I hope that all is well with you and your family, and I am looking forward to your next article.

Wishing you a very nice weekend. ]]>

Just seen this. Now approved.

Nic ]]>

this is beneath you.

I provivded a link, one that returned hundreds of thousands sites.

Now, be a big boy and refine my query if you dont wish to wade through the larger body of results.

As if my point is any less valid if there are only tens of thousands of citations?

Geez mosh, its just sad.

You present a selected subset of three responses and attempt to argue that the term isnt commonly used in peer reviewed literature?

That it doesnt exist in the current lexicon?

That my contentions were in some way incorrect?

“I cited the number of hits for that phrase to make the plain that it is widely used in peer reviewed articles published by those who practice applied statistical analysis,

that its been the subject of quite a bit of research,”

Seems like 96 or 220K is relevant 220K is rather Yamal like

More funny is that I use the very link provided to do the search and get condemn for having bad faith.

too funny

]]>I have used values of h of 50 and 5 which gives a look, respectively, at low and higher frequency smoothing of the original temperature reconstructions for the regions of Arctic, Antarctica, Asia, Australasia, Europe, North America and South America. The series analyzed here were for the periods of 1500-2000 for first 5 series above and for 1500-1979 for North America and 1500-1995 for South America. It should be noted here that these reconstructions are for land only regions. Pages 2 K has rendered the data used elsewhere for these reconstructions into reconstructions of their own using 3 methods for reconstruction which are namely: Composite after centering and standardizing, Bayesian Hierarchical and Paired Correlation. This exercise by Pages allow us to see what differences can arise from the choice of reconstruction method. The Pages data is linked here: ftp://ftp.ncdc.noaa.gov/pub/data/paleo/pages2k/DatabaseS2-Regional-Temperature-Reconstructions.xlsx

I also added a presentation of the LWP smoothed series using the kernel type Epanechnikov in place of the normal kernel type that was used in all the other presentations. That provides 3 parameter variations in looking at temperature reconstruction data, i.e. smooth, kernel type and reconstruction method.

Here as like for the Northern Hemisphere analysis shown above and the claims in the Abram paper, the low frequency presentations of these regional reconstructions (h=50) shows a rather smooth trend from the 1800s time period up to the near present time – except for the Antarctica series which we might expect to be different. The higher frequency presentations (h=5) show structure in the series which in turn indicates a meandering series with upward and downward rises and dips into a time much closer to present time and then a regime change upward in some cases as early as 1900 or somewhat later followed in a most regions by leveling off of that regime trend upward.

The kernel type used differently for the PaiCo reconstruction shows that that choice can make a subtle difference in showing smoothed series structure.

I was surprised by the differences that the Pages 2 K application of the 3 different reconstruction methods could make in the final presentation of the reconstruction, and particularly with the longer term trends. I rechecked my calculations to insure I had not made an error here. The link here: https://www.blogs.uni-mainz.de/fb09climatology/files/2012/03/Pages_2013_NatureGeo_Sup.pdf explains the 3 reconstruction methods used in this comparison.

I have not thoroughly searched the literature about these method comparisons but I would be interested in explanations for the differences.

https://www.dropbox.com/s/11ssbozjyjfwb0t/Poly_Smooth_Reg_Recon.xlsx?dl=0

]]>I have recently been looking at publicly available global temperature reconstruction data and found that most of the series as presented are already smoothed. In fact the Northern Hemisphere plot I showed above in my previous post has an obviously smoothed plot from Mann-Jones 2003 which is further smoothed in my analysis. It becomes a difficult to impossible task to show variations from different smoothing functions once the series is already smoothed.

I have also been looking at regional temperature reconstructions from Pages 2K where the Pages 2 K people have taken proxy data from previous reconstructions and used 3 different methods of reconstructing these data and to present in plots. The methods were: Composite after centering and standardizing, Bayesian hierarchical and PaiCo. Those methods for some regional reconstructions can produce very different looking plots of the reconstructions. I would like to show those differences in this thread if you are around, Nic, to take this future post with a link to my Dropbox out of the inevitable moderation. I also show some differences that can occur in plot appearances when the kernel type is change in the locally weighted polynomial function used in SiZer.

]]>Sorry, I’ve not looked at this thread for a few days. Now approved. ]]>