GHCN v3.3 Regions Closeup Current vs Baseline

In a prior posting I made a nice area proportional global map of stations in current use (then 2015) vs those in the Baseline (more or less as GISS uses it). The only problem is that it is hard to see just what stations come and go in any given continent / country. Then, in another posting, I made “scatter graphs” of all the temperatures by region, but that still doesn’t tell you what came from where. (I’m working on a way to make the baseline vs current more visible but so far have an incompatibility block on Python where I can either get to the database OR have the statistics and color facilities I want from Python 3.x)

So for now I’ve made “Now” (the 2015 most current data in GHCN v3.3) vs Baseline maps for each “Region” (roughly continent) with the exception of Ships At Sea (since there are none currently in use). This, at least, lets you look at each Region and see just what is being measured and what it is being compared to.

So, with that, I’m just going to paste the maps in without much comment. Do note that I “cut off” a half dozen stations at the far end of Russia on the Asia region since they ran over into “the other side of the globe” and I’d need to have the whole globe shown again, at least on the E/W axis. For Region 5, Australia / Pacific, the same problem arises but with a lot more stations. So for them there is a Region 5.1 map and a Region 5.2 map. One for the positive longitudes and one for the negative.

Each map has variable scale. I make no attempt to keep things area proportional nor dimensionally accurate. (Well, for Region 5.2 I did, keeping a 2:1 ratio of Long:Lat as a bunch of islands doesn’t have an outline you can recognize). The goal was separation of the dots, not natural proportions.

These are “Now over Baseline” with transparency in the “Now” that are red. This means overlapping dots ought to turn purple in color. There are some of those in high density places, but much better separation than in the global view.

Region 1 – Africa

Region 1 - Africa Now Over Baseline GHCN v3.3

Region 1 – Africa Now Over Baseline GHCN v3.3

It looks like the blue thermometers are abandoning the stable Tropical and Temperate areas and migrating to the rim of The Sahara… I’d put a smiley on that as it is intended as a humorous metaphor, except that it does look like that as a source of bias. Those blue areas will have “temperatures” fabricated from the red areas via the Reference Station Method, then compared to make the “anomalies” that are used to scare the children with Global Warming Boogeymen.

Region 2 – Asia

Region 2 - Asia Now Over Baseline GHCN v3.3

Region 2 – Asia Now Over Baseline GHCN v3.3

Russia & India always very sparse. China got a good thinning out. (I wonder if the ones kept are in major growing industrial centers…) Then Japan and South Korea, even in close up, are mostly blue. What’s that all about? Zooming in for a closeup:

Region 2 Japan South Korea Close Up Now Over Baseline GHCN v3.3

Region 2 Japan South Korea Close Up Now Over Baseline GHCN v3.3

It does look rather like abandoning the interior while keeping coastal sites. Interior South Korea can be darned cold. I wonder how high the mountains are there? Japan has a row of mountains down the spine, but it does look like some red dots survive there. Enough? Who knows.

Region 3 – South America

Region 3 - South America Now Over Baseline GHCN v3.3

Region 3 – South America Now Over Baseline GHCN v3.3

Brazil very under sampled. The Spanish Language coastal areas over sampled. Erosion at the tip of Patagonia.

Region 4 – North America

Region 4 - North America Now Over Baseline GHCN v3.3

Region 4 – North America Now Over Baseline GHCN v3.3

What can I say? Canada and Alaska a mess. Mexico looks like most of it is gone. Ditto Central America and a bunch of the Caribbean. Looks like the Texas Panhandle got dropped, but lots of data from the Bos-Wash corridor and down into Appalachia. California looks like the mountains get tossed out along with down near San Diego, while some coastal areas are kept.

Region 5 – Australia / Pacific

Region 5 - Australia Now over Base GHCN v3.3

Region 5 – Australia Now over Base GHCN v3.3

Looks like a lot was dropped in what the Australians like to call mountains ;-) (Hey, Sierra Nevada and Rockies guys… Real Mountains ™ ;-) then also New Zealand looks to have dropped mountain stations and kept coastal. I think that’s a pattern… GHCN seems to regularly keep high cold places in the baseline and coastal mild stations in the “now”. Can you really compare the tops of the New Zealand Alps with the sunny beaches? Even with anomaly Reference Station Fabrication? AND have 1/10 C accuracy?

Region 5 - Americas South Pacific Now over Baseline GHCN v3.3

Region 5 – Americas South Pacific Now over Baseline GHCN v3.3

Didn’t know exactly what to call this chunk. Polynesia covers too much. South Pacific is a bit vague. I chose “Americas” as it is closer to the Americas than to Asia, just to make it clear what side of the Pacific it is on. ( I know I could use East / West but they change depending on where the viewer stands…) It would be interesting to get current data for the islands that were dropped (blue) and see if they don’t show any warming…

Region 6 – Europe

Region 6 - Europe Now over Baseline GHCN v3.3

Region 6 – Europe Now over Baseline GHCN v3.3

It looks like they are comparing Portugal to Spain and Poland to Germany, given when there are Baseline vs Now stations. Then Turkey and the Austro-Hungrians get a good thinning out. Ireland runs to the coast, while the UK gets the blues.

Region 7 – Antarctica

Region 7 - Antarctica Now Over Baseline GHCN v3.3

Region 7 – Antarctica Now Over Baseline GHCN v3.3

You can sort of make out the Peninsula and the perimeter of Antarctica. Not a lot of deep inland data.

In Conclusion

I’m not going to include all the code. It is basically that prior global map with Lat vs Long plotted, but with the area displayed adjusted and a regional filter in the sql statement. Here’s the bits that matter:

sql="SELECT I.latitude, I.longitude FROM invent3 AS I 
INNER JOIN temps3 as T on I.stationID=T.stationID 
WHERE T.region='5' AND year>1949 AND year<1981 AND I.stationID 
NOT IN (SELECT I.stationID FROM invent3 AS I INNER JOIN temps3 as T 
ON I.stationID=T.stationID WHERE  year=2015)GROUP BY I.stationID;"

Note I’m not plotting the cosine latitude, just regular latitude. Then you plug in a region number on the filter and adjust the dimensions of the graph displayed:

plt.xlim(-180,-90)
plt.ylim(-60,30)

Per what the maps tell me: overall, it looks to me like way too much playing with the thermometers and not enough comparing A Good Thermometer in the past to THE SAME Thermometer in the present.

IMHO, any thermometer not in the present needs to be removed from the comparison “Baseline”. It is no longer a real instrument.

Subscribe to feed

Advertisements

About E.M.Smith

A technical managerial sort interested in things from Stonehenge to computer science. My present "hot buttons' are the mythology of Climate Change and ancient metrology; but things change...
This entry was posted in AGW Science and Background, NCDC - GHCN Issues and tagged , . Bookmark the permalink.

11 Responses to GHCN v3.3 Regions Closeup Current vs Baseline

  1. KeefInLondon says:

    Blimey!
    This is completely bonkers. How on earth can any one pretend that the changes to the stations used do not significantly influence the results?
    Well done EM.

  2. E.M.Smith says:

    @KeefInLondon:

    Especially when it was preferentially stations in “High Cold Places” that got nuked:
    https://chiefio.wordpress.com/2019/02/12/ghcn-v3-3-stations-by-altitude-by-years-or-mountains-what-mountains/

  3. Bill in Oz says:

    Congratulations EM. Yes, the same pattern is clear again & again. The global temperature is based on an unrepresentative sample which is heavily biased towards the parts of the Earth where humans live.

    And as we already know, in those parts of the planet we have an impact on the temperature as well via the UHIE wffect..

    By the way the Bureau of Misinformation has just released another guide to ‘past’ temperatures in Australia with ‘homogenisation’ meaning that the past in Australia was cooler than thought before.

    However this is based on only 57 weather stations that go back to 1910. I wonder what your map would show for Australia if those 57 stations were plotted.

  4. H.R. says:

    Keef: “How on earth can any one pretend that the changes to the stations used do not significantly influence the results?”

    Oh it’s easy enough when your bloated paycheck depends upon it… and you have no conscience… and you believe the end justifies the means.

    Doesn’t hurt if you’re a useful idiot, either.

  5. Steven Fraser says:

    @EM: This is fine work. Bravo! I think it would make an excellent presentation, too.
    What happens if you isolate the stations which are removed post-baseline (the trimmed ones), and remove them from the baseline, too? For example, does it move the baselined temp average?

  6. Steven Fraser says:

    @EM, continuing… as if they were never there to begin with. Not thinking so much about the red v blue, but the effect on the monthly temp anomalies.

  7. E.M.Smith says:

    @Steven:

    I tried to do that once in about 2011 with GIStemp, but the code was so fragile it blew up if you removed too many stations. Figured I’d get back to it someday… Now GIStemp has been rewritten in Python, so maybe less full of historic artifacts and more sturdy…

    Part of why I’m bothering with python… for a re-porting effort, maybe.

    The big issue is that each entity (GISD, NCDC, NOAA, Hadley, Aus. BOM,…) does their own version of The Reference Station Method, so IF you roll your own, they say it isn’t like theirs, and they don’t all share the code…

    I’m thinking the first step would just be showing trends in the two sets of stations. The problem there being to get current data for the dropped stations…

    So at the moment I’m working on a temp trend using ONLY the present stations. Did a rough draft DB layout this morning. Figuring out processing now…

  8. Steven Fraser says:

    @EM: Interested to see that, too,

  9. Steven Fraser says:

    …Testing the hypothesis that the baseline is currently biased low in comparison to the current.

  10. Octave Fiddler says:

    IS it just me? I am not seeing any ‘purple’ dots, where the red overlays the prior blue dot location.

    Glad to see Cheifio back in the saddle again.
    I still remember those graphs with ‘hair’…

    If I had to guess, I would imagine that those ‘fugitive’ Higher Altitude stations are rather far from nearest neighbor, and thus a lot of ‘areal’ coverage is getting discarded;
    (and replaced by the homogenizer with warmer stations).
    i.e., those HA stations have disproportionate influence compared to their modest numbers,
    and their removal is (as you suggest) a bigger dealio than might appear ‘at a glance’.

    Am sure you are working on a way to spread out that NA data.
    No way to see how many are in the LA group compared to the HA group.

    Really happy that you’ve decided to again focus your unique talents on this issue.
    Cheers
    OF

  11. Octave Fiddler says:

    So, do those dropped stations stop recording data?
    (Why would they stop in sync with Hadley?)
    Or do they still exist and are being ignored?
    YIkes
    OF

Anything to say?

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.