This is a bit of an odd project from NOAA and Cornell. A more practical name might be “spliced” but I think they just like being able to say:
“If it positively, absolutely, has to be extreme, use ThreadEx”
“SpliceEx” doesn’t sound cool.
At least they realize how surface temperature records get fractured. Of course, while the method is useful, the data representing extremes may very well be crap in some instances. The problem these guys have (and NOAA in general) is that they seem to have no idea of the conditions which contribute to the measurement extreme. They just data mine. For temperature records, UHI and microsite bias is an issue that has not been investigated on a station by station basis.
For example they thread together Baltimore’s three stations as one of the threaded records.
But as we know, Baltimore’s two most recent stations have serious temperature bias issues, as outlined here and here at WUWT. So what good are threaded records if you don’t know the quality of the extremes data?
The links section at the bottom of the article has all the data files. Visit the URL below to try out the interface. – Anthony
From http://threadex.rcc-acis.org/
Summary
ThreadEx is a project designed to address the fragmentation of station information over time due to station relocations for the express purpose of calculating daily extremes of temperature and precipitation.
There are often changes in the siting of instrumentation for any given National Weather Service/Weather Bureau location over the observational history in a given city/region. As a result, obtaining a long time series (i.e., one hundred years or more) for computation of extremes is difficult, unless records from the various locations are “threaded” or put together. This has been done, but different approaches and combinations of stations have resulted in confusion among data users and the general public about what constitutes an official daily extreme record.
In consultation with NOAA’s National Climatic Data Center (NCDC) and the National Weather Service (NWS), the Northeast Regional Climate Center (NRCC) has evaluated station relocations and built “threads” for 270 locations that are published in NCDC’s Local Climatological Data using NOAA daily data sets. The data sets used for this project include NCDC DSI3210, DSI3200, DSI3206 and DSI3205. In addition, NRCC has been able to extend over sixty station threads back further in time using daily data contributed by local NWS offices, state climate offices and regional climate centers. An ongoing process of adding daily data from the old NWS Climate Record Books is also underway. These data have been digitized, passed through quality control and used to extend threads back even further in time for over fifty stations to date (see revision history for details).
Methodology for Developing Threads
The record of a currently active station was used as the starting point for a station thread. This station’s current record was used as far back in time as possible, taking precedence over a closed station’s record during any periods of overlap. A search was conducted to identify other weather stations in the area that could be used to extend the thread further back in time. In this process, preference was given to Weather Service/Bureau stations (that were not themselves LCD stations). The thread was extended back in time as far as possible using NOAA daily data available in digital form. Partner input was solicited and this local expertise was used to fine-tune the station threads.
Version 5.1 – released 26-March-2010.
h/t to WUWT reader Pamela Gray
Can’t seem to get these links to work.
The revision history link takes you to a page that has data file and inventory links.
They are broken.
But all they are doing is taking the worst affected station data (the most recent) in big UHI cities and making the historical record bounce up to meet the bad data.
They are not doing what they should be doing: Correct for UHI.
Looks like a gold mine of data.
55 stations within 50 miles of San Antonio.
Beeville (arguably one of the best sites in Texas) is not listed.
Now to download and compare to B91’s.
And once the record has been run back as far as possible, it will be homogenized. In other words, the farther back the record goes, the more the data will be adjusted downwards, suggesting serious warming is happening. Does anyone believe any of their products now? (I’m turning cynical in my dotage).
When I first saw this, all kinds of bells, whistles, and flags when up. Just eyeballing the data across states, I saw bias in record max temps for day and night that favored newer stations, while record lows favored older stations. However, occasionally, and in particular for Oregon, I saw record temps still standing for way back when. These “threads” are ripe for picking up bias and should prove to be a gold mine of information for those interested in station move bias. My mind designed a statistical analysis study that would blow the top off UHI straight away. These “threaded” data records could prove to be a point for our side.
I can’t understand why all the analysis is so focused on the absolute level of the temperature — when what we are really interested in is the change — ie are temperatures going up or down.
If we are looking at the level of change, I cannot see we need to splice at all. The analysis should surely be based on the % change between this years temperature and the next year for each year, and then the trend is the accumulation of these percentages. (ie there is also no need for all the controversy over the base either).
If the analysis was done this way then if a station changed site (or thermometer or anything else significant) you deal with it by having a two period break — and even don’t try to link the two. (ie if the change happens in year X then the first % change record goes up to year X-1, and the second starts at the % change between X and X+1)
I realise this would not deal with longstanding systematic changes like UHI, but it would stop the need to make all these adjustments which revise history backwards all of the time. It would also mean that reasonably short-lived thermometers could be used (though I presume there would be a minimum length below which it would not be worth going.)
If this is totally garbage — could someone please explain to me why … because the approach being used at the moment — with all its attendant problems — seems so counter-intuitive to me.
In data processing terms, it’s not a bad idea. If the thread carries the metadata about what changed (instrumentation change, site move, asphalt repaving…). Each event creates a thread and potentially each thread could have it’s own unique adjustment curve. That way the adjustments are kept parallel with the the raw data and not in the computer programs of the various modelers. It might reduce the need to synthesize missing values from other stations. It would be possible, if you have the station history and the time and knowledge to build an adjustment series for the station. For example every 10 years when the census is taken, you create a new thread with a population (UHI proxy) adjustment. Or a volcanic eruption event. And it there’s bad siting at a station, a fine grained adjustment can sit next to the raw data in the thread.
Devils in the details, of course but conceptually it’s a pretty clever way to retroactively add information from site audits (surfacestations.org) and historical information. Not fail proof, by any means but better than what we have now and reasonably extensibile. If Katla was to erupt and affect temperatures for some stations and not others or by varying degrees, that could be noted, site by site in ‘threads’
Yes, anomaly can be interesting, but range is a more useful metric to track climate change. We live in flora and fauna, and that flora and fauna has developed adaptations to the climate range it resides in. Flora and fauna that cannot survive that range dies out. All other metrics simply demonstrate weather pattern variation with each climate zone. Knowing what the range is doing can tell us whether or not climate zones, and not just weather pattern variations, are indeed changing.
It is also useful to note that daytime Tmax and Tmin, along with nighttime Tmax and Tmin may not be true range values if taken at the same time of day or night. The actual max or min might have occurred before or after the temp was recorded, unless the system of recording is continuous. My back porch thermometer automatically records and stores the max and min, regardless of day or night or clock hour. If I were I to write the reading down at noon and midnight, I may not be able to capture the true range.
For sustainable agriculture, range is critical. If absolute range is truly changing, then I need to make large changes in what I produce or I will not be in business very long. Otherwise, within season variations (IE anomalies) related to all weather pattern variation parameters tell me whether or not I need to irrigate more often (among other things of course).
Range is also a more direct identifier of possible UHI, much more so than temperature anomaly in my opinion.
“Tom in Texas says: July 14, 2010 at 8:26 pm
Looks like a gold mine of data.”
Yep, lots of data in one place – only $170.
I don’t like this Thread-Ex stuff. Back in the old days, most of the temperature records for the major metro areas were taken downtown on rooftops, where it’s artificially warmer. For instance, the old Custom House rooftop site used to run several degrees warmer than BWI until they shut it down. How can you compare records from BWI to the Custom House rooftop? If it hit 105 at BWI, it probably would have been like 109 at Custom House.