Surfacestations.org hits the 1000 mark

I’m pleased to announce that the surfacestations.org project has now surveyed over 1000 of the 1221 USHCN stations in the USA, putting the percentage of the survey at over 82% now.My sincere thanks to the many volunteers who stepped up recently to survey additional stations in Texas, New Mexico, Arizona, Colorado, and many other states.

Here is what the coverage looks like as of 7-14-09

Here is the breakdown by state. Note that 5 states are now 100% completed.

Here is a chart to show the table data above:

Note the states that are lacking the most in coverage are Virgina, Missouri, Texas, South Dakota, Maryland, Wisconsin, West Virginia, Nebraska, Pennesylvania, Iowa, and Idaho. These states are all below 80%. The states I would most like to get more surveys from are Missouri, Virginia, and Upstate New York which has a cluster of stations untouched. Arrangements have been made to get three stations in Texas this coming weekend: Alice, Falfurrias, and Rio Grande City, so anyone who is considering Texas can cross those off the list.

For those that wish to help getting the final stations here is a Google Earth KML file that will help you locate the remaining stations yet to be surveyed in the USA.

Download Google Earth KML file here (sincere thanks to Gary Boden for his preparation of this)

Just download it (right click, save as) and then drag and drop to Google Earth, which can be downloaded free here.

The coordinates are mostly accurate, but it is always a good idea to get station location descriptions from NCDC’s MMS database also. Often they can be located easier by description than by coordinates. Note the the KML file has descriptions also along with the COOP ID number to help you get a match with NCDC’s database.

If you wish to help in surveying the remaining stations, go to the surfacestations.org project and complete the signup process.

I know that many people have been waiting for an analysis of the data. That is in process right now and a paper suitable for peer review is being prepared. I’ll answer the most obvious question ahead of time, and that is: no I will not be posting the results here first. After the paper has been set for publication, and within the rules of the journal, the paper, all of the data, methods, and results will be made public for anyone who wishes to replicate the work or to challenge it. There will be no hidden folders marked “censored” or incomplete MATLAB code. I’ll post the Full Monty once cleared by the journal.

That being said, given the tasks ahead for me, I’ll be posting far less frequently on WUWT. In the meantime, please be patient and let me finish this up with my co-authors.

Some people have wondered why I have taken two years now before going into data analysis mode. There are a couple of reasons.

1- Getting the best stations. The number of well sited stations are so few, getting enough to do a valid comparison to the poorly sited stations was a challenge.This is why I’ll continue to ask for additional surveys until we reach a publication deadline. There are so few “best” stations that even adding a handful more will be statistically significant. So please, keep up the surveys.

2- Coverage. I wanted to be absolutely certain that I had an undisputably large enough sample both in percentage volume and in spatial distribution. There were some folks who did some analysis using data from early in the project, such as John V at about 30% (with very poor spatial distribution) and the recent NCDC Talking points memo at 43%  NCDC “thought” they had the most current data, but they don’t have it, nor did they ask before attempting that analysis. That was an error on their part, and they are aware of it now. I’ve been in touch with the principal investigators at NCDC.

3- Patience is a virtue. If I had done analysis at 30 or 40%, as many suggested I do, and the analyzed results suggested that “siting mattered significantly” to the accuracy of the US Temperature record, I would be immediately vilified for having an inadequate sample, and rightly so. Interestingly, no such criticisms have been levied at NCDC by the AGW blog community for their results in the “talking points memo” at 43%, or at John V at 30%. Yet those results are being held up as examples of valid results by some. A double standard for statistical significance is something we’ve seen before in examples demonstrated by Steve McIntyre and others. Yet even without the statistical analysis, it is clear that the USHCN has not been well maintained. NOAA/NWS has closed many stations that we have highlighted, and even some we haven’t. Most recently Telluride, CO which is another story. If nothing else, this project is helping to get the USHCN network cleaned up. NOAA agrees in practice, as does NCDC, otherwise the US Climate Reference Network (USCRN) would not have been created nor would there be an HCN modernization program if the USHCN was in an acceptable condition.

I wish to thank everyone who has helped in making this project continue to the level of coverage it has. Regardless of the outcome of the analysis, whether it shows that siting matters or it does not, one thing can always be said with pride: this survey is a one of a kind volunteer accomplishment that NOAA couldn’t do themselves.

It has been a long road, fraught with roadblocks, frustration, and criticisms.  I appreciate everyone who has helped me along the road.