reasons given for any of them.

Still, good of them to make this info available, even if it took pressure from ACT and the Coalition to make them do it.

You need to register and can then download all you want and play with it to your hearts content.

If you want to have a go at working out your own adjustments (which are necessary to get a record over time – don’t beleive anyone who clai9ms otherwise) there is a link on the NIWA site to publications which include methodologies for doing that.

You will have to make subjective decisions about which method to use in each case and it will take a lot of work. But if you do this honestly I am sure your results will be similar to NIWA’s.