An Intriguing Look At Temperatures In Australia

In this earlier posting:
https://chiefio.wordpress.com/2019/02/07/ghcn-global-thermometers-sine-globe-with-labels/

I made a couple of maps showing the Baseline era stations in blue, and the year ending (2015) stations in red. This lets you see what is being compared to what. Then a question from Larry L. got me to give a quick “thumbnail” of how the global temperature stuff is done.

Basically, last time I looked, GIStemp had gone to 16,000 “grid boxes” into which they stuff a fabricated “temperature”. Since there are only a bit over 2500 current thermometers, and 7000 at the max ever, most of those boxes really have no data. Those that DO have data may have it adjusted based on a variety of things. To fill the boxes the “Reference Station Method” is used to create a temperature guess based on some other thermometer up to 1200 km away.

Well, looking at Australia it had a huge baseline population of thermometers, but very little left now. That got me wondering “What would it look like to just toss all the thermometers on one graph?” Since the stuff is being made up based on all of it, let’s see all of it. At the same time, looking at all the months mushed together would not show trend well. So I decided to pick just one month. In this case, their hot one of January. It is plotted as a scatter plot just after the two globe views of baseline / current.

Year runs up and down. I tried to make it go side to side, but for some unknown reason that throws an exception. Perhaps I’m over running some size limit in the array functions or “whatever”. To some extent I found this more useful anyway as it causes you to think about the data differently.

With that, the graphs:

The Globe – GHCN v3.3 Baseline only Stations vs 2015 “now” Stations

I’ve put both the baseline and the “now” stations on the same map, but those only NOW are red ( 1/2 transparency) while those only in the baseline are blue (full color density). This means a red on top of a blue will give a purple, but a whole lot of reds will tend to hide the blue, and out in the boonies you can see the light red vs the blue as what was dropped vs kept.

GHCN v3.3 NOW in red 50% transparent over Baseline in Blue

Then, since the RED tends to swamp the blue where there’s a lot of thermometers now, I’ve also made one with red first, then blue as 50% transparent over it.

GHCN v3.2 Baseline blue 50% transparent over 2015 “now” in red

Australia

Here is ALL January data for country 501 – Australia. Scatter plot.

Australia all January temperatures

Look at it for a minute. Think about it. See what you think it says. Then read my ideas a bit further down. Next is July, their winter.

GHCN v3.3 Australia July Temps

Think about it a bit too. Remember the “baseline” used by GIStemp is 1950 to 1980 while Hadley use 1960 to 1990 last I looked.

My Ideas?

It sure looks to me like the top end is NOT any hotter, and in fact a bit cooler. Then it looks like a bunch of “medium cool” data sources just go away at that hard cut off about 1990. Now I know they claim to use “anomalies” so that magically removes and all biases ever from anything they do (/sarc;) but really? Any “warming” in the actual body of the data only looks like it shows up as a dropping of thermometers reporting cold values.

There’s still some really cold source (likely the island off to the south), but a bunch of the “middling cold” just goes away.

Overall the dispersion reduced in both directions. There is NO hotter hot happening. It is only the center of mass of the temperatures that has been moved via pruning out a bunch of data sources. Any “warming” isn’t from getting hotter, it is a statistical artifact from the data distribution change.

That’s what it looks like to me.

I’m going to try a couple of other countries and see if any of them look interesting too. As I find interesting bits, I’ll append them to this posting.

I’m not going to post all the code as it is very repetitive. Just one example. Then you change the country code and month to change the graph (and make color and range limits something that works for that data).

# -*- 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 Australia JULY")
plt.ylabel("Year")
plt.xlabel("Temp C")
plt.xlim(-5,30)
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.country=501 AND T.deg_c>-90 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()[0]   # or xs = data.T[0] or  xs = data[:,0]
    ys = data.transpose()[1]
    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()

Update Canada

Yes, an update already. Canada looks to have similar issues, though the data slew more (oddly, it looks to me like it was slewing toward colder then got chopped). I did change the program to plot both graphs at once.

Canada July Temperatures GHCN v3.3

Canada January Temperatures GHCN v3.3

It does look like the same basic issue to me. Lots of “middling cold” stations in the baseline, then dispersion cut down in the present stations. No actually hotter records, just loss of some places providing colder ones. Statistical artifacts, not actually warmer temperatures.

Here’s the code that does the 2fer graphing. It pauses on each graph until you close that graph, so you can study and then save it. In the original the SQL is all one line, I’ve pretty printed it here with linebreaks so it is easier to read.

# -*- 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 Canada July")
plt.ylabel("Year")
plt.xlabel("Temp C")
plt.xlim(0,30)
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.country=403 AND T.deg_c>-90 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()[0]   # or xs = data.T[0] or  xs = data[:,0]
    ys = data.transpose()[1]
    print("after the transpose")

    plt.scatter(xs,ys,s=1,color='red',alpha=1)
    plt.show()

    plt.title("GHCN v3.3 Canada JAN")
    plt.ylabel("Year")
    plt.xlabel("Temp C")
    plt.xlim(-40,10)
    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.country=403 AND T.deg_c>-90 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()[0]   # or xs = data.T[0] or  xs = data[:,0]
    ys = data.transpose()[1]
    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()

I ought to make this take passed parameters for Country Code and such, but I’m not going to do every country… Though I might do a couple more somewhere “interesting”…

Subscribe to feed