In the posting on Australia, in a comment by KeefInLondon (h/t) he suggested:
Very interesting. And surprising. So both Australia and Canada show a distinct reduction in stations around 1990. I wonder why? This is about the time the GHCN was created. Maybe the initial collection and load of the data was more extensive than subsequent additions?
Australia also has another discontinuity at 1910 (it looks like). Canada doesn’t though, and instead there is steady increase in stations, which is what you expect really.
It would be interesting to see if the same patterns show up in some other countries. It would be worth including a couple of small countries too.
Well, I’m going to do that. In fact, I’d already made a dozen or so graphs of other countries including some of those smaller ones. Some are interesting. However, it lacks context. So I’ve decided that first, I’m going to just post up ALL the data in EACH “region” (or roughly, continents plus ships at sea). That, then, gives you the comparison context for any given country. Each country will be a part of these continent scale blobs, so seeing the big trend will let you then look for “who is doing that” inside that blob as countries and groups get graphed.
I’m going to do them in reverse order, since Region 8, Ships, is the most interesting. But first, here’s the region number / name set:
mysql> SELECT * from continent; +---------+-----------+--------+---------------------------+ | version | ascension | region | region_name | +---------+-----------+--------+---------------------------+ | Gv3 | 7Sept2015 | 1 | Africa | | Gv3 | 7Sept2015 | 2 | Asia | | Gv3 | 7Sept2015 | 3 | South America | | Gv3 | 7Sept2015 | 4 | North America | | Gv3 | 7Sept2015 | 5 | Australia Pacific Islands | | Gv3 | 7Sept2015 | 6 | Europe | | Gv3 | 7Sept2015 | 7 | Antarctica | | Gv3 | 7Sept2015 | 8 | Ship Stations Ocean | +---------+-----------+--------+---------------------------+ 8 rows in set (0.05 sec) mysql>
I’ll put one copy of the code for one region at the end of the posting. With that, off to the graphs:
Region 8 – Ship Stations in the Ocean
I’ve been playing around with coloring the dots by winter vs summer with the month associated reversing in the southern hemisphere. For these graphs, I’m leaving January BLUE and June RED just because South America and Africa span the equator. Only Australia / Pacific and Antarctica will be “backwards” in color with winter RED / summer BLUE. So just keep that in mind. For the Ships, I think most of the stations are well into the northern hemisphere so the color scheme is appropriate.
THE biggest thing is just that there isn’t much ship data and it doesn’t span much time at all. Roughly a bit over 20 years. So it will show up in the “Baseline” used by GHCN and Hadley, but will have no comparable data to compare to it. Whatever is compared will be from something else. Unless satellite data are added, that will be a land station on an island almost certainly at an airport now serviced by jet aircraft.
Note that the scale on the blue graph above starts at -10 C while this one starts at 5 C. I could fix all the scales at one wide width, but then it is harder to see the dispersion / changes as most temperatures would be in a tight band. This spreads the dots out more so you can see more detail. This matters more in the more dense graphs below.
Notice that nothing gets much above 25 C / 77 F. At about 84 F hurricanes / cyclones form and move massive amounts of heat to above the Troposphere. Ships also tend to stay away then and certainly don’t stop to take the temperature when the wind is kicking up and the waves turning to monsters.
I question what those ships were measuring when they measured something below 0 C. Perhaps ice breakers in the Arctic… So a small “dig here” for LAT/Long and such on those data points.
Since the overall lifetime of data is nearly nothing, it’s hard to say much about it other than that it isn’t here now so any Ship data is rather useless as a baseline and can only be compared to fabricated fantasy values.
Region 7 – Antarctica
In the region of the Baseline (1960 to 1990) there are a bunch of data between -20 C and -10 C yet nothing after that. Most likely some station was abandoned, or no longer reporting, or just dropped from the record. There are a couple of outlier data at about 5 C to 8 C which, given that ALL of Antarctica is ice, must be an error or measuring jet exhaust or near a building. (Or perhaps exposed to the direct sun, January being summer there). Other than those there is NO rise of the hot end above “just about 0 C”.
Again we see that the top end doesn’t move. Hanging around just about -3 C. Between 1960 to 1980 we have some “middling way cold” at about -40 to -30 that is missing in the data after 1990. A “step change” like that is NOT slow global warming, that’s instrument change.
Region 6 – Europe
My comments will become a bit more thin from here on down. Partly as the above graphs are most interesting. Partly as the things observed become repetitive. Top ends NOT moving steadily to the right. Loss of a “middle cool” chunk that is present in the Baseline and missing in the data after 1990. Not much in the way of “skew” of the mass of data.
Overall, a “scatter / gather” as thermometers show up, then suddenly get pruned about 1990, then an interesting drift of colder happening about -10 C after 2000 A.D. It looks like Europe had a few cold winters recently. Then any “warming” isn’t at the top end, it is from loss of cold data from the Baseline stations.
Here we DO start to see a bit of skew to the data. The top end does rise and the center of mass slowly drifts a bit right. But it happens in 2 steps. First as a step change in about 1950 as the Baseline stations got added, then after 1990 as station change continues. There is also a step change warmer on the low end with the thermometer changes of about 1990, but not much after that. I would suspect mostly Urban Heat Island effects as Europe is very densely populated.
Region 5 – Australia / Pacific Islands
This will include New Zealand and a bunch of islands scattered over the Pacific, but the mass of it will be Australian stations.
This first, January, graph has ONE dot about 1998 at -20 C that is certainly an error. I’ve left it in, even though is squashed the rest of the graph off to the left. Remember that, though blue, this is the summer season in Australia.
Again a very clear lack of rise at the right hand side. Loss of stations about 1990 thins out data between 10C and 20C.
For the Australia winter, we can see a bit of bifurcation into those places that get cold in winter, and the tropical islands that don’t change. There are a few outlier high reports about 35 C about 2000+ that I can’t explain, but would bear investigating as likely errors, jet wash at the airports, or perhaps a time of still air and solar heated tarmac at the airports. Overall, there’s a pretty hard wall at 30 C or 86 F where Cyclones remove the heat.
There is a block of about 0 C to 10 C temperatures in the Baseline years that “goes away” when the thermometers are removed.
Region 4 – North America
Given the density of stations in North America, it is particularly interesting. Again we have a hard wall at the warmer right hand side, and drop out of middle cold stations from the baseline era.
There’s a little bit of visible slew of the right side into the baseline era, then it just thins out into dots and loses pattern. The lower side (left side) has strong slew toward colder into the Baseline batch, Then it thins out as stations are lost, but also we pick up some cold outliers. Not sure how to interpret that.
Region 3 – South America
Again remember that a lot of South America is south of the equator and the rest is very near it, so while blue in color, this is the summer graph. There’s an interesting spreading out as thermometer counts rise to about 1950, then both the high end and low end narrow somewhat.
Again the major feature is the loss of 0 C to 10 C range after about 1985.
Again we see the fairly hard cap at about 30 C where water driven effects limit rise. There’s a general thinning after 1980, but it is more diffuse.
Region 2 – Asia
Absolutely straight sides on the right..
Thinning of the low end after the baseline thermometers are lost.
This one is interesting in that both the cold and the warm side show some spreading out but the cold side is getting colder faster! Then again we get a drop out in the -5 C to 5 C range post 1990 thermometer drops.
Region 1 – Africa
Strongly split between north of the Sahara and south of the Sahara and Equator, so we get bifurcated graphs. Has a few crazy outliers, but not much trend in any of it.
Thermometers come and thermometers go, but nothing of trend is visible. Clearly THE biggest thing in the data is Thermometer Change, not Climate Change.
There is an almost imperceptible slew of the right side adding about 1 C to 2 C of warmth (and a lot more thermometers) but at the same time the cold side gets about 10 C colder, then it has a spike down to -10 C to -20 C spots in the Baseline (wonder what station that was…) which then disappears again abut 1990, and in the late 90s we get a general thinning of the data below 20 C to date.
So that is ALL the REAL data, from the unadjusted data set.
I’m not seeing the things you would expect from a general warming trend of the globe across all continents and seasons. It just is NOT showing hotter anywhere, really. At most, there is less extreme cold is some places; but most of it is from the step change loss of thermometers.
Given all the chaos in the data from instrument changes / thermometer loss, there is simply NO WAY to find a 1/2 C “warming signal” in that mess that has any validity at all. You are far far more likely to find artifacts of instrument change.
Here’s a sample of the code I ran. Each varies only on the “region=” number and the heading on the graph. Again, I’ve pretty printed the SQL statement, but it is all on one line in the actual code. Note that I’ve turned plt.xlim(xx,xx) into a comment so each graph just chooses its own range:
# -*- coding: utf-8 -*- import datetime import pandas as pd import numpy as np import matplotlib.pylab as plt import math import MySQLdb plt.title("GHCN v3.3 Ships July") plt.ylabel("Year") plt.xlabel("Temp C") #plt.xlim(0,40) plt.ylim(1850,2020) try: db=MySQLdb.connect("localhost","root","OpenUp!",'temps') cursor=db.cursor() sql="SELECT T.deg_c, T.year FROM invent3 AS I INNER JOIN temps3 as T on I.stationID=T.stationID WHERE T.region=8 AND T.deg_c>-90 AND T.deg_c<50 AND T.month='JULY' ;"' print("stuffed SQL statement") cursor.execute(sql) print("Executed SQL") stn=cursor.fetchall() data = np.array(list(stn)) print("Got data") xs = data.transpose() # or xs = data.T or xs = data[:,0] ys = data.transpose() print("after the transpose") plt.scatter(xs,ys,s=1,color='red',alpha=1) plt.show() plt.title("GHCN v3.3 Ships JAN") plt.ylabel("Year") plt.xlabel("Temp C") # plt.xlim(-40,30) plt.ylim(1850,2020) sql="SELECT T.deg_c, T.year FROM invent3 AS I INNER JOIN temps3 as T on I.stationID=T.stationID WHERE T.region=8 AND T.deg_c>-90 AND T.deg_c<50 AND T.month=' JAN' ;" print("stuffed SQL statement") cursor.execute(sql) print("Executed SQL") stn=cursor.fetchall() data = np.array(list(stn)) print("Got data") xs = data.transpose() # or xs = data.T or xs = data[:,0] ys = data.transpose() print("after the transpose") plt.scatter(xs,ys,s=1,color='blue',alpha=1) plt.show() except: print "This is the exception branch" finally: print "All Done" if db: db.close()