GHCN v3.3 vs v4 Asia Anomaly

The GHCN v3.3 vs v4 Series

Well, I got it done. A whole lot of tedious work, but here it is. Yes it is long, but just remember all you have to do is look at the graphs… I have a brief conclusion all the way down at the bottom.

This posting will continue the series on GHCN v3.3 vs. v4 anomalies by presenting the graphs for Asia. Prior postings in this series are here:

A semi-random sample from around the world:

https://chiefio.wordpress.com/2019/04/09/ghcn-v3-3-vs-v4-selected-country-anomaly-differences/

South America – Region 3

https://chiefio.wordpress.com/2019/04/18/ghcn-v3-3-vs-v4-anomaly-south-america/

North America – Region 4

https://chiefio.wordpress.com/2019/04/19/ghcn-v3-3-vs-v4-anomaly-north-america/

Australia & Pacific Islands, region 5 (I included region 7, Antarctica, as well):

https://chiefio.wordpress.com/2019/04/28/ghcn-v3-3-vs-v4-anomaly-australia-pacific-islands/

Asia Climate Zones

Here’s the Koppen Climate graph for the World (from the Wiki) so you have something for comparison.

Köppen-Geiger Climate Map for 1980-2016

Köppen-Geiger Climate Map for 1980-2016

Countries Of Various Regions

This SQL program counts up how many countries are in each continent (“region’) then lists them in decending order:

SELECT COUNT(cname), region 
FROM  country GROUP BY region 
ORDER BY COUNT(cname) DESC;
MariaDB [temps]> source countryperregion.sql
+--------------+--------+
| COUNT(cname) | region |
+--------------+--------+
|           61 | 1      |
|           57 | 6      |
|           37 | 5      |
|           36 | 2      |
|           31 | 4      |
|           16 | 3      |
|            1 | 7      |
|            1 | 9      |
+--------------+--------+
8 rows in set (0.00 sec)

You can see that I’m basically working this list in reverse order, from fewest to most, with the exception that I skipped over Asia (2) to do Australia & Pacific Islands (5) first. There was only one more in A&P than in Asia, and Asia has the complication that in v3.3 both Russia and Kazakhstan are split between Europe and Asia while in v4 they are just treated as whole countries, so I assigned them to Asia. This means the graphs for those two countries will mostly show the effect of instrument change on an anomaly calculation.

Countries Of Asia

So this posting will cover the 36 countries of Asia, with 2 graphs each, so 72 separate graphs (or attempts at graphs – some countries have been missing in one data set or the other).

This little SQL program lists the countries of Asia and sorts them by name:

SELECT cnum, abrev,region, cname 
FROM country 
WHERE region=2 
ORDER BY cname;

So here’s the list of countries. You will notice that “Asia” includes some parts of the Middle East, but not others. Saudi Arabia, but not Israel. Why? Don’t ask why…. Here’s the slice of this report for Europe that includes Israel.

| 622  | IS    | 6      | Israel    

So Israel, Jordan, Lebanon are all “Europe” while Yemen, Saudi Arabia, UAE, Oman are Asia. Go figure…

It is my intent to group these countries so that those nearer each other in climate type are presented near each other on the page, to make comparisons easier.

As Russia and China dominate Asia, and have their own complex climate zones, they will be done first as stand alone graphs. Then the Asian FSU (Former Soviet Union) States near Russia. Then India and those countries clustered around it, covering the Himalayas and nearby, followed by the Far East Nations around China. Then finally those countries of the Near East that border Europe I’ll work the various Asian Islands in as seems appropriate.

MariaDB [temps]> source Asia.sql
+------+-------+--------+-------------------------------+
| cnum | abrev | region | cname                         |
+------+-------+--------+-------------------------------+
| 201  | AF    | 2      | Afghanistan                   |
| 202  | BA    | 2      | Bahrain                       |
| 203  | BG    | 2      | Bangladesh                    |
| 216  | BM    | 2      | Burma                         |
| 204  | CB    | 2      | Cambodia                      |
| 205  | CH    | 2      | China                         |
| 299  | HK    | 2      | Hong Kong                     |
| 207  | IN    | 2      | India                         |
| 208  | IR    | 2      | Iran                          |
| 209  | IZ    | 2      | Iraq                          |
| 210  | JA    | 2      | Japan                         |
| 211  | KZ    | 2      | Kazakhstan                    |
| 206  | KN    | 2      | Korea, North                  |
| 221  | KS    | 2      | Korea, South                  |
| 212  | KU    | 2      | Kuwait                        |
| 213  | KG    | 2      | Kyrgyzstan                    |
| 214  | LA    | 2      | Laos                          |
| 234  | MC    | 2      | Macau S.A.R                   |
| 235  | MV    | 2      | Maldives                      |
| 215  | MG    | 2      | Mongolia                      |
| 217  | NP    | 2      | Nepal                         |
| 218  | MU    | 2      | Oman                          |
| 219  | PK    | 2      | Pakistan                      |
| 220  | QA    | 2      | Qatar                         |
| 222  | RS    | 2      | Russia                        |
| 298  | SH    | 2      | Saint Helena [United Kingdom] |
| 223  | SA    | 2      | Saudi Arabia                  |
| 224  | CE    | 2      | Sri Lanka                     |
| 236  | TW    | 2      | Taiwan                        |
| 227  | TI    | 2      | Tajikistan                    |
| 228  | TH    | 2      | Thailand                      |
| 229  | TX    | 2      | Turkmenistan                  |
| 230  | AE    | 2      | United Arab Emirates          |
| 231  | UZ    | 2      | Uzbekistan                    |
| 232  | VM    | 2      | Vietnam                       |
| 233  | YM    | 2      | Yemen                         |
+------+-------+--------+-------------------------------+
36 rows in set (0.01 sec)

MariaDB [temps]>

Notice that Saint Helena is listed here as in Asia. Since it is in the South Atlantic, that is clearly an error. I’ll be doing a bit of “looking up stream” to see where that error is introduced. A quick look at v3 shows no Saint Helena and for v4 there was a need to make “country code” numbers for those that were missing in v3.3; so most likely that is my error and I’ll need to change the country number from 298 to 1xx (and Africa will go up by one, while Asia goes down by one, on country counts). Given that Saint Helena has no data in v3.3 it will not have any effect on the graphs here to omit it.

So in fact we’re looking at 35 countries and 70 graphs. Though really, since 5 of these countries were in the semi-random world tour set, it’s only 30 new countries and 70 new graphs. Yes, I’m trying to convince myself it’s not that much work ;-)

A Bit Of Tech Talk on Process

As an aside on process, I have a “model program” for each of the two graphs. I copy those for each country, then edit the country abbreviation and name to be each of those on the list for that continent. Doing that all by hand is a bit tedious, so some of it is automated. Here’s the script that makes a copy of the model for each country:

for N in AF BA BG BM CB CH HK IN IR IZ JA KZ KN KS KU KG LA MC MV MG NP MU PK QA RS SA CE TW TI TH TX AE UZ VM YM
do
	echo doing:
	echo cp cnACdelta.py cn${N}delta.py 
	cp cnACdelta.py cn${N}delta.py 
	echo cp a3v4deltaAC.py a3v4delta${N}.py 
	cp a3v4deltaAC.py a3v4delta${N}.py 
done

It is nmed “dup” and there’s one of them in each directory where I’m processing a continent. So I just copy in the model and the old version of “dup”, then change the list of country abbreviation to be those for that continent. Finally, running ./dup makes all those program copies.

chiefio@odroidxu4:/SG/xfs/chiefio/Py3/Aasia$ ./dup
doing:
cp cnACdelta.py cnAFdelta.py
cp a3v4deltaAC.py a3v4deltaAF.py
doing:
cp cnACdelta.py cnBAdelta.py
cp a3v4deltaAC.py a3v4deltaBA.py
doing:
cp cnACdelta.py cnBGdelta.py
cp a3v4deltaAC.py a3v4deltaBG.py
doing:
cp cnACdelta.py cnBMdelta.py
cp a3v4deltaAC.py a3v4deltaBM.py
[...]

It might not seem like much, but when you are talking something like 240 countries, and 2 graphs each, and something like 25 characters of destination program names per country (not to mention typing 480 model program names correctly), well, that’s about 6000 characters of destination names alone. Something will end up a typo… So one little script and it’s easy to validate each 2 char abbreviation against the list of that continent.

Similarly, typing 70 names to edit the program is tedious, so I have a similarly small script to toss me into the editor for each:

chiefio@odroidxu4:/SG/xfs/chiefio/Py3/Aasia$  cat edem 
for N in AF BA BG BM CB CH HK IN IR IZ JA KZ KN KS KU KG LA MC MV MG NP MU PK QA RS SA CE TW TI TH TX AE UZ VM YM
do
	vi a3v4delta${N}.py 
	vi cn${N}delta.py 
done

So that lets me just concentrate on matching the two letter Abbreviation to the name of the one I’m editing, and making sure I’ve typed the right country name for that Abbreviation (from the panel I have open with that report of “abrev” and “cname” in Mariadb). MUCH less typing and much less chance for errors or typos to creep in.

Here’s a listing of those programs in that directory, just waiting for me to slog through the editing of them to their new country name and abbreviation in the selection and reporting portions:

chiefio@odroidxu4:/SG/xfs/chiefio/Py3/Aasia$ ls
a3v4deltaAC.py	a3v4deltaKN.py	a3v4deltaTI.py	cnHKdelta.py  cnNPdelta.py
a3v4deltaAE.py	a3v4deltaKS.py	a3v4deltaTW.py	cnINdelta.py  cnPKdelta.py
a3v4deltaAF.py	a3v4deltaKU.py	a3v4deltaTX.py	cnIRdelta.py  cnQAdelta.py
a3v4deltaBA.py	a3v4deltaKZ.py	a3v4deltaUZ.py	cnIZdelta.py  cnRSdelta.py
a3v4deltaBG.py	a3v4deltaLA.py	a3v4deltaVM.py	cnJAdelta.py  cnSAdelta.py
a3v4deltaBM.py	a3v4deltaMC.py	a3v4deltaYM.py	cnKGdelta.py  cnTHdelta.py
a3v4deltaCB.py	a3v4deltaMG.py	cnACdelta.py	cnKNdelta.py  cnTIdelta.py
a3v4deltaCE.py	a3v4deltaMU.py	cnAEdelta.py	cnKSdelta.py  cnTWdelta.py
a3v4deltaCH.py	a3v4deltaMV.py	cnAFdelta.py	cnKUdelta.py  cnTXdelta.py
a3v4deltaHK.py	a3v4deltaNP.py	cnBAdelta.py	cnKZdelta.py  cnUZdelta.py
a3v4deltaIN.py	a3v4deltaPK.py	cnBGdelta.py	cnLAdelta.py  cnVMdelta.py
a3v4deltaIR.py	a3v4deltaQA.py	cnBMdelta.py	cnMCdelta.py  cnYMdelta.py
a3v4deltaIZ.py	a3v4deltaRS.py	cnCBdelta.py	cnMGdelta.py  dup
a3v4deltaJA.py	a3v4deltaSA.py	cnCEdelta.py	cnMUdelta.py  edem
a3v4deltaKG.py	a3v4deltaTH.py	cnCHdelta.py	cnMVdelta.py

I could probably do this with some kind of passed parameters instead, but this lets me re-run any one report later for QA and / or updated data reasons. Also, some of the reports / graphs need different ranges on the MAX / MIN and some just make the one graph for v4 when v3 data are nonexistent. As a TB of disk is about $25 these days, it isn’t like the disk space matters.

The a3v4—- programs are producing the Anomaly version 3.3 and version 4 graphs. The cnXXdelta.py programs are producing the ComparisoN or Delta / Difference of the anomalies (Difference) graphs. Examples of that code are in prior postings in this series.

So it is just edit that code for each of those files to have the right “XX Countryname”, and run each one from inside “idle3”, save the graphs, load them here, look at them, and ponder what they have to say. “Only a few days work” ;-)

Graphs of v3.3 vs v4

Do bear in mind that in theory these data sets are the same historic thermometers around the world. It isn’t really possible to go back to 1890 on some Pacific islands or to China and install a new thermometer. That means that there is little theoretical reason for “cooling the past”. Furthermore, I use a very pure and strict form of anomaly processing. I do NOT use a baseline. (Any references to “baseline” are to the periods used by GIStemp or Hadley). I only ever compare a given location to itself, and only inside the same month. So, for example, I would add up all the July temperatures in Central Park NYC and divide by the number of instances. That’s the average temperature for that location in July. Then the anomaly is calculated by subracting that from the reported temperature in each of those records. THE only reason for an anomaly value for that record is that it is different from the average temperature recorded in that location, in that month. This avoids the issue of the “baseline period” being set in a known cold period and it avoids changes in the average instrument inventory from changing the anomaly.

Then those anomaly values by instrument-month are averaged together for a given country. We are constantly told by “Climate Scientists” that using Anomalies makes it irrelevant what instruments are in the inventory and when. Well, OK then, that, by their argument, means that ALL the difference in the anomalies in these two data sets are down to their processes. Or, IMHO more likley, it just illustrates that using anomaly processing does not remove instrument selection bias.

The “Anomaly” graph plots the two sets of average anomalies for each country. At least in theory they ought to be identical. Yet they are not. So just as clearly that demonstrates that something other than CO2 and something other than history is busy at work changing the data and the results.

The “Difference” graph for each country just subtracts one set of average anomaly data from the other, to make it easy to just see the changes that took place between the two versions. Just how much “diddle” is in the data that makes it through anomaly processing.

RS Russia

GHCN v3.3 vs v4 RS Russia Difference

GHCN v3.3 vs v4 RS Russia Difference

So we have 1 1/2 C range of change due to processes or instrument selection, with some bits of the past cooled by 1 C.

Looking at the actual anomalies below, the present hot years are only about 1/2 C hotter than the hot years fo the 1800s, but the cold years are essentially gone. Though most peculiarly, only with sudden onset about 1995. That is NOT the gradual accumulation of radiative heating from CO2. As CO2 has a reduction of sensitivity with rising concentration, most of any “warming” effect ought to have been before 2000, not after it.

GHCN v3.3 vs v4 RS Russia Anomaly

GHCN v3.3 vs v4 RS Russia Anomaly

UPDATE:

I’ve worked out a method of combining the 2 “countries” of European Russia and Asian Russia used in GHCN v3.3 and prior for a “apple to apples” comparison with the v4 data (where there is only one ‘RS’ Russia). Those two graphs will follow this comment. One observation on this is simply that even using “anomalies”, instrument change MATTERS. “Climate Scientists” often claim that since they use anomalies, the exact instruments in use does not matter. Compare these graphs to the ones above. They are different. All processed with anomalies, and the Asian thermometers are the same data, so the difference is all down to “instrument selection”.

Russia With Europe & Asia v3.3 Combined

GHCN v3.3 vs v4 Russia Anomaly Difference Europe & Asia Combined

GHCN v3.3 vs v4 Russia Anomaly Difference Europe & Asia Combined

GHCN v3.3 vs v4 Russia Anomalies Europe & Asia Combined

GHCN v3.3 vs v4 Russia Anomalies Europe & Asia Combined

One obvious difference is just that the oldest data is all from Europe, so there are now v3.3 dots in the early years of the Anomaly graph. Another is that the “difference” graph is less scattered. That makes sense as there is now less difference in total instrument data. It does hint that a dispersed or wide ranging Difference Graph can be an indicator of a lot of instrument change; implying you are measuring instrument change more than climate changes.

CH China

GHCN v3.3 vs v4 China Difference

GHCN v3.3 vs v4 China Difference

A little dip into the start of the baseline window, and recent larger changes, but over all China is fairly flat. The general cooling of the past is a bother. Just wrong, even if only a 1/10 C or two.

The actual anomalies of the “warm now” are not signficantly different from the warm 1940s or that spot about 1860. We do, again, see a loss of cold excursions and a step function in about 2000. I wonder if that’s about when they installed the new electronic thermometers, or just about when they were promised lots and lots of money from the “Paris Accords”?

GHCN v3.3 vs v4 China Anomaly

GHCN v3.3 vs v4 China Anomaly

MG Mongolia

GHCN v3.3 vs v4 MG Mongolia Difference

GHCN v3.3 vs v4 MG Mongolia Difference

Wow! Just wow. A full 1 C of “cooling of the historical record”, then about 1/4 C cooling in the middle, and a bit of warming at the end. So while the news from Mongolia has been about historic losses of cattle due to the very severe cold, the GHCN is claiming it is the hottest ever. I don’t believe it.

Note that the Anomaly graph has an expanded range. The central continent areas are prone to wider temperature ranges. So according to these anomalies, it was about 5 C colder than now up to the 1950s… yet we’re having the nomadic herdsdmen threatened by cold now. Reality and the data diverge.

Then there’s just the question of causality. CO2 is supposed to account for about 1/2 C to at most 1 C of warming. Here we’ve got 5 C. So what caused the other 4 C of warming? Might it have caused the whole 5 C?

Now imagine a virtical line about 1990 (or put a ruler to your monitor). Before that point, no warming. After that point, a sudden Jump Up in temperatures and big loss of cold years. How does CO2 do nothing over most of the accumulation period then have onset of effect at a point in time? Hardware / instrument changes can do that. Building new airports around thermometers can do that. Processing changes can do that. CO2 not so much. We will see a lot of that in the graphs below.

GHCN v3.3 vs v4 MG Mongolia Anomaly

GHCN v3.3 vs v4 MG Mongolia Anomaly

KZ Kazakhstan

GHCN v3.3 vs v4 KZ Kazakhstan Difference

GHCN v3.3 vs v4 KZ Kazakhstan Difference

Interesting that it is mostly cooling the whole record with more cooling recently.

Looking at the actual anomalies, again with an expanded range, there’s 4 C to 5 C of “warming”, but mostly as a loss of cold excursions after the year 2000. Once again I must ask: Why did CO2 wait until the year 2000 to warm up Kazakhstan? Then how does it cause so much warming in excess of theory?

GHCN v3.3 vs v4 KZ Kazakhstan Anomaly

GHCN v3.3 vs v4 KZ Kazakhstan Anomaly

UZ Uzbekistan

GHCN v3.3 vs v4 UZ Uzbekistan Difference

GHCN v3.3 vs v4 UZ Uzbekistan Difference

This is an interesting one. It looks like someone was embarrassed and tried to “fix up” the data to not look so wrong. Raise the baseline years by almost 1/2 C and cut the present by a full 1 C? Since we can get 1.5 C of variation just from data variation (remember, this is supposedly the same “good data” from the same places & times…) doesn’t that mean the error bars on the data are bigger than the 1/2 C of “CO2 warming” folks are trying to find in it?

Looking at the actual anomalies we see a familiar pattern. Warm years now about like the late 1800s. Loss of cold years / anomalies, but only after about 1990 / 2000. So here’s your choices: EITHER CO2 is entirely beneficial as all it is doing is removing the most horribly cold years when things die, or the data are crap and diddled after about 1990.

GHCN v3.3 vs v4 UZ Uzbekistan Anomaly

GHCN v3.3 vs v4 UZ Uzbekistan Anomaly

TX Turkmenistan

GHCN v3.3 vs v4 TX Turkmenistan Difference

GHCN v3.3 vs v4 TX Turkmenistan Difference

Not much in the Difference graph. Mostly just a general slight reduction of history. There are a couple of years that get changed by -1.5 to -1.25 C but I suppose they had a good reason. Though if those years were that far off, what confidence can we have in the other years?

The Anomaly graph is a familiar pattern. Nothing at all happening until about the year 2000, then a sudden onset loss of log going years and a “flip” up at the tail. And all that during the time the satellites were measuring a “pause” in the warming…. Wonder which one is lying… /sarc;

GHCN v3.3 vs v4 TX Turkmenistan Anomaly

GHCN v3.3 vs v4 TX Turkmenistan Anomaly

AF Afghanistan

GHCN v3.3 vs v4 AF Afghanistan Difference

GHCN v3.3 vs v4 AF Afghanistan Difference

Under almost constant warfare, I doubt they have any good data. Then the changes are 2.5C in range.

The actual anomalies show a hot year in the ’50s about the same as now. A big drop out during one of their wars. A very short record in two disjoint segments. And what looks more like a step function on the return of measuring after 2000.

Would I trust a tend line fited to that mess to mean anything? Nope. So why average it in with other data and think it will mean anything better?

GHCN v3.3 vs v4 AF Afghanistan Anomaly

GHCN v3.3 vs v4 AF Afghanistan Anomaly

KG Kyrgyzstan

GHCN v3.3 vs v4 KG Kyrgyzstan Difference

GHCN v3.3 vs v4 KG Kyrgyzstan Difference

Changes ranging over almost a 3 C range. This is expected to be precise to 1/10 C? When the same place same dataset changes a few C version to version?

Anomalies show tops more or less the same, with periodic ripple, until about that 1990 to 2000 period, then the familiar sudden loss of cold going excursions and a step up in highs. I really do want to know how CO2 knew to stay inactive until 1990…

GHCN v3.3 vs v4 KG Kyrgyzstan Anomaly

GHCN v3.3 vs v4 KG Kyrgyzstan Anomaly

TI Tajikistan

GHCN v3.3 vs v4 TI Tajikistan Difference

GHCN v3.3 vs v4 TI Tajikistan Difference

Despite changes with a range of 3 C, the actual anomalies don’t have that “Tailored Flip” look after 2000 like the others. There was something of a “flip” but it looks like the changes took out much of it. It has something of the same “shift” after about 1990-2000 but it looks like the red dots have adjusted out the flip. Then just the whole shape of things is unlike all the neighbors.

GHCN v3.3 vs v4 TI Tajikistan Anomaly

GHCN v3.3 vs v4 TI Tajikistan Anomaly

IN India

GHCN v3.3 vs v4 India Difference

GHCN v3.3 vs v4 India Difference

Looks like India isn’t with the “adjusting” program ;-)

There is a bit of jump after 2000.

GHCN v3.3 vs v4 India Anomaly

GHCN v3.3 vs v4 India Anomaly

MV Maldives

The Maldives have no data for GHCN v3.3 (odd, that, given they are the frequent poster child for sea level rise despite no damage from “seal level rise” and building a new tourist airport with lots of big jets… So all we get is this “anomaly” graph. Wonder when the big jet port went in…

MariaDB [temps]> SELECT cnum, abrev, region, cname FROM country WHERE abrev='MV'
    -> ;
+------+-------+--------+----------+
| cnum | abrev | region | cname    |
+------+-------+--------+----------+
| 235  | MV    | 2      | Maldives |
+------+-------+--------+----------+
1 row in set (0.37 sec)

MariaDB [temps]> SELECT COUNT(deg_C) FROM temps3 WHERE country=285;
+--------------+
| COUNT(deg_C) |
+--------------+
|            0 |
+--------------+
1 row in set (34.98 sec)

MariaDB [temps]> 

Looks like after a jump about 1995, it is flat.

GHCN v3.3 vs v4 MV Maldives Anomaly

GHCN v3.3 vs v4 MV Maldives Anomaly

CE Sri Lanka

GHCN v3.3 vs v4 CE Sri Lanka Difference

GHCN v3.3 vs v4 CE Sri Lanka Difference

Not much change version to version. Then sort of dead flat until just about 1985, then the jump up happens.

GHCN v3.3 vs v4 CE Sri Lanka Anomaly

GHCN v3.3 vs v4 CE Sri Lanka Anomaly

PK Pakistan

GHCN v3.3 vs v4 PK Pakistan Difference

GHCN v3.3 vs v4 PK Pakistan Difference

The most recent data changed the most. Why would the most recent be the most wrong? Then we again have the circa 2000 jump.

GHCN v3.3 vs v4 PK Pakistan Anomaly

GHCN v3.3 vs v4 PK Pakistan Anomaly

NP Nepal

GHCN v3.3 vs v4 NP Nepal Difference

GHCN v3.3 vs v4 NP Nepal Difference

Just sort of screwy. 1.5 C of random changes. Then again with the flat data to about 1995 when things jump up. Looks like those two big up changes after 2000 were to erase the two low temperature excursions in the prior record.

GHCN v3.3 vs v4 NP Nepal Anomaly

GHCN v3.3 vs v4 NP Nepal Anomaly

BG Bangladesh

GHCN v3.3 vs v4 BG Bangladesh Difference

GHCN v3.3 vs v4 BG Bangladesh Difference

Some cooling added in the past (smoohting out the 1900 and 1930 warm peaks) then the recent data has a lot of change. Mostly upward.

The anomalies are essentially flat until about 1985, when things “Do The Jump!” (Wonder if there’s any ’50s rock song by that name…. There is the Van Halen song “Jump”…) Mostly via changing the data between v3.3 and v4.

GHCN v3.3 vs v4 BG Bangladesh Anomaly

GHCN v3.3 vs v4 BG Bangladesh Anomaly

BM Burma

GHCN v3.3 vs v4 BM Burma Difference

GHCN v3.3 vs v4 BM Burma Difference

Not much change. Not much going on.

So if the whole world is warming, and some of it by 3C to 4C, from CO2, why is Burma not?

GHCN v3.3 vs v4 BM Burma Anomaly

GHCN v3.3 vs v4 BM Burma Anomaly

TH Thailand

GHCN v3.3 vs v4 TH Tailand Difference

GHCN v3.3 vs v4 TH Tailand Difference

Nice “dip” added in the baseline era data.

Then the Anomaly graph: Then the basically flat black dot non-trend gets turned into a turning upward curve. Largely via no low going excursions and a narrowing of range so only the upside values remain.

GHCN v3.3 vs v4 TH Tailand Anomaly

GHCN v3.3 vs v4 TH Tailand Anomaly

LA Laos

GHCN v3.3 vs v4 LA Laos Difference

GHCN v3.3 vs v4 LA Laos Difference

About 1.5 C of changes bouncing around.

Looking at the anomalies, it looks like the 1940s were wicked cold. Were they really? Not seeing that in nearby places.

Not much trend in the top excursions until that same old 1995 or so point, when we get the loss of low going excursions and a few more upward fliers.

GHCN v3.3 vs v4 LA Laos Anomaly

GHCN v3.3 vs v4 LA Laos Anomaly

CB Cambodia

GHCN v3.3 vs v4 CB Cambodia Difference

GHCN v3.3 vs v4 CB Cambodia Difference

Data in 3 segments. The past segments gets a big cooling while the present gets a bit of lift.

The anomalies show now trend inside the data segements. Only “jumps” between them.

GHCN v3.3 vs v4 CB Cambodia Anomaly

GHCN v3.3 vs v4 CB Cambodia Anomaly

VM Vietnam

GHCN v3.3 vs v4 VM Viet Nam Difference

GHCN v3.3 vs v4 VM Viet Nam Difference

About 1/2 C of cooling the past, 1/2 C of warming the resent data.

Not much trend in the data until the 1990’s, and with the 1900 highs about like then. Then the most recent data jumps up.

GHCN v3.3 vs v4 VM Viet Nam Anomaly

GHCN v3.3 vs v4 VM Viet Nam Anomaly

MC Macau

GHCN v3.3 vs v4 MC Macau Difference

GHCN v3.3 vs v4 MC Macau Difference

This one is pretty dramatic. 1/2 C of cooling of the past, and up to 1 C of warming of more recent data, and you too can turn a flat non-trending black dots into a trending red dots. Oh, and erase those embarrassing low years we had recently.

GHCN v3.3 vs v4 MC Macau Anomaly

GHCN v3.3 vs v4 MC Macau Anomaly

HK Hong Kong

Hong Kong has no data in version 3.3 for GHCN. Now you might wonder how we had North Korean data, but could not managed to get Hong Kong data from when it was a Crown Colony and the British recorded everything. But we have it now, since it’s a Communist China possession. So we only get the one graph of GHCN v4 data anomalies.

MariaDB [temps]> SELECT cnum,abrev, cname FROM country WHERE abrev="HK";
+------+-------+-----------+
| cnum | abrev | cname     |
+------+-------+-----------+
| 299  | HK    | Hong Kong |
+------+-------+-----------+
1 row in set (0.00 sec)

MariaDB [temps]> SELECT COUNT(deg_C) FROM temps3 WHERE country=299;
+--------------+
| COUNT(deg_C) |
+--------------+
|            0 |
+--------------+
1 row in set (34.03 sec)

MariaDB [temps]> 

Tops not going up, but we are losing some of the cold excursions. Yet one more country down, and in the same water, we have Taiwan with a spike up then a cooling trend in that same 1990 to now period.

GHCN v3.3 vs v4 HK Hong Kong Anomaly

GHCN v3.3 vs v4 HK Hong Kong Anomaly

TW Taiwan

GHCN v3.3 vs v4 TW Taiwan Difference

GHCN v3.3 vs v4 TW Taiwan Difference

Interesting that the v3.3 data had an end in the 1960s then one year pops up, no warmer in the ’80s, so it goes silent. Then in the v4 anomaly below, we have a more complete record. What does it show?

About a 1 C rise from about 1920 to the early 1940s (now cooled in the v4 version), then flat until almost the year 2000 but with a general dropping of the highs. I wonder if that is why those years were being left out of v3.3? Then as we see many many times in this dataset, a big jump up in about 2000. Followed by a long drift back down to again just about the norm.

So we have warming in the 1920s to 40s when CO2 can not have been causal, then no warming until 2000 when CO2 was supposed to be causing warming, then a big jump up in 2000 in a very non-CO2 step function, followed by about 15 years of cooling when CO2 is supposed to be making us just burn up.

Frankly, to my eye that looks a whole lot more like cyclical weather patterns and ocean current shifts, layered on top of some UHI as Taiwan industrialized, and then an instrument or processing change in about 2000. It does not look o me like a steady up bias from a well distributed evenly increasing warming blanket of gas.

Then look down at Japan, in the same flow of currents in the same ocean, and it has a different pattern of warming and cooling periods…

GHCN v3.3 vs v4 TW Taiwan Anomaly

GHCN v3.3 vs v4 TW Taiwan Anomaly

JA Japan

GHCN v3.3 vs v4 Japan Difference

GHCN v3.3 vs v4 Japan Difference

Well. Japan can only manage 1/4 C of cooling of the early baseline period. The Japanese must feel guilty about changing historical data…

High excursions about the same until around 1980, but the low excursions moderate. Then we get a jump up into a basically flat block with one hot data point at the end. None of that looks like a gradual increase from CO2 with decreasing effect to incremental additions as they claimed physics would predict.

If I had to pick a “most likely cause”, I would pick industrialization UHI and concrete / tarmac trimming the low end excursion from the areas with thermometers, then an equipment change artifact in the swap to electronic instruments in that 1985-2000 era (varies by country).

GHCN v3.3 vs v4 Japan Anomaly

GHCN v3.3 vs v4 Japan Anomaly

KS South Korea

GHCN v3.3 vs v4 South Korea Difference

GHCN v3.3 vs v4 South Korea Difference

Ah, the Classics. Just a smooth gentle 1/2 C colder past… Korean tech is neat and careful like that.

Prior to the changes, the black dots have a minor warming, likely from UHI and industrialization, up until about 1985. Then again a jump up and once again the tops are staying about flat; so that 2000 hot had to be brought back to trend…

GHCN v3.3 vs v4 South Korea Anomaly

GHCN v3.3 vs v4 South Korea Anomaly

KN North Korea

GHCN v3.3 vs v4 KN North Korea Difference

GHCN v3.3 vs v4 KN North Korea Difference

A whole degree C of changes just from dataset change. Then the anomaly graph shows basically nothing happening until about 1985 with The Jump.

GHCN v3.3 vs v4 KN North Korea Anomaly

GHCN v3.3 vs v4 KN North Korea Anomaly

SA Saudi Arabia

GHCN v3.3 vs v4 Saudi Arabia Difference

GHCN v3.3 vs v4 Saudi Arabia Difference

1.5 C range of variation? I wonder if they play with their thermometers much?…

Then a behaviour that’s becomming painfully familiar. Nothing much at all happening until about 1995 when The Jump happens and we lose low going excursions as the range narrows to nearly constant non-variablity. Wasn’t there a report of Snow in Arabia for the first time in a life time? Ought there not be some cold years in their recent history?

GHCN v3.3 vs v4 Saudi Arabia Anomaly

GHCN v3.3 vs v4 Saudi Arabia Anomaly

IZ Iraq

GHCN v3.3 vs v4 IZ Iraq Difference

GHCN v3.3 vs v4 IZ Iraq Difference

Not many changes, but there is data dropout from war years.

The anomaly graphs show a 3.5 C range of their recent temperatures. So why are they so variable and their neighbor, Saudi Arabia, is unchangable?

Essentially flat through 1980. (Was CO2 on vacation then? Maybe Sadam scared it away… /sarc;) Then a jump up on the new data (and new electronic instruments?)

GHCN v3.3 vs v4 IZ Iraq Anomaly

GHCN v3.3 vs v4 IZ Iraq Anomaly

IR Iran

GHCN v3.3 vs v4 IR Iran Difference

GHCN v3.3 vs v4 IR Iran Difference

Iran gets up to 2.5 C of range in the changes.

Then that painfully familiar behaviour of “nothing changes” until a sudden jump up. In this case in 2000. Wonder if they were having trouble buying new Electronic Thermometers until then?…

GHCN v3.3 vs v4 IR Iran Anomaly

GHCN v3.3 vs v4 IR Iran Anomaly

KU Kuwait

GHCN v3.3 vs v4 kU Kuwait Difference

GHCN v3.3 vs v4 kU Kuwait Difference

Again we have 2.5 C range of “change from data choices”. So how do we know that the 1/2 C of “Global Warming” is from such data choices?

Then the anomaly has a familiar pattern. Flat until a step function “Jump Up” in 2000. Yet Oman jumps up in about 1980.

GHCN v3.3 vs v4 kU Kuwait Anomaly

GHCN v3.3 vs v4 kU Kuwait Anomaly

MU Oman

GHCN v3.3 vs v4 MU Oman Difference

GHCN v3.3 vs v4 MU Oman Difference

About a 1 C range of change from data choices. Then theirs a jump up about 1980, and then a smaller one about 2000. Oddly, the “fix” of the red dots remove that jump in 1980 and make it more trend… Finally, with snow in the desert in recent years, we have recent data as the “Hottest Evah”. Really?

GHCN v3.3 vs v4 MU Oman Anomaly

GHCN v3.3 vs v4 MU Oman Anomaly

BA Bahrain

GHCN v3.3 vs v4 BA Bahrain Difference

GHCN v3.3 vs v4 BA Bahrain Difference

Then there’s the old small cool of the past every so often, kick up the present a bit.

The result over who knows how many years of it? A 3C to 4C hotter “now” than in the past. Yet looking at nearby Oman, it has half the warming over the period of overlap (about 1 C from 1960 instead of 2 C).

GHCN v3.3 vs v4 BA Bahrain Anomaly

GHCN v3.3 vs v4 BA Bahrain Anomaly

QA Qatar

GHCN v3.3 vs v4 QA Qatar Difference

GHCN v3.3 vs v4 QA Qatar Difference

Another one with 1.5 C of data selection change.

Then the anomaly graph is again basically flat to 2000 when about a 1 C Jump happens and we’re Suddenly Warm. I’d like to know how CO2 accounts for that. I’d also like to know what instrument changes might have happened then. Anyone get into Air Travel in a big way then? Qatar Air promoting tourism?

GHCN v3.3 vs v4 QA Qatar Anomaly

GHCN v3.3 vs v4 QA Qatar Anomaly

AE United Arab Emirates

GHCN v3.3 vs v4 AE United Arab Emirates Difference

GHCN v3.3 vs v4 AE United Arab Emirates Difference

A full 1 C of recent cooling of the data. Perhaps becuase a 2 C Jump Up in 2000 was standing out in the data? Need to blend it in more?

GHCN v3.3 vs v4 AE United Arab Emirates Anomaly

GHCN v3.3 vs v4 AE United Arab Emirates Anomaly

YM Yemen

GHCN v3.3 vs v4 YM Yemen Difference

GHCN v3.3 vs v4 YM Yemen Difference

Then we have poor Yemen. Not enough money for new thermometers?

They get their “Jump” in about 1920. Then it’s just dead flat since then. But their data end in about 1990. But I’m sure that they will get another nice big jump after their war ends and new thermometers are installed…

GHCN v3.3 vs v4 YM Yemen Anomaly

GHCN v3.3 vs v4 YM Yemen Anomaly

In Conclusion

Looking at those graphs, I just can’t see any way to justify the claim that it is caused by CO2. CO2 is a slow gradual effect over decades (per the claims) and with stong DECAY of effect with concentration. It doesn’t “do nothing” until 2000 then give a 1 C to 2 C step change. It doesn’t “trim off the cold years” and only narrow the range of variability while not raising the warmer years in one country, while doing something quite different in the country right next door.

To me, this looks a whole lot more like some minor UHI effects up until a Step Change happens. That the change is all about the same decade or two, but does vary by country, makes me think of instrument changes to electronic thermometers, and the growth of airports and jet traffic with thermometers at the airports.

In short, the land temperature data are crap and reflect poor location of instruments and change of instruments more than anything else.

Then, just the way that the anomalies can change by a few Degrees C with shifts of whatever processing is different between v3.3 and v4 OR a change of what instruments are in the record: That means that claiming 1/2 C of “Global Warming”, without addressing that several C of instrument change artifacts, is just daft. There is simply no way to claim you know anything from these data, even with anomaly processing, with that level of change from version to version of what is supposed to be the SAME temperature at the SAME place and mostly with the SAME historical instruments and their data.

Then, that these data using anomaly processing can vary so much; that puts the lie to the claim that you can ignore instrument change by using anomalies. You just can’t. Change the instruments, change the results, it’s that simple.

Since Africa and Europe have a lot more graphs (about double North America), and this was more than enough workload, I think I’ll divide them up into 2 postings each. Otherwise it is likely to just hit the “burnout” point.

Subscribe to feed

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, Global Warming General, NCDC - GHCN Issues and tagged , , , , . Bookmark the permalink.

12 Responses to GHCN v3.3 vs v4 Asia Anomaly

  1. Serioso says:

    An interesting question [for me] is whether [nighttime] lows are more affected than [daytime] highs, per the theory [or observations?] of Freeman Dyson. You have the data: Is this a question of interest to you?

  2. E.M.Smith says:

    I have the GHCN, and some USHCN monthly average data, some of the MAX and MIN data though I’ve not unpacked it, and maybe some “few year old” daily data from the site scrape I did way back when – but that is TB of stuff and on a large archive disk on my file server (that is powered off 99% of the time as that is the safest data archive ;-)

    So I might have the needed day vs night data, depending on what’s in those data sets.

    But I’ve not made any database tables nor loading programs for the daily data. There’s a lot of it… nor have I done anything with the monthly MIN / MAX data (as the one time I tried about a decade ago it seemed very ‘thin’, as in a lot missing. I’d likely need to first validate that I got a full copy before trying again, and learn what is different in how it is packaged.)

    So while I’d be interested in day vs night, I’m pretty sure you are talking weeks of work to get to where I could do something useful with it.

    There’s also the issue that using MIN vs MAX isn’t the same as DAY vs NIGHT. To really get good day vs night, you would need to use the hourly data from those sites that collect it. That I know I do not have, and I’m not sure how to get it.

    I DID sit up ll night a couple of very cold winter nights a few years back collecting data on rate of cooling and such, but it didn’t go anywhere. (Just told me what I already knew – it gets colder until the sun rises…but in a non-linear way – that you can get from the nice graphs at Wunderground without sitting in the cold all night… Though I did find that non-airport thermometers cooled faster and more than those at the local airport ;-)

    So yeah, I have the interest. Just not a lot more time to start another major project… (Maybe when Europe and Africa are done… but I presently was planning to add V1 and V2 to this set and then make some spot checks of all 4 versions for comparisons. Leveraging this set for “interesting places to look”)

    One other problem is the processing done on the data. I did a posting many years back about the “QA” process. It compares a station with nearby ones and tosses out the data if it is “too different”, then estimates a replacement via the Reference Station Method. One glaring problem was that they used the same acceptance band for both high and low excursions; yet we KNOW that low going excursions can go further and faster than high going ones. So by definition using the same bandpass on both sidebands will toss more low going valid data than high going. How much of a problem I can’t say. But that complicates the question of impact of any CO2 effect on lows vs highs.

    THE big problem is just getting really unchanged unhomogenized unfutzed with raw data. The entire system from start to end is set up to change the data. Even these “unadjusted” data come with a disclaimer that the upstream national providers might well have adjusted it. “Adjustment” in their parlance is ONLY a limited set of their processes that actually change the data. The other changes are not “adjustments”….

    One Positive Note:

    We know for a fact that UHI, concrete, pavement, etc. raise the lows A LOT more than they change the highs. This is known from the gathered data and from direct experiments. It is also known by every motorcycle rider who has been out on a cold country road and warmed up considerably on entry to towns ;-) I’ve experienced it hundreds of times.

    One particularly memorable evening I was riding past flood irrigated peach orchards. They were quite cold and damp. Then I’d go past a pasture or ploughed dirt. Warmer. Then another orchard, cold. Then I entered town and it was about 10 F warmer and nice!

    So one clear point is that “land use” matters a very great deal to night temperatures. Many degrees worth.

    Yet riding by those same places on a hot “100 F+” day and it was far far more uniformly hot.

    My GUESS is that convection in the heat keeps things more uniform, while the night time has a race condition between the evaporation, radiation, and thermal mass of a given area; and those with lots of hot thermal mass and no evaporation stay hot, while those with evaporation and low thermal mass and a good radiative surface, get way colder faster. But there is no “downward convection” so those pools of air tend to stay where they are made.

    It was so strong an effect that after the 2nd or 3rd time I experienced it, I started taking extra warm gear for night rides in the country and planning routes to avoid orchards and spend more time in urban areas or on freeways…

    Admiring the problem:

    Now, how to disambiguate those effects from the hypothetical of a CO2 effect, that’s an interesting question for which I don’t have much answer. It would need data from a pristine rural area without land use changes nor irrigation; and those are non-existent. Not a lot of people put thermometers out in the desert away from all urban locations, airports, and farming; and especially not 100 years ago. Even in Death Valley they changed the visitor center a lot (growth, more paving) and then moved the instrument to a location with more solar heating / wind blocking.

    It would take a site survey of global thermometer history to find a site without any of the mentioned changes. “Data, data, I need more data!” comes to mind.

    So an awfully long winded way of saying “I’m interested but I suspect it can’t be answered with the data available.”

  3. Larry Ledwick says:

    I can also see those land use temperature effects on my car thermometer as I drive home at night. River bottoms and the near by valley can be as much as 5 deg F cooler than the higher road nearby with the same semi-rural land use.

    Get heavy fog down wind of a reservoir as I come out of the Boulder Creek drainage.
    Light industrial area near my work place is warmer than the nearby farm land by a couple degrees F.

    There is also a warm band that parallels major roadways (4 lane 55-65 mph traffic) due to waste heat from the cars. That warm band extends about 1/4 mile or so on each side of the roadway / highway.

    My understanding is that cooling continues until about 1 hour after sunrise, because net energy is still negative due to cold sky and very low angle of the sun on the ground. Standing upright you have almost normal incidence to the early sun, but the ground has almost no direct heating due to low angle and long shadows. Of course you also have cloud effects regarding heating as they can act as both shades and reflectors.

    Highest current output on solar panels occurs in partly cloudy clear sky conditions when the large clouds are close to but not quite covering the sun. In that case the indirect illumination from the very bright cloud can substantially increase current output on the solar panel as illumination is higher than totally clear sky direct sun.

  4. Bill in Oz says:

    Em I have reposted your conclusion of Jo Nova blog site here in Oz : http://joannenova.com.au/2019/05/what-will-the-carbon-tax-cost-bill-shorten-says-only-a-lying-charlatan-would-ask/#comment-2136595

    looking through the charts I was wondering about Japan, the Koreas & Taiwan from 1900 through to 1945. They are all in the same region of Asia and pretty close to each other. So similar weather processes would have effected them in similar ways. And. at a technical level, the same Japanese meteorological service was doing the weathet bureau readings in that period. And using the same rules and protocols to do so. So there should be a degree of ‘uniformity’ in the charts for these now 4 states – Japan, Taiwan North Korea & South Korea.

    But there isn’t.

    Which supports the the theme that the temperature data has been screwed around.

  5. Bill in Oz says:

    A similar thought occurs to me about the countries that made up the USSR from ~ 1920- 1990. Russia, Khazakstan, Turkmenistan, Kirghizstan etc.. The former Imperial Russian weather service was operating across all these areas of the USSR…And the USSR weather bureau developed out of this..

    Sooo. differences on the charts showing up after 1990 are probably as much to do with political changes having an impact on how the data was collected and handled.

  6. Pingback: W.O.O.D. – 11 May 2019 | Musings from the Chiefio

  7. rms says:

    Superb. Well done. I knew you’d get there. I’m very impressed with the end result, and the methodology used. The methodology is important IMHO as it’s world-class, proper, and using contemporary tools an technology vs. what I see so-called “climate scientists” doing.

  8. Pingback: Wow. Look at This | Musings on Interesting Things

  9. Ian W says:

    My concern with the ‘warmer lows’ especially as you state in UHI and land use effects is the claim to have a ‘daily average temperature’. When all that has been done is take the mathematical mean of max temperature and min temperature. The warmer lows will then lead to a rise in this mathematical mean, which is called a rise in average temperature. That rise is then added to the maximum temperatures projected forward, and thus we get to the we are all going to boil panic. When in reality there is just a few hundredths of a degree less cooling.

  10. Pingback: GHCN v3.3 vs v4 Anomaly Graphs – Europe | Musings from the Chiefio

  11. Pingback: GHCN v3.3 vs v4 Anomaly Graphs – Europe – Climate Collections

  12. Pingback: GHCN v3.3 vs v4 – Top Level Entry Point | Musings from the Chiefio

Comments are closed.