GHCN v3.3 vs. v4 Anomaly Graphs – Africa

The Climates of Africa

Africa presents a couple of problems for climate trend discovery. For one, it has this giant desert in the middle right next to one of the great wet tropical rain forests of the world. Then it has a Mediterranean Coast that is in the Northern Hemisphere along with a temperate end in the Southern Hemisphere. Seasons can be inverted, or non-existent.

Here’s the Koopen Climate map:

Koppen Climate Map of Africa

Koppen Climate Map of Africa

Looking at that you can see that Morocco and part of Algeria are comparable Mediterranean coastal, but the rest of Algeria is a hot Sahara climate. What really needs to be done is to identify thermometers in the same climate zones and only compare them. But as this is just a start, we start with comparing whole countries.

I’ve done a general grouping of countries into bands that more or less follow the map of climate zones. I was not rigorous about it. So some countries might be more properly compared to a different set of nearby countries. Let the map be your guide for your own comparisons.

In general, I start with the Sahara / bit of Mediterranean band, then scan down through the Tropical and into the Subtropical and Temperate. One “confounder” is the southern desert in places like South West Africa (now Namibia) and Botswana, who are more like each other than anything around them; and with seasons 180 degrees out of phase with the northern desert.

So to some extent my groupings were just to make the process more orderly rather than strict climate matches. But it ought to put comparable places next to similar neighbors most of the time.

I’ve made a quick first comment on each of these countries. Of neccesity, given the number, these are at best a cursory look and some sniditude sprinkled in. This group desperately needs some “Crowd Sourced” scrutiny of the graphs. I’ve flagged a couple that are particularly dodgy, and noted The Usual “drop the baseline 1/2 C raise the present about 1/2 C” and the frequent “The Jump” about 1990-2000 (that likely correlats with MMTS rollouts, IMHO… but needs a good “Dig Here!” for each country.

One other theme is the frequent 1C to 2.5 C range of “change to history”. IF our v3.3 data were really that crappy in 2015, what evidence it is any less crappy now? How do you find 1/2 C of “Global Warming” from CO2 inside 2 C of “random error” and maybe another 1 C of “thermometer changed moved near buildings for the wire”?

It just looks to me like the data are crap and being “massaged” with each release to fit a narrative. That’s my opinion; I hope you will look at the graphs and form your own.

With that, here’s Africa:

The Countries

Here are the countries of Africa per GHCN. You will note many of the abbreviations do not match the names. That is due to the names changing over time. WA West Africa becomes Namibia. UV Upper Volta becomes Burkina Faso. Etc.:

MariaDB [temps]> source bin/Africa.sql
+------+-------+--------+-------------------------------------------------+
| cnum | abrev | region | cname                                           |
+------+-------+--------+-------------------------------------------------+
| 101  | AG    | 1      | Algeria                                         |
| 102  | AO    | 1      | Angola                                          |
| 103  | BN    | 1      | Benin                                           |
| 104  | BC    | 1      | Botswana                                        |
| 161  | IO    | 1      | British Indian Ocean Territory [United Kingdom] |
| 105  | UV    | 1      | Burkina Faso                                    |
| 106  | BY    | 1      | Burundi                                         |
| 107  | CM    | 1      | Cameroon                                        |
| 108  | CV    | 1      | Cape Verde                                      |
| 109  | CT    | 1      | Central African Republic                        |
| 110  | CD    | 1      | Chad                                            |
| 111  | CN    | 1      | Comoros                                         |
| 112  | CF    | 1      | Congo (Brazzaville)                             |
| 154  | CG    | 1      | Congo (Kinshasa)                                |
| 113  | IV    | 1      | Cote D'Ivoire                                   |
| 114  | DJ    | 1      | Dijibouti                                       |
| 115  | EG    | 1      | Egypt                                           |
| 199  | EK    | 1      | Equatorial Guinea                               |
| 116  | ER    | 1      | Eritrea                                         |
| 117  | ET    | 1      | Ethiopia                                        |
| 198  | EU    | 1      | Europa Island [France]                          |
| 143  | FS    | 1      | French Southern and Antarctic Lands [France]    |
| 118  | GB    | 1      | Gabon                                           |
| 150  | GA    | 1      | Gambia, The                                     |
| 119  | GH    | 1      | Ghana                                           |
| 120  | GV    | 1      | Guinea                                          |
| 121  | PU    | 1      | Guinea-Bissau                                   |
| 197  | JU    | 1      | Juan De Nova Island [France]                    |
| 122  | KE    | 1      | Kenya                                           |
| 162  | LT    | 1      | Lesotho                                         |
| 123  | LI    | 1      | Liberia                                         |
| 124  | LY    | 1      | Libya                                           |
| 125  | MA    | 1      | Madagascar                                      |
| 126  | MI    | 1      | Malawi                                          |
| 127  | ML    | 1      | Mali                                            |
| 128  | MR    | 1      | Mauritania                                      |
| 129  | MP    | 1      | Mauritius                                       |
| 163  | MF    | 1      | Mayotte [France]                                |
| 130  | MO    | 1      | Morocco                                         |
| 131  | MZ    | 1      | Mozambique                                      |
| 132  | WA    | 1      | Namibia                                         |
| 133  | NG    | 1      | Niger                                           |
| 134  | NI    | 1      | Nigeria                                         |
| 165  | RE    | 1      | Reunion [France]                                |
| 166  | RW    | 1      | Rwanda                                          |
| 196  | SH    | 1      | Saint Helena [United Kingdom]                   |
| 136  | TP    | 1      | Sao Tome and Principe                           |
| 137  | SG    | 1      | Senegal                                         |
| 138  | SE    | 1      | Seychelles                                      |
| 139  | SL    | 1      | Sierra Leone                                    |
| 140  | SO    | 1      | Somalia                                         |
| 141  | SF    | 1      | South Africa                                    |
| 148  | SU    | 1      | Sudan                                           |
| 167  | WZ    | 1      | Swaziland                                       |
| 149  | TZ    | 1      | Tanzania                                        |
| 151  | TO    | 1      | Togo                                            |
| 168  | TE    | 1      | Tromelin Island [France]                        |
| 152  | TS    | 1      | Tunisia                                         |
| 153  | UG    | 1      | Uganda                                          |
| 169  | WI    | 1      | Western Sahara                                  |
| 155  | ZA    | 1      | Zambia                                          |
| 156  | ZI    | 1      | Zimbabwe                                        |
+------+-------+--------+-------------------------------------------------+
62 rows in set (0.25 sec)

Sixty Two is a LOT of countries and that’s 122 graphs. That’s one long posting. Due to that, I’m going to divide Africa into three parts. North Africa near the Mediterranean including the Sahara, as many of those nations extend into the desert; then Equatorial that tends to be Tropical Rain Forrest and will include any islands of the Northern Hemisphere; and finally Southern that tends to a bit more temperate and will include the islands of the Southern Hemisphere Africa.

This may not end up an equal number of nations in each posting, but it will tend to group together those nations with similar environments. I’ve started near the Atlantic coast of the north, taking the coastal route to Eritrea, then we return across the Sahara and proceed along the tropical central coast, cross again the continent to the Indian Ocean side, returning in a weaving motion across the Southern Horn of Africa. Then we do through all those islands in the South and out into the Indian Ocean, but also including those in the Atlantic.

Northern:         Equatorial:              Southern:

MR Mauritania     CV Cape Verde            AO Angola
WI Western Sahara SG Senegal               WA Namibia
MO Morocco        GA Gambia, The           BC Botswana    
AG Algeria        PU Guinea-Bissau         ZA Zambia
TS Tunisia        GV Guinea                ZI Zimbabwe    
LY Libya          SL Sierra Leone          MI Malawi
EG Egypt          LI Liberia               MZ Mozambique
ER Eritrea        IV Cote D'Ivoire         WZ Swaziland
DJ Dijibouti      GH Ghana                 SF South Africa
ET Ethiopia       TO Togo                  LT Lesotho  
SO Somalia        BN Benin   
SU Sudan          NI Nigeria               Islands:   
CD Chad           CM Cameroon              MA Madagascar
NG Niger          EK Equatorial Guinea     MF Mayotte [France]
ML Mali           TP Sao Tome and Principe CN Comoros
UV Burkina Faso   GB Gabon                 JU Juan De Nova Island [France]
                  CT Central African Rep.  EU Europa Island [France]   
                  CF Congo (Brazzaville)   TE Tromelin Island [France]
                  CG Congo (Kinshasa)      RE Reunion [France]
                  RW Rwanda                SE Seychelles
                  BY Burundi               MP Mauritius
                  UG Uganda                IO British Indian Ocean 
                  KE Kenya                    Territory   
                  TZ Tanzania              FS French Southern
                  SO Somalia                  and Antarctic Lands [France]
                                           SH Saint Helena [United Kingdom]

Should you need to look up where one of these countries is located, here is the Political Map of Africa (click or open in a new tab to embiggen):

The Graphs

Northern Africa

MR Mauritania

GHCN v3.3 vs v4 MR Mauritania Difference

GHCN v3.3 vs v4 MR Mauritania Difference

A pattern we have seen often, and will see often again. The “Baseline Period” used by NOAA / GISS and Hadley from about 1950 to 1990 gets cooled in the move from v3.3 to version v4 of the data, and the more recent periods get warmed.

Looking at the actual anomaly plotted below, we see the same pattern. A “dip” in the baseline period and a rise after. Oddly, they often ignore time prior to the Baseline, so here we see the temperature anomalies were about as warm as now in the 19-teens to 1930s. Also note that a line, laid across the tops, would show more a compression of the range of the Baseline; while the recent “warming average” is almost entirely a result of the loss of cold going excursions, not higher hot years. Then, the last couple of dots are near or below zero. That says that the latest data are below average and a bit cold.

So “Global Warming” has left the country, at least for now.

GHCN v3.3 vs v4 MR Mauritania Anomaly

GHCN v3.3 vs v4 MR Mauritania Anomaly

But at lest now you can see why 3/4 C of “tilt” was added. Cooling that hot past by 1/4 C and bumping up the recent period by 1/2 C.

WI Western Sahara

Right next door, almost surrounded by Mauritania, we have a nearly flat “trend” with adjustments all over the place. Though maybe I ought not call them adjustments. Thi sis the “unadjusted” data. It’s just had a bit of change of the past, that’s all. Just a bit of diddle…. /sarc;

GHCN v3.3 vs v4 WI Western Sahara Difference

GHCN v3.3 vs v4 WI Western Sahara Difference

Cooling he Basline by about 1/2 C; warming the more recent data by up to 1.5 C then ignoring the present. I think this one needs a bit more of a “Dig Here!”

GHCN v3.3 vs v4 WI Western Sahara Anomaly

GHCN v3.3 vs v4 WI Western Sahara Anomaly

MO Morocco

GHCN v3.3 vs v4 MO Morocco Difference

GHCN v3.3 vs v4 MO Morocco Difference

A nice 1/2 C “dip” added about 1980 in the Baseline. We see that in the anomaly graph too. Then the very recent couple of years have up to a 2 C colder anomaly. We did have snow in those mountains IIRC. So a small pull down of temperatures just prior would help smooth things out. Then fix the trend change by putting that dip in the Basline.

GHCN v3.3 vs v4 MO Morocco Anomaly

GHCN v3.3 vs v4 MO Morocco Anomaly

AG Algeria

GHCN v3.3 vs v4 AG Algeria Difference

GHCN v3.3 vs v4 AG Algeria Difference

I don’t really know what to make of this one, but it is odd. That must have been quite a surprising cold plunge in 1975-76. A 2 C plunge not seen before or since. Wonder if it made the newspapers? /sarc;

Lay a line across this anomaly graph at about the 0.8 C line and we have the “Dip” in the Baseline Period, but the highs are about the same between 1800s, 1950s ish and now. Thogh the most recent data are a -1 dot rather like prior cold drops. It will be interesting to see if that data point is slowly warmed over time… Since all the “warming” after about 2000 is missing cold excursions, not way high hot years, that last cold year will be an embarrassment… It also looks like they are trying to erase that 19-teens cold blip, just warming it some to make a smoother trend. Wouldn’t want it to look like a 60 year cycle with 1975. Then most of the current “warm” would be just the 1940’s + 60 years…

GHCN v3.3 vs v4 AG Algeria Anomaly

GHCN v3.3 vs v4 AG Algeria Anomaly

TS Tunisia 5

GHCN v3.3 vs v4 TS Tunisia Difference

GHCN v3.3 vs v4 TS Tunisia Difference

Ah, another one of these. Crazy all over the place changes of a 2 C range. So does that mean v3.3 has a 2 C random error in it and we can’t trust the official GHCN to +/- 1 C? Doesn’t that cover ALL of “Global Warming” and then some? We do get a nice generalized cooling from about 1940 through the Baseline Period then a move to quasi random changes around zero after that.

Looking at the Anomaly Graph, this is one of those station sets that was basically dead flat until about 1995, then makes a “Step Function” jump by about 1.5 C to 2 C to make the “Duck Tail” at the end. Now I was told that CO2 has been causing a general slow warming since about 1950 and that it has a fall off in effect with concentration; so that means we ought to see MORE effect early one and less recently.

In NO case ought CO2 “warming” hide in the bushes until 1995, then jump out all at once. BOO!

That’s what instrument change or data manipulation does.

I note in passing that the last data point is about -1.2 C and dead center of the historical range, so looks like “Global Warming” can evaporate in one year anyway…

GHCN v3.3 vs v4 TS Tunisia Anomaly

GHCN v3.3 vs v4 TS Tunisia Anomaly

LY Libya

GHCN v3.3 vs v4 LY Libya Difference

GHCN v3.3 vs v4 LY Libya Difference

Very similar observations to the above. We have a gentle cooling of the Baseline Period (with some quasi-random ‘jitter’ but more dots below zero than above.) Then about 1/2 C of “juice” being added after 1990 and one spot even taking a 1.5 C Boost.

Looking at the Anomaly plot, we see a narrowing of range in the Baseline Period, no highs, less extreme cold excursions (likely due to more instruments being averaged I’d guess)then about 1995 to 2000 “The Jump!” by 1 C to 1.5 C into the narrow range Duck Tail shape. Then the most recent years data very cold again. But don’t worry, it can get a 1.5 C boost “later” in the next revision, just like about 2002 did in this one. For now, the average “trend” is preserved and nobody will notice it was cold this past year.

GHCN v3.3 vs v4 LY Libya Anomaly

GHCN v3.3 vs v4 LY Libya Anomaly

EG Egypt

GHCN v3.3 vs v4 EG Egypt Difference

GHCN v3.3 vs v4 EG Egypt Difference

A very odd Difference Graph. Mostly cooling things across the board, but with a spike in a couple of recent years. Still, the past and Baseline Period cooled more than the 1990 to 2000 interval. The odd bit being how much after 2000 was reduced. Looking at the Anomaly Plot, it is almost like they were removing the prior Duck Tail flip up. BUT tossing in two dots “Way crazy high” at almost 2 C. I guess on average that will preserve the trend. Likely good for “homogenizing” into nearby areas too.

Again it all evaporates in the most recent year, and a line at about the 0.8 C level just lays right on top of the regular hot limit. Other than the “adjusted out” v3.3 spots and the “adjusted in” v4 dots in the 20-teens. I also again note thatwith that much “error subject to administrative change” in the data, about 1.5 C in the recent best data, how are we to find 0.5 C of “Global Warming” in an error band of 1.5 C?

GHCN v3.3 vs v4 EG Egypt Anomaly

GHCN v3.3 vs v4 EG Egypt Anomaly

ER Eritrea

GHCN v3.3 vs v4 ER  Eritrea Difference

GHCN v3.3 vs v4 ER Eritrea Difference

Speaking of error bands… 3 C range to the changes in the “unadjusted” data? Really?

Then the old data is in a down trend with relatively wide volatility, and we have a gap, then the newer data is 3 hot by 1 to 2 C and one cold by almost 3 C. Well at least they kept the volatility this time…But 3 high and 1 low does not a trend make. It’s just bouncing between the normal bounds… though that one cold down spike was about 1 C colder than any before. So is “right next door” cooling too? Well, not Egypt per the newly massaged data, but what about Dijibouti?

GHCN v3.3 vs v4 ER  Eritrea Anomaly

GHCN v3.3 vs v4 ER Eritrea Anomaly

DJ Dijibouti

GHCN v3.3 vs 4 DJ Dijibouti Anomaly

GHCN v3.3 vs 4 DJ Dijibouti Anomaly

A closer look at Dijibouti is here:

https://chiefio.wordpress.com/2019/05/18/dijibouti-why-what-changed/

What caught my eye, right off, was that most of the time the data are just lifted about 1/10 C with a stright line preserved. What could justify a perfectly flat adjustment for almost all the data,,. but then hava a big “dip” in the baseline period and a nice “lift” at the more recent end? ALL for “historical” data.

It just looks wrong and suspicious. And it is. (see the link).

Then the bulk of any rising trend in the anomalies is a match to the changes. What? It’s all in the changes of historical data? WT? Then, right next door is Ethiopia (in fact, Dijibouti snuggles it nicely as does Eritrria and Somalia wraps around it. The changes ought to be similar with the Ethiopia being moderted by the wetter climate but the coastal nations moderated by the local seas. Yet they are not the same.

GHCN v3.3 vs 4 DJ Dijibouti Anomaly

GHCN v3.3 vs 4 DJ Dijibouti Anomaly

ET Ethiopia 10

GHCN v3.3 vs v4 ET  Ethiopia Difference

GHCN v3.3 vs v4 ET Ethiopia Difference

What a whack the peak temperatures of the 1950s took! A 1 to 2 C “chop” and those annoying past hot years are a nice rising trend instead. Way to patch up that baseline hot time. Then a lot of other tempertures nudged up a little on each side (with a bit more warming of the present) to balance it out in the average of changes.

On the anomaly plot, with those black 1950s dots moved dow to “on trend”, the only anoying prior heating is in the 19-teens to 20s and nobody cares about that far back.

Yet even wiht that, a flat line at about the -0.8 C point and another at about +0.8 pretty much bound the bulk of the (now) trendless data up until that “Step Function” jump in about 2000. From that point onward, no year is a cold year. Really? Would be interesting to spot check that against local news reports. Even then, it leaves the question of how CO2 did nothing until the year 2000 and then suddenly got busy.

GHCN v3.3 vs v4 ET  Ethiopia Anomaly

GHCN v3.3 vs v4 ET Ethiopia Anomaly

SO Somalia

GHCN v3.3 vs v4 SO Somalia Difference

GHCN v3.3 vs v4 SO Somalia Difference

A 3 C range to the “Fiddle” in the most recent data. So we’e got a declared 3 C error band on known data? If not “error” then why was it changed? We again have the pattern of 1950-90 range the changes are on average negative, while befoe and after things tend to get lifted.

Looking at the anomaly plot, not a lot of trend in it. Even with the fiddle. Then again, being lawless place of conflict not much chance of a new thermometer post 1990 nor much chance of data.

GHCN v3.3 vs v4 SO Somalia Anomaly

GHCN v3.3 vs v4 SO Somalia Anomaly

SU Sudan

Leaving the Coastal band and heading back into the hard desert, we would expect more range to things.

GHCN v3.3 vs v4 SU Sudan Difference

GHCN v3.3 vs v4 SU Sudan Difference

What we get looks like an attempt to erase some embarssing hot years back in the 1920s to ’40s. A good solid 1/2 C to 3/4 C of “cool” added in. But isn’t that MORE than the CO2 “warming” we’re looking to find? If we can tilt the table by that much between two versions OF THE SAME HISTORICAL DATA, doesn’t that make it impossible to say that any 1/2 C “warming” isn’t “fiddle”?

Looking at the anomaly plot, with the v3.3 kept in, it’s a dead flat line across the tops at about 0.6 C right up to 2000, then a sudden jump. Across the bottoms is a bit more volatile (to be expected) with a line about -1 C catching most of the cold excursions with only the tip of the spikes going to -1.5 C right up until about 1995, then the cold excursions suddenly end. The whole block of data post 2000 looks like it is just shifted up 1 C and volatility reduced. A well distributed gas causing long slow heating over 60 years would not be a “step function” in about 2 to 5 years.

GHCN v3.3 vs v4 SU Sudan Anomaly

GHCN v3.3 vs v4 SU Sudan Anomaly

CD Chad

GHCN v3.3 vs v4 CD Chad Difference

GHCN v3.3 vs v4 CD Chad Difference

Ah the familar pattern. Cool the basline period by 1/4 C, pump up the recent data by 1/2 to 1 C, preto chango “instant Global Warming!”

In anomalies we again have flat to falling until we reach 2000, then “The Jump!” ™ but 1 C to 2 C after that.

GHCN v3.3 vs v4 CD Chad Anomaly

GHCN v3.3 vs v4 CD Chad Anomaly

NG Niger

GHCN v3.3 vs v4 NG Niger Difference

GHCN v3.3 vs v4 NG Niger Difference

Not much happening in the changes. Generally about 1/2 C of semi-random and 1 C range of extreams. What’s the anomaly plot look like?

Very wide range of volatility in the distant path, narrow waist in the 1950 to 1960 part of the Baseline Period, then spreading back out but with highs no higher than past highs until post 2000. When we again lose all cold years and the highs get bumped up in the changes. This during the “pause”. Then that last data point erases all that loverly built up heat “trapped” by all those changes to the “unadjusted” data…

GHCN v3.3 vs v4 NG Niger Anomaly

GHCN v3.3 vs v4 NG Niger Anomaly

ML Mali

GHCN v3.3 vs v4 ML Mali Difference

GHCN v3.3 vs v4 ML Mali Difference

What a wild ride of changes history has had here! A 2 1/4 C range near the ’30s! Everything before about 1980 cooled, after that a mix.

The Anomaly polt looks like the present about the same as the 1930s and 1900 in v3.3 so they need to cool them about 1/2 C, then clip a little bit lower in the Baseline Period (note the “usual” low volatility “waistline” about 1950-70 with NO hot years. After that, it looks like about 2000 they stop having lows as low as the Baseline and highs that go nicely higher than the newly lowered 1930s. They have even managed to keep enough variation in the year to year data to avoid that low volatility tapering Duck Tail look. Nicely done! Whoever manicured Mali needs a bonus or at least a night out on the expense account.

GHCN v3.3 vs v4 ML Mali Anomaly

GHCN v3.3 vs v4 ML Mali Anomaly

UV Burkina Faso 15

GHCN v3.3 vs v4 UV Burkina Faso Difference

GHCN v3.3 vs v4 UV Burkina Faso Difference

Formerly Upper Volta, so has UV as the country code. The Difference Graph shows cooling of 1950 to 1970 in the Baseline Period, then warming, if gentle, afterwards. We’ve even got a couple of down spike adjustments in the recent data of about -1.2 C (but few of them.

What do anomalies look like? Not much, really. We still have the 1940s about as hot as now, the latest data point at dead zero, so no warming left. If the baseline had not been cooled (dropping those black dots to the red ones) we’d have almost no trend at all. A line across the bottoms at about -.75 C catches most of the low point limit until about 1995 when suddenly lows turn up and never come below the zero line again.

Personally, I most strongly suspect the MMTS rollout has instrumemnts with more concrete, buildings and “stuff” nearby as they are on a wire leash to a building. That, alone, would knock out a lot of the low excursion potential.

GHCN v3.3 vs v4 UV Burkina Faso Anomaly

GHCN v3.3 vs v4 UV Burkina Faso Anomaly

Equatorial Tropical Africa:

Being at the Equator, plus or minus a little, these ought to all be about the same “shape” and about the same changes. It’s hot, humid, and tropical pretty much all over. Usually with an ocean not too far away.

CV Cape Verde

GHCN v3.3 vs v4 CV Cape Verde Difference

GHCN v3.3 vs v4 CV Cape Verde Difference

A pattern that is becoming painfully familiar. A downward spike / revision during the Baseline Priod (that drop of 1/2 to 1 C between about 1976 and 1982; then rises in the best, most recent, post 1990 data of up to 1.5 C. So just what was so screwed up in our temperature records from 1990 to date that it needs 1 C and up to 1.5 C of “fixing”? And IF it were that screwed up, how can we know that the “fix” is right? Furthermore, when we know that hte ’60 and early ’70s were cooler (the Great Pacific Climate Shift happened about 1976 bringing warmer water to North America, at least, and shifting the climate – ALL Naturall – via ocean pattern changes) then why does it need MORE administrative cooling?

Looking at the anomal plot we can see why. It’s a small Island and isn’t warming up without some help. Basically ranging between about -1 and +1 anomaly with no trend. So time to take some “tucks” in the baseline and try to lift the more recent data a bit by erasing those cold years in v3.3 data.

GHCN v3.3 vs v4 CV Cape Verde Anomaly

GHCN v3.3 vs v4 CV Cape Verde Anomaly

SG Senegal

In the transition from Desert to Subtropical. On the coast, and wraped around The Gambia. Whith changes galore in the “Unadjusted” data.

GHCN v3.3 vs v4 SG Senegal Difference

GHCN v3.3 vs v4 SG Senegal Difference

About 1.5 C of changes. Looks like a taylor taking “a tuck here, let it out a bit there, fit is just fine!”

Can’t really see the purpose of the changes other than perhaps to hide that all the “Warming” happens post 2000 and in the changes to the data.

Looking at the anomaly plot, up to the 19-teens it’s has a range (volatility) of about 2.5 C, then suddenly that drops to about 1.5 C, then in the “Baseline Period” there’s a chunk from about 1950 to 1970 where it is down to 1 C. I really do think that a great deal of “Global Warming” comes simply from ignoring the fact that an average tends to suppress range and there were a LOT more thermometers in the baseline period than either side of it. Take out the range in the baseline, put it in a known cold period, then clip via QA some cold going excursions in the recent data and you get “Instant Global Warming!”… just about anywhere.

Looking at the anomaly plot, a line layed on the tops is basically flat to about 1985. As the bottoms rise about 1 C largely from volatility suppression. Then we “Do the jump” about 2000, and in v4 had to erase that high spot so “now” could be warmer than “then”. This is a frequent pattern too. Cooling the “just a while ago” to erase The Pause. Even with that, the bulk of any statistical “warming” is coming from reduced volatillity and removal of low excursions. At best, that’s all beneficial. At worst, it’s corrupted data.

IMHO the anomaly graph tells the tale. A huge PC Jump in 2000, rising a range of almost 4 C from just a year or two prior. Gee, was something political happening about 2000? Anyway, the current data is spot on 0 delta or lower, so it became necessary to erase that “bump” in 2000 to avoid endorsing The Pause, or worse, showing cooling.

Regarless, in the last 4 years all the “Global Warming” has left the building…

GHCN v3.3 vs v4 SG Senegal Anomaly

GHCN v3.3 vs v4 SG Senegal Anomaly

GA Gambia, The

GHCN v3.3 vs v4 GA Gambia Difference

GHCN v3.3 vs v4 GA Gambia Difference

THis one is fascinating to me. So much cooling of the recent data. 1.5 C worth. Are the data really crappy to the 1.5 C error band? If not, why the change? If so, how can we believe the 1/2 C of imputed “warming”?

GHCN v3.3 vs v4 GA Gambia Anomaly

GHCN v3.3 vs v4 GA Gambia Anomaly

PU Guinea-Bissau

Senegal, The Gambia, Guinea-Bissau, Guinea; they are all of a sort. Wrapped around each other and on the Atlantic edge of the African Sahara. Only Senegal is in the transition from desert to sub-tropical, and then only the northern half. By Ginea-Bissau, things ought to be far more alike than different. But their anomaly plots are all over the place.

GHCN v3.3 vs v4 PU Guinea-Bissau  Difference

GHCN v3.3 vs v4 PU Guinea-Bissau Difference

Not much changing other than erasing some cold years in the 1940s and 1970s (we see more of that below). Then we hit the “High Quality Data” of the recent 1990s to date and “fixes” have about a 2.25 C range. So are you saying we can’t trust readings to within 2 C? Even in the modern data? In the anomaly plot that erases a hot year in the late 90s and a cold year in the 2000s.

Other than the “Dip” in the Baseline Period where volatility drops to about 1/3 C, the record is basically flat to the discontinuity at 2000. Then we get The Jump! ™ with some years 2 C higher but the bulk just shifted about 3/4 C. Wonder if they got a new electronic thermometer then… Just sayin’… Looks like it is near more concrete now.

GHCN v3.3 vs v4 PU Guinea-Bissau  Anomaly

GHCN v3.3 vs v4 PU Guinea-Bissau Anomaly

GV Guinea 20

There are also way too many Guinea’s around the world. This is the one that’s on the coast next to Sierra Leone.

GHCN v3.3 vs v4 GV Guinea Difference

GHCN v3.3 vs v4 GV Guinea Difference

The same general pattern we see way too often. A subtile 1/10 or 2/10 C cooling of the general past, a bigger up to 1/2 C “dip” in the Baseline Period from about 1970 to 1985 in this case (Baseline for GISS and Hadley ranging from 1950-1980 and from 1960 to 1990 respectively). Then more volatile warming of records after 1990.

The anomaly plot follows the changes more than anything real, IMHO.

GHCN v3.3 vs v4 GV Guinea Anomaly

GHCN v3.3 vs v4 GV Guinea Anomaly

SL Sierra Leone

GHCN v3.3 vs v4 SL Sierra Leone Difference

GHCN v3.3 vs v4 SL Sierra Leone Difference

They can ignore those 1875 era warm reports, but the more recent 1900s and 1920s to 40s had to be cooled to trend better. Then the Baseline Period has a nice low volatiity narrow range and when volatility returns in the recent data, we get the Jump about 1995 to 1 C warmer. Only problem is that it’s supposed to be 1/2 C and it was supposed to start in abut 1940 with significant warming in the 1980s when Hansen was going BSC with Congress. You are not supposed to be flat to 1995 then do a Jump!

GHCN v3.3 vs v4 SL Sierra Leone Anomaly

GHCN v3.3 vs v4 SL Sierra Leone Anomaly

LI Liberia

GHCN v3.3 vs v4 LI Liberia Difference

GHCN v3.3 vs v4 LI Liberia Difference

This one is amusing. They ease into a 1/2 C drop of the baseline about 1980, then work back up to warming 1/2 C to 3/4 C toward the end of it. One wonders why until the anomaly plot is inspected. Cold data in the later 1990s had to be erased, then the early 1990s “lifted” a bit to avoid the apearance of a cooling trend. Then we “take a gap” followed by a “splice” on of new data that’s slightly warmed. Overall, it tilts the past a bit lower and smooths it out so a nicer “stick” to the “blade” spliced on.

Still needs work though. Not much acutal warming going on. After all, Sierra Leone just next to you manged a whole 1 C of spectular warming. Surely you can cook another 1/2 C into things.

GHCN v3.3 vs v4 LI Liberia Anomaly

GHCN v3.3 vs v4 LI Liberia Anomaly

IV Cote D’Ivoire

In the same strip of coastal and as those below, but with a very different shape to the graphs.

GHCN v3.3 vs v4 IV Ivory Coast Difference

GHCN v3.3 vs v4 IV Ivory Coast Difference

Minor cooling of the data until 1970, then a minor warming after. Erasing a cold year just after 2000.

I can see why they needed to do that. Other than a cold dip in the “New Little Ice Age” era of the 70s, they have no rising trend. So pull the past down a couple of tenths, push the present up a smidge, and you get a tiny trend. Enough to join the club.

GHCN v3.3 vs v4 IV Ivory Coast Anomaly

GHCN v3.3 vs v4 IV Ivory Coast Anomaly

GH Ghana

GHCN v3.3 vs v4 GH Ghana Difference

GHCN v3.3 vs v4 GH Ghana Difference

Right next to Togo. Get’s a very odd set of changes. Pumping up the 1900-1920 interval by up to 2 C, then about 1.25 C of range to sporadic “fixes” after 1990. I thought we were supposed to have good thermometers then?

Looking at the anomaly plot, looks like an attempt at volatility reduction in the deep past and trying to form a “stick” for the “blade” to be on, instead of a volatility wedge. 1950 to 1970 just crazy low volatility. All the highs and lows gone. Really? Then the reason for the odd ‘recent cooling’ becomes clear. Can’t have The Pause so need to cool down those 90s so the present reports look hot. From about 1960 to date you can almost fit a straight line to the lower bound, with a very clear removal of low going anomlies. QA process run amok?

GHCN v3.3 vs v4 GH Ghana Anomaly

GHCN v3.3 vs v4 GH Ghana Anomaly

TO Togo 25

A little slip of a country right up against Benin. They get a very similar “fix” too.

GHCN v3.3 vs v4 TO Togo Difference

GHCN v3.3 vs v4 TO Togo Difference

Talk about your blatant Fiddle Faddle. Cool the 1950 to 1990 Baseline Period exactly to the years by about 1/4 to 1/2 C increasing, then BANG! right on the end of it, start a gratual warming pattern up to 1/2 C but with a couple of “fliers” of +1 C and almost +2 C. Besides, now folks are going to notice that the 19-teens were about as hot as all the other recent data points but that one. Look, it would be easer to just erase the teens… Oh, wait…

The anomaly plot shows the pull down in the Baseline Period, and the erasure of the cold years in the 2000s. Then you really think you can sell a +2 C end point? Really? That’s too soon. Not till a couple of more years. Heck it’s 3 C higher than the low part of the Baseline.

GHCN v3.3 vs v4 TO Togo Anomaly

GHCN v3.3 vs v4 TO Togo Anomaly

BN Benin

A little sliver of a country on the coast. But such crude fiddle faddle.

GHCN v3.3 vs v4 BN Benin Difference

GHCN v3.3 vs v4 BN Benin Difference

Chop out the cold 1970s, drop the Baseline Period 1/2 C, then bounce up the present by up to 1/2 C. Just crazy obvious what you are doing. Look at Nigeria (your neighbor) below. They hide it much better.

In the anomaly plot we can see the easure of the cold ’70s, and then pulling the 80s and 90s down by 1/2 C to make a cooler baseline, finally the “The Jump!” ™ in 2010 to “make a trend” and finally leaving the most recent data lone so it doesn’t stand out. Really guys? That’s the best you can do? This just hollers “Data Diddle”, you now. Folks are gonna talk. I’m gonna do a “Dig Here!” on it. Is that really what you wanted?

GHCN v3.3 vs v4 BN Benin Anomaly

GHCN v3.3 vs v4 BN Benin Anomaly

NI Nigeria

GHCN v3.3 vs v4 NI Nigeria Difference

GHCN v3.3 vs v4 NI Nigeria Difference

Compared to Cameroon just below, Nigeria is doing a vern subtle job of it. Only a minor cooling of the pre-1940s past by about 1/4 C, taking out a bit of cold in 1940 to make it look more like a flat stable past, Nice low volatility baseline period with highs suppressed, and then a gap and instant subtle “Global Warming” of an appropriate 1/2 C after 2000 with suppression of cold going anomalies.

Nicely Done Nigeria! Now go talk to Cameroon. They are an obvious mess with ham handed Diddle.

GHCN v3.3 vs v4 NI Nigeria Anomaly

GHCN v3.3 vs v4 NI Nigeria Anomaly

CM Cameroon

Can you say “Trying too hard”? I knew you could. Just look at that Difference Graph. A 2 C range to the changes. That’s just trying too hard to please. Warming everything after 1995 by an average of about 1 C is just showing off too much.

GHCN v3.3 vs v4 CM Cameroon Difference

GHCN v3.3 vs v4 CM Cameroon Difference

Then what do we see in the Anomaly Plot? Bit of cold coming out of the LIA. Then a long gap in the warm 1920s to 1950s, essentially dead flat to about 1995 with the “pinched waist” we see a lot around the Baseline Period especially prounounced in the 60s. Then, while the black dots stay flat, the New Improved Warmed!!! Red Dots make a great rising Duck Tail Anomaly. Beautifully tapering volatilty to near zero at an almost 2 C Peak Of Roasting hot! But other countries in Africa report a cold recent couple of years. And CO2 was supposed to be warming us most between 1950 and 1990, and after 2000 was The Pause in satellite data. Dear Cameroon, I think you are trying a bit too hard. Had you backed off the changes to a more ignored 1/2 C you would be in line with CO2 Theory and also not standing out like an over enthusiastic 3rd grader.

Just look back up at Nigeria. They did it much better. A very hard to spot gentle 1/2 C of post 2000 fudge. Look, you share a border with them and the same coastline, weather, and climate. People will talk if you are too different. Sober up, man! And don’t diddle the thermometer data so much, get a room where folks wont see you.

GHCN v3.3 vs v4 CM Cameroon Anomalies

GHCN v3.3 vs v4 CM Cameroon Anomalies

CT Central African Republic

Just below Chad and the desert. Just inland from Cameroon. In the Tropical Savanna.

GHCN v3.3 vs v4 CT Central African Republic Difference

GHCN v3.3 vs v4 CT Central African Republic Difference

Despite huge 1.5 C range changes in the historical data between v3.3 and v4, not much “Global Warming” in the anomaly plot. Here, too, they gently cool the past by about 1/4 C (but put some +/- variation in the shift) but, LOOK! A new trick! They just drop and erase those pesky warm 1940s! How elegant. Folks can’t see what isn’t there, after all. The cold 1970s get the “fix” eraser too. The net result being to remove a natrual hot to cold cycle and replace it with a generally 1/2 C cooler past. Then, after 2000, some wide ranginhg “Putty and Spackle Work” to tidy up the tail. Still needs some work but at lest now it looks like some warming instead of just a cycle going nowhere.

GHCN v3.3 vs v4 CT Central African Republic Anomaly

GHCN v3.3 vs v4 CT Central African Republic Anomaly

EK Equatorial Guinea

No data in v3.3 so all we get is the v4 anomaly plot. Which is a short record with about zero. A couple of cool years around 2002 and 2005, otherwise not much of note.

MariaDB [temps]> SELECT COUNT(deg_C) FROM anom3 WHERE abrev='EK';
+--------------+
| COUNT(deg_C) |
+--------------+
|            0 |
+--------------+
1 row in set (0.70 sec)
GHCN v3.3 vs v4 EK  Equatorial Guinea Anomaly

GHCN v3.3 vs v4 EK Equatorial Guinea Anomaly

TP Sao Tome and Principe 30

Sleepy little island off the coast. Nothing happening. No Global Warming. No changing. Ignored.

GHCN v3.3 vs v4 TP Sao Tome and Principe Difference

GHCN v3.3 vs v4 TP Sao Tome and Principe Difference

Difference graph dead flat. Anomaly plot dead flat. Better stop taking their data in the 1980’s…

GHCN v3.3 vs v4 TP Sao Tome and Principe Anomaly

GHCN v3.3 vs v4 TP Sao Tome and Principe Anomaly

GB Gabon

GHCN v3.3 vs v4 GB Gabon Difference

GHCN v3.3 vs v4 GB Gabon Difference

This one is just a classic. Anomaly Plot of black spots not rising, so cool the past by a few 1/10s C and after 2000 put some juice in it, baby! By 1/2 to 3/4 C. Go for it!!!

Then the anomaly plot, while flat to falling until the year 2000, suddenly does “The Jump!” ™ by, wait for it, 1/2 to 3/4 C. Instant Global Warming! Buy it now, in stores everywhere!!! /sarc;

GHCN v3.3 vs v4 GB Gabon Anomaly

GHCN v3.3 vs v4 GB Gabon Anomaly

CF Congo (Brazzaville)

Too many “Congos”. This one is the one on the other side of the river from the former Zaire and closer to the coast. It is a strip between the coastal Gabon and the more inland D.R. Congo.

GHCN v3.3 vs v4 CF Congo Brazzaville Difference

GHCN v3.3 vs v4 CF Congo Brazzaville Difference

The Difference Graph is a bit bizzare. 1.5 C range of change. A couple of 1/10s C of cooling the deep past, a bump up of 1/2 C of the start of the Baseline Period, then a pull down of 1/2 C to 1 C in the second half. Finally a pump of up to 1/2 C in the recent data. Error much?…

The anomaly plot showd the typical low volatility “pinched waist” effect around 1960. Manicured data or just the result of averaing a lot more thermometers so any variation gets suppressed? Need to count up the thermometers by year here. The past is volatile to the 2 C range, 1960 about 1/3 C then volatility expands back out to about 1 C of range. looks like they trimmed out that low pulse in the 1970 range then lifed the recent data to remove The Pause. Ending on a hot point (despite other countries around that end of Africa having a cold year…)

GHCN v3.3 vs v4 CF Congo Brazzaville Anomaly

GHCN v3.3 vs v4 CF Congo Brazzaville Anomaly

CG Congo (Kinshasa)

This one is the former Zaire. Central to the south of Africa, where it can be homgenized in to almost anywhere in the southern 1/3 of the continent.

GHCN v3.3 vs v4 CG Congo Kinshasa Difference

GHCN v3.3 vs v4 CG Congo Kinshasa Difference

Rather dramatic changes to the historical data between v3.3 and v4. We’ve got nothing changed until about 1975 Then changes with a 1.75 C spread, mostly adding cooling in the Baseline Period until about 1990, then shifting to warming after 2000 with most of the changes in the first 1/2 C of Plus range.

The anomaly plot? Very strange look to it. That “pinch” in 1960 looks like someone took out too much range of warming. Then the big “suck down” of the black spots into the 90s is a bit obvious. Yet still the plot is basically dead flat to 2000. Even staying near the zero line to about 2008. Then BANG! 1 C of “Global Warming”! Looks very much NOT like CO2 gradually warming the planet since 1940, and a lot more like “Instrument Change” or “Diddle”. Pick one.

GHCN v3.3 vs v4 CG Congo Kinshasa Anomalies

GHCN v3.3 vs v4 CG Congo Kinshasa Anomalies

RW Rwanda 35

No data in v3.3 so all we get is the v4 Anomaly Plot.

MariaDB [temps]> SELECT deg_C FROM anom3 WHERE abrev='RW';
Empty set (0.49 sec)

MariaDB [temps]>

Not in version 3.3, so no Difference Graph, just the V4 anomaly plot.


MariaDB [temps]> SELECT mean3 FROM yrcastats WHERE abrev='RW';
+-------+
| mean3 |
+-------+
|  NULL |
|  NULL |
|  NULL |
|  NULL |
[... more NULLS]
|  NULL |
|  NULL |
+-------+
62 rows in set (0.06 sec)

MariaDB [temps]> 

So the present is almost exactly zero warming ( +/- 1/2 C each side of zero over a couple of years) and the only notable point is that it was quite cold in 1970-1985. “The Jump” from 2000 to about 2015 is a nice touch, that being the period of “The Pause” and all (must erase The Pause…), but couldn’t hold onto it in recent data. A “Watch Here” for future “fixing”…

GHCN v3.3 vs v4 RW Rwanda Anomaly

GHCN v3.3 vs v4 RW Rwanda Anomaly

BY Burundi

GHCN v3.3 vs v4 BY Burundi Difference

GHCN v3.3 vs v4 BY Burundi Difference

Just not with the program in Burundi. Cooling in the Difference Graph of about 1/10 C. That’s all you can do?

Then the anomaly plot is about as dead flat as you can get. Let’s just forget those old years (they can be homgenized in from somewhere cooler) and cut off new data after 1990 (homogenize that in from somehwere with a better Duck Tail flip) and use this just as Baseline cool.

GHCN v3.3 vs v4 BY Burundi Anomaly

GHCN v3.3 vs v4 BY Burundi Anomaly

UG Uganda

GHCN v3.3 vs v4 UG Uganda Difference

GHCN v3.3 vs v4 UG Uganda Difference

This one is a hoot. In this case, look first at the Anomaly plot below. BIG gap between about 1980 and 1990 (I think they were having a war then). Volatility at the start is about 2 C range. It narrows to about 1 C in 1970. Then there’s a cold year. Now notice that the Diffrence Graph above ends in 1980. It is changed to 1/2 C colder. That’s the usual “cooling the Baseline Period”. When readings return, in the 90’s, they are a solid 1/2 C and sometimes 1 C higher. So all of “Global Warming” here happened in the 10 year gap. Mind the gap!

GHCN v3.3 vs v4 UG Uganda Anomaly

GHCN v3.3 vs v4 UG Uganda Anomaly

KE Kenya

GHCN v3.3 vs v4 KE Kenya Difference

GHCN v3.3 vs v4 KE Kenya Difference

Taking a little tuck in the hot 1920’s to 30’s of about -1/2 C (got to get rid of that pesky warm past). Then jittery around zero up to 1960 with a steady cooling of the Baslne Period up to 1990, a bit of an overshoot on the cooling hisotry to 2000. At that point the changes go more volatile with about 1 C range to the “error correction” – and here I thought we were supposed to have electronic thermometers accurate to 1/100 C in this millenium… SO why a 1 C “diddle” in data from this last decade?

The anomaly chart is amusing too. The old hot past prior to 1930 gets pulled down hard. Volatility between about 1940 and 1975 is kept way low at around 1/2 to 3/4 C. Then in 1980 the v3.3 data takes a rocket ride up of 1.3 C or so. Guess that stood out so it gets pulled back down keeping the most recent data as “warming”. Can’t have it “Pause”, you know. Then that cold year around 1998 gets erased with a huge 1.25 C adjustment / fix / fudge / “whatever you call changes in the unadjusted data”…

Would be interesting to check newspapers from there, then, to see if there was a cold year reported.

GHCN v3.3 vs v4 KE Kenya Anomaly

GHCN v3.3 vs v4 KE Kenya Anomaly

TZ Tanzania

GHCN v3.3 vs v4 TZ Tanzania Difference

GHCN v3.3 vs v4 TZ Tanzania Difference

What do you do with a country with no trend? Add one! So the Basline Period from about 1975 to 1990 gets pulled down 1/2 C, then you bump up after 2000 by 1/2 C to 1 C, and what does the anomaly plot do? Gives you warming, sort of, if you don’t look back at the 1920s.

I like the way it’s warm in the 1920s, then we have a drop into about 1975, then a wobble but still low in the 1990. One wonders where CO2 was all those decades of cooling…. Then, suddenly, in 2000 and a bit “Global Warming!!” bursts onto the scene in one year. We pop up in just a couple of years by about 1 C to 1.5 C, but caan’t hold onto it. The most recent data points being colder than the average… I guess that means “Tanzania needs work” ;-)

GHCN v3.3 vs v4 TZ Tanzania Anomaly

GHCN v3.3 vs v4 TZ Tanzania Anomaly

Southern Horn Of Africa:

AO Angola

GHCN v3.3 vs v4 AO Angola Difference

GHCN v3.3 vs v4 AO Angola Difference

Wow! A full on pumping of the last half dozen years. A nice small cooling of the deep past.

Then the anomaly plot shows a nice cooling of the deep past… But wait, were not other places warmer in the 1920s to 1940? This shows Angola cooler. Then a sudden jump of 1 C that holds flat until 1990, then another jump to 2 C warmer “now”. But CO2 doesn’t cause 2 C of warming. That’s what they say is in our future scare story…

GHCN v3.3 vs v4 AO Angola Anomaly

GHCN v3.3 vs v4 AO Angola Anomaly

WA Namibia

GHCN v3.3 vs v4 WA Namibia Difference

GHCN v3.3 vs v4 WA Namibia Difference

So per this Difference graph, Namibia has 1.5 to 2 C of “error” that had to be taken out from v3.3 to v4.

Again we have flat to declining through 1980 on the Anomaly plot, then a step up of about 2 C and a widneing dispersion into the present. Not in keeping with CO2 theory but looks more like instrument change at a point in time.

GHCN v3.3 vs v4 WA Namibia Anomaly

GHCN v3.3 vs v4 WA Namibia Anomaly

BC Botswana

GHCN v3.3 vs v4 BC Botswana Difference

GHCN v3.3 vs v4 BC Botswana Difference

A general systematic cooling of the data prior to 1980, then wide ranging volatile but sparce changes recently.

The anomaly plot is essentially dead flat to 1980, then a Step Function higher by 1 C to 2000, then it drops back a bit but without much trend.

GHCN v3.3 vs v4 BC Botswana Anomaly

GHCN v3.3 vs v4 BC Botswana Anomaly

ZA Zambia

GHCN v3.3 vs v4 ZA Zambia Difference

GHCN v3.3 vs v4 ZA Zambia Difference

Was there somehing crazy happening in Zambia between 1990 and now to justify didding their recorded data by a 2.5 C range of Diddle? In the recent and supposedly best data?

The anomaly plot is a mess too. General cooling from 1920 to 1980 by roughly 2 C (so 2 C of temperature change can be “Natural Causes”, right?) up to 1980. Then it rockets up by 2 C and volatility goes wild. How can this happen here and not in any country around it?

GHCN v3.3 vs v4 ZA Zambia Anomaly

GHCN v3.3 vs v4 ZA Zambia Anomaly

ZI Zimbabwe 45

GHCN v3.3 vs v4 ZI Zimbabwe Difference

GHCN v3.3 vs v4 ZI Zimbabwe Difference

Cooling the 1930s to 1970, so I guess they had it hot then, Then 1/2 C of warming the data until the approach to 2000, when the changes go way nutty. A 2 C range of “fixes”. So 2 C error band, right? Right?…

The Anomaly plot has general cooling through 1980, a rise back to normal into about 1990 (now “lifted” by 1/2 C of change of the data for v4) and then a mess at the end.

GHCN v3.3 vs v4 ZI Zimbabwe Anomaly

GHCN v3.3 vs v4 ZI Zimbabwe Anomaly

MI Malawi

GHCN v3.3 vs v4 MI Malawi Difference

GHCN v3.3 vs v4 MI Malawi Difference

This one is just a mess. Difference graph with a 2 C range in the “Latest And Greatest” temperture anomalies. Otherwise just a touch of cooling of the Baseline Period.

Anomaly plot showing nothing much at all happening until 1980, then “The Jump” through 2000, that they then had to erase with these changes so “now” would look warmer, but it wasn’t, so it had to get juiced too. Now we’ve got about 4 years pegged high with near zero volatility and 2 C of warming.

So really, how does CO2 do that? How does it do nothing until 1980, “Then Jump!” up by 2 C, all while other countries are saying last year was a cold year?

GHCN v3.3 vs v4 MI Malawi Anomaly

GHCN v3.3 vs v4 MI Malawi Anomaly

MZ Mozambique

GHCN v3.3 vs v4 MZ Mozambique Difference

GHCN v3.3 vs v4 MZ Mozambique Difference

A real mess of a Difference graph. A full 2 C range to the changes in historical data. So was GHCN v3.3 bogus to a range of 2 C? Was it? If not, why the changes?

Basline gets a nice 1/2 to 1 C dropout. 2000 gets a 1/2 to 1 C boost.

The anomaly plot shows the erasure of a warm 1960s and a lifting of the present to make a trend. ALL the warming comes out of the changes of the data.

GHCN v3.3 vs v4 MZ Mozambique Anomaly

GHCN v3.3 vs v4 MZ Mozambique Anomaly

WZ Swaziland

Not in v3.3 so no Difference graph:

MariaDB [temps]> SELECT COUNT(deg_C) FROM anom3 WHERE abrev='WZ';
+--------------+
| COUNT(deg_C) |
+--------------+
|            0 |
+--------------+
1 row in set (0.69 sec)

MariaDB [temps]> 

Nice dropout in the Baseline Period, so this data will be compared with a cold artificial baseline from nearby data (where “nearby” can be 1200 km away) in the GIStemp / Hadley methods.

Otherwise, not any real warming in the anomaly plot.

GHCN v3.3 vs v4 WZ Swaziland Anomaly

GHCN v3.3 vs v4 WZ Swaziland Anomaly

SF South Africa

GHCN v3.3 vs v4 SF South Africa Difference

GHCN v3.3 vs v4 SF South Africa Difference

A general small cooling of the past up until about the start of the Baseline Period. Then it widens out to about 1/2 C of range to the changes.

The anomaly plot is essentially dead flat to 1980 (so where was “Global Warming” then, eh?) with some fast cycles at about the length of a sun spot cycle. Then it pivots up, mostly on higher lows, to 2000 when those changes help raise the highs just enough higher to make a trend.

GHCN v3.3 vs v4 SF South Africa Anomaly

GHCN v3.3 vs v4 SF South Africa Anomaly

An interesting thing to ote about South Africa, is that depsite all those changes between the two sets, the net change is nearly zero. Only is the time of change important…

MariaDB [temps]> SELECT AVG(mean4-mean3) FROM yrcastats WHERE abrev='SF' ;
+------------------+
| AVG(mean4-mean3) |
+------------------+
|        -0.094088 |
+------------------+
1 row in set (0.07 sec)

So take the average of ALL the diffrences between v3.3 and v4, it is nearly zero. It truncates to 0.0 C but rounds to 9/100 C. Effectively zero. Yet all the data change. Isn’t that a remarkable thing, for so much change?

LT Lesotho 50

GHCN v3.3 vs v4 LT Lesotho Difference

GHCN v3.3 vs v4 LT Lesotho Difference

Interesting that there’s so little data in v3.3 for the Differenc graph, yet so much more v4 in the Anomaly Plot. The changes have a 1 C range to them, so again I must ask “If 1 C is administrative diddle, how can you claim 1/2 C of ground truth Climate Change?”. You have here a statement that there’s 1 C of error band. Own it.

Not much warming going on in the anomaly plot. More nearly cooling recently.

GHCN v3.3 vs v4 LT Lesotho Anomaly

GHCN v3.3 vs v4 LT Lesotho Anomaly

Southern Hemisphere African Islands:

MA Madagascar

GHCN v3.3 vs v4 MA Madagascar Difference

GHCN v3.3 vs v4 MA Madagascar Difference

Next to, and largely surrounded by, the islands that follow; Madagascar ought to look a lot like them. But it doesn’t. Guess the French and the English don’t have as much influence there. The Difference graph shows little being changed until 1940, just a couple of patches of change. Then we get 1/4 C of cooling until about 2000 when it swaps to warming a smidgeon.

The Anomaly plot is fascinating. This is what I would expect an island located in a place with cyclical current changes to look like. A fairly consistent about 1/2 C range, but one that wanders cyclically up and down about 2 C. Looks like about an 80 year cycle to me. I don’t see much being wrong with Madagascar data. Even the 1/2 C of “adjusting the unadjusted” in the Difference chart doesn’t manage to bias it too much.

This would imply that other islands with short records might well show a 1 C rise from 1980, but simply lack the older data to show it is just a regular cycle.

GHCN v3.3 vs v4 MA Madagascar Anomaly

GHCN v3.3 vs v4 MA Madagascar Anomaly

MF Mayotte [France]

GHCN v3.3 vs v4 MF Mayotte Difference

GHCN v3.3 vs v4 MF Mayotte Difference

The Difference graph looks like they needed to remove the hot excursion from 1998 so as to let “now” be hotter. Generally a nice steady 1/4 C cooling of the past added in, and then recent data warmed just 1/4 C. Gotta love the precision and finesse of the French.

The anomaly plot is also nicely finished. Polishing off those rough high excursions of the late 1990s, putting a wonderfully fine Duck Tail on the present. Narrowing the volatitlity range of the 1950 of about 1 C down to a slim 1/4 C in the present. Finally, the coup de gras, a stellar 1 C hotter “now”. No cold year for them.

GHCN v3.3 vs v4 MF Mayotte Anomaly

GHCN v3.3 vs v4 MF Mayotte Anomaly

CN Comoros

GHCN v3.3 vs v4 CN Comoros Difference

GHCN v3.3 vs v4 CN Comoros Difference

Nothing much happening in the change departement but some slight rollling until the 1970s, then we start getting warming added to about 1/2 C. Guess they got the memo that the goal was 1/2 C of “Global Warming”.

The anomaly plot reflects that with things nicely stepping up about 1/2 C to the cutoff in 1980… for version FOUR data. The v3.3 continues on, but not warming. I guess all they could do was truncate that non-warming and let it have some warm “homogenized” in from other islands with bigger airports…

GHCN v3.3 vs v4 CN Comoros Anomaly

GHCN v3.3 vs v4 CN Comoros Anomaly

JU Juan De Nova Island [France]

MariaDB [temps]> SELECT COUNT(deg_C) FROM anom3 WHERE abrev='JU';
+--------------+
| COUNT(deg_C) |
+--------------+
|            0 |
+--------------+
1 row in set (0.34 sec)

MariaDB [temps]> 

A French island with a very short record. It looks like a nice smooth trend, but is mostly located after the point where longer records “take a knee” and turn upward. Over about 40 years, has near 1.5 C of rise. That’s not the CO2 thesis, so what is it?

GHCN v3.3 vs v4 JU Juan De Nova Island Anomaly

GHCN v3.3 vs v4 JU Juan De Nova Island Anomaly

EU Europa Island [France] 55

MariaDB [temps]> SELECT COUNT(deg_C) FROM anom3 WHERE abrev='EU';
+--------------+
| COUNT(deg_C) |
+--------------+
|            0 |
+--------------+
1 row in set (0.73 sec)

A nice steady rise, of about 1.5 C, for the lows. Similar for the highs. Yet the most recent dot is below zero anomaly, so essentially average. The French do a nicer job of sculpturing the data, IMHO.

GHCN v3.3 vs v4 EU Europa Island Anomaly

GHCN v3.3 vs v4 EU Europa Island Anomaly

TE Tromelin Island [France]

GHCN v3.3 vs v4 TE  Tromelin Island Difference

GHCN v3.3 vs v4 TE Tromelin Island Difference

Again with cooling the past, going almost neutral near the later baseline period, then a bit of lift to the present. / recent data. The anomaly graph also has a nice steady rise to it, now augmented by the recent changes. . BUT the pivot comes about 1990. That’s when the range decreases from about 1 C to about 1/2 C and down closer to 1/4 C in the last few years. The anomly does a small jump then, and another just after 2000. Looks like the changes are partly to erase those step functions and smooth out the “warming”.

GHCN v3.3 vs v4 TE  Tromelin Island Anomaly

GHCN v3.3 vs v4 TE Tromelin Island Anomaly

RE Reunion [France]

GHCN v3.3 vs v4 RE Reunion Difference

GHCN v3.3 vs v4 RE Reunion Difference

Nice adjustment pattern to the “unadjusted”! Pull the past down by about 1/4 C up to the 2000 pivot point, throw in some random spots up a smidge, then warm the post 2000 smootly up to 1/4 C. That way you bag a nice 1/2 C of warming, spot on the prediction of “Global Warming”. Just by a minor re-write of the history from THE definitive data set of 2015. So will they say it was crappy prior to 2015?

The anomaly plot is almost as nicely done. They do still have the narrowing volatility Duck Tail problem. Things start out cold and with a range of about 1 C. This continues to about 1985 – 1990 when range decreases, the lows rise, and the highs start rising about 1998. Interesting to note that the high here is over 1 C high while the other places have generally shown a cold recent year. Someone missed the memo…

GHCN v3.3 vs v4 RE Reunion Anomaly

GHCN v3.3 vs v4 RE Reunion Anomaly

SE Seychelles

GHCN v3.3 vs v4 SE Seychelles Difference

GHCN v3.3 vs v4 SE Seychelles Difference

In the same water as most of the rest of these islands, yet with a very differnt profile or “shape” of the anomaly plots. The Difference graph puts a bit of ‘lift’ into the period from 1950 to 1970, then starts lowering things. But not by much. My guess would be that the period near 1940 is ‘way cool’ so they can lift it some to balance their statistics on warming vs cooling adjustments, while not changing any trend.

The overall Anomaly graph has a generally flat upper bound at about +1/2 C until near 2000. Then there are some nearly zero volatility years and it does the turn upward by about 1/2 C more. The bottoms would be characterized by a line about -1/2 C with occasional excursions down to -1 1/3 C in about 1905 and 1945. Those excursions then end and the bottom makes all the “warming” trend as volatility narrows significantly. I note in passing tha the most recent data are at zero. So “Global Warming” has left the building.

GHCN v3.3 vs v4 SE Seychelles Anomaly

GHCN v3.3 vs v4 SE Seychelles Anomaly

MP Mauritius

GHCN v3.3 vs v4 MP Mauritius Difference

GHCN v3.3 vs v4 MP Mauritius Difference

Not a lot changes, but maybe that’s because they already have a beautiful Duck Tail flip in their anomaly plot? These island have a full 2 C of difference between the old lows and the new lows. As that is also way more than the CO2 prediction or claim to date, I once again wonder “Where does it come from?” The anomalies are basically dead flat with highs at about 1/4 C above the zero line and lows about 3/4 C below up until 2000. THen there’s a clear pivot upward and compression of range from about 1 C to about 2/10 C. Has the climate of Mautitius been remarkably constant the last decade? Not a single cool year? Really?

GHCN v3.3 vs v4 MP Mauritius Anomaly

GHCN v3.3 vs v4 MP Mauritius Anomaly

IO British Indian Ocean Territory 60

GHCN v3.3 vs v4 IO British Indian Ocean Territory Difference

GHCN v3.3 vs v4 IO British Indian Ocean Territory Difference

An odd tweek here. A general cooling of most of the past, then a set of alternating “about zero” and about 1/10 C of higher, then another near flat about zero, but with 1/10 or so below zero. Just odd.

The current anomaly is “about zero” so no accumulated “Global Warming” while there is a blip up by about 1 C in 1998. Back at zero about 1982, then I guess the “cooling the past” is there to try to remove some of the warmth from 1960 ish and make some kind of trend.

GHCN v3.3 vs v4 IO British Indian Ocean Territory Anomaly

GHCN v3.3 vs v4 IO British Indian Ocean Territory Anomaly

FS French Southern and Antarctic Lands [France]

GHCN v3.3 vs v4 FS  French Southern and Antarctic Lands Difference

GHCN v3.3 vs v4 FS French Southern and Antarctic Lands Difference

Not sure exactly where these are, but we once again have the “cool the past warm the present” changes in the data. There are some token adjuments upward (in this “unadjusted” data) in the early years, the bulk of the data points x distance is below the zero line prior to 2000, then it goes to predominantly above. Likely another good candidate for closer scrutiny.

The anomaly graph only really has “trend” in the changed red spots. The black spots rise some from 1970 to 1980 but then are basically dead flat to about 2015 when they end. The “warming” is in the changing to higher in the v4 version of history.

GHCN v3.3 vs v4 FS  French Southern and Antarctic Lands Anomaly

GHCN v3.3 vs v4 FS French Southern and Antarctic Lands Anomaly

SH Saint Helena [United Kingdom]

Looks like it’s not in v3.3 ( I wonder where it was accounted then, if anywhere?)

MariaDB [temps]> SELECT COUNT(deg_C) FROM anom3 WHERE abrev='SH';
+--------------+
| COUNT(deg_C) |
+--------------+
|            0 |
+--------------+
1 row in set (0.69 sec)

MariaDB [temps]> SELECT mean3 FROM yrcastats WHERE abrev='SH';
+-------+
| mean3 |
+-------+
|  NULL |
|  NULL |
|  NULL |
[... repeated NULL...]
|  NULL |
|  NULL |
|  NULL |
|  NULL |
+-------+
128 rows in set (0.06 sec)

MariaDB [temps]> 

Here’s the Anomaly graph. Nice steady rise to it…

GHCN v3.3 vs v4 SH Saint Helena Island Anomaly

GHCN v3.3 vs v4 SH Saint Helena Island Anomaly

Though it looks like about 2 C+ of rise. Rather a lot more than the supposed 1/2 C from “Global Warming” we’ve been told has happened. Wonder what caused it?

In Conclusion

I put most of my conclusions in the introduction. This one was grueling to do. So much so that even the Fire Fox Spell Checker, even running on the Odroid N2 with 4 GB of memory, crapped out about 3/4 of the way through and stopped giving errors messages, so “here there be typos”…

But now I’m done.

No, I’ve not done my usual QA read through for errors, typos, and consistency. I’ll do that later, maybe. After a couple of days off ;-)

Next up is Europe, and while there are fewer countries than Africa, it is almost as much. I may break it up into a couple of parts just to avoid the Mammoth Posting problems.

Just be glad all you have to do is look at a few graphs and read a comment or two about them ;-)

My overall impression of the graphs and the data is that there is a “Tayloring” operation going on. The changes are NOT just a little fix up here and a correction there. It looks to me like it has direction and purpose. Cool the Baseline Period. Cool warm past periods. Warm the recent data UNLESS it is too high in the last 2 decades, then you cool them so the nearest data can look warmer in comparison. Stamp out cold periods in the middle. Remove cool periods recently if not already suppressed. The quesiton that remains for me is just: “Is that an accident from ignoring the effects of Instrument Change, or a deliberate planned act?”

With that, I’m taking a break. Over to you folks for more analysis.

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 NCDC - GHCN Issues and tagged , , , . Bookmark the permalink.

19 Responses to GHCN v3.3 vs. v4 Anomaly Graphs – Africa

  1. rms says:

    wow.

  2. Steven Fraser says:

    @EM: Using the mk-1 eyeball, these charts have distinctive flavors of ‘look’, ranging from the almost ‘scatterplot’ to the highly regular matching. Very interesting.

  3. E.M.Smith says:

    @rms:

    Thanks for that! I need all the courage I can to repeat this on Europe that has almost as many countries… It is rather a lot of work, but the output is, I think, worth it.

    @Steven Fraser:

    Yeah. That’s what got me hooked enough to continue this for the whole world. The data as anomalies just look so “manicured”. There’s a couple of different patterns, but nearly none of them look like what you would expect in a natural system

    1) The natural range of variability of a place ought not change dramatically over the decades. Sure, a little change is to be expected, but if the “natural range” is about 2 C on the annual average anomalies, it ought not drop to 1/2 C in a year or two, then pop back up a few years later.

    2) “Errors” ought not be so concentrated that the period from 1950 to 1990 gets about 90%+ cooling and years after 2000 get about 100% warming. Similarly, the collapse of volatility (range) ought not be so tightly focused on the Baseline Period.

    3) “Step Functions” of 1 C and even 1.5 C are NOT what is expected from CO2, yet they are common. They never arrive before 1985, well after “warming” ought to have shown up (Hansen was testifying then, remember…).

    4) Why would natural variation cause so many places to “take a dip” just in the baseline period, then have volatility compression take out only low going anomalies and taper range into a very small point.

    5) Why are the same kinds of changes to the data seen in the Sahara Desert, the Tropics, and islands in the sea?

    6) WHY is the version 3.3 data, that we were told up until 2015 was Golden, Pristine, PERFECT for determining the fate of nations and $BILLIONS / year, over turning whole economic orders; why is it so crappy now that it needs 1 C, even 2 C of “fixes” across all countries and all years? Changing every single value for most countries. If it was that crappy, why should we believe v4 is any less crappy?

    It’s all just a huge shout of “Diddled Data” disastrous drek.

  4. Bill in Oz says:

    EM Just spent an hout gazing at charts and reading your comments.. Bugger ! Mind glazing stuff… But I’m glad you persisted !
    You write : ” there is a “Tayloring” operation going on. The changes are NOT just a little fix up here and a correction there. It looks to me like it has direction and purpose. Cool the Baseline Period. Cool warm past periods. Warm the recent data “.

    OK Point firmly established.

    Now the important question is “Who did it ?”
    Are you saying that the weather staff is each of these countries fiddled the data ? Or are you saying that the science staff at who ever puts together the global data did it ? And are you saying that the introduction of electronic thermometers in the 1990’s early 2000’s did it ?

    or it it a mix of all three ?

  5. A C Osborn says:

    It is obvious from your results that each set of “Data Collectors” have absolutely no faith in the Raw data and in what their predecessors did to correct it.

    Oh aren’t they all the same people?
    Blatant FRAUD and totally illegal in the US.

  6. Simon Derricutt says:

    EM – it took a while to even read all of that and to absorb the data, and it’ll need at least another couple of read-throughs to make sense of it. Presumably both V3.3 and V4 are advertised as raw and uncorrected data, so the pattern of the changes (I was going to call them corrections, but that’s the wrong word) seems to be a calculated fraud.

    It seems a comparison of V2 to V3.3, and V2 to V4, could be interesting, but it will also be pretty difficult because it’s unlikely that the datasets will be directly comparable and so there’s a lot of work in investigating each station. It may however be possible to select just a few stations where the same location is in all 3 databases and see how the “raw” data has changed. It looks like the main problem is going to be getting data that you’re sure is what was actually recorded at the time and hasn’t been changed. About the only way of being certain of that is to get the original hand-written logs and key them in – one big task.

    There’s a chance that your V2 data hasn’t been diddled that much, but even so it seems we couldn’t be sure of that.

    If this is the source data, it’s not surprising that the models are giving the wrong answers. That means that we’re spending a load of money to fix a problem that isn’t there, and so can’t be fixed. There’s a tacit admission that the error-bars are way higher than admitted, since the “corrections” can be 2°C or more. I’d suspect that the corrections are done using an algorithm, given the overall pattern.

    Getting the measurements a bit wrong is something scientists expect and can deal with. Buggering the data because it doesn’t fit the theory just isn’t allowed, and shouldn’t be allowed. “Filling in” temperatures from weather stations up to 1200km away is very obviously wrong, since I can get several degrees different by moving 10m away from a point, and by the time you get any considerable distance it’s just a guess. If the data is missing, it’s missing.

    In reality, it looks like the late 1930s were a bit warmer than the 2000s, and that calling 2016-2019 the “hottest evah” is basically a lie. Seems that the statistical methods used are invalid, and that also the data-diddling is lying to us. As in 1984, though, where the government holds all the official data, and inconvenient data is put into the memory-slot, it’s hard to prove anything.

  7. E.M.Smith says:

    @Simon:

    The v2 data format, stnID and all are almost 100% the same as v3. The reason I took on the v3.3 vs v4 “up front” was it is the “Hard Lift”. With it now working, adding the v2 data is fairly simple (making all the graphs not so much…). I have copies of v2 (and even v1) data squirreled away, so it will happen. Just throttled by my personal speed doing the work and rate of other folks deciding to use the code themselves and “Get ‘er done”. IIRC, v1 Station ID was somewhat different. but not too hard to convert to a common basis.

    One of my “someday ideas” is to just graft the latest parts of each series onto the next earlier so as to limit the buggerage and see what it looks like. The assumption being that they would not want to bugger the recent data too much as “folks might notice” and to remove the “cooling the past”. (Though a couple of countries here show recent data also subject to change…) In this way I would use v1 up to about 2000 (or whenever it ended) then v2 until about 2010 (or whenever it ended) and then v3 until v3.3 then after 2015 v4. That ought to preserve as much of the pre-diddle past as possible at each step.

    Also on my “someday” list is a mapping of Station IDs so that you can plot all the changes for any one station on a single graph. Basically have a spaghetti graph for, oh, NYC Central Park thermometer with 4 lines on it in 4 colors showing the thermometer wiggling over time as the past cools, the baseline sinks, and the present grows a duck tail…. but that depends on the creation of a New Unique Station ID to which each of the GHCN stnIDs in each era can be mapped. That’s a 7600 or so task for v1 through v3, then the added stations in v4 can likely be ignored as they are not in the earlier versions.

    FWIW, I think one of THE biggest problems is in the widely believed as an Article Of Faith that “using anomalies means instrument change doesn’t matter”. The “Climate Scientists” are happy to mix and match instruments, have thousands enter and leave the record, etc. and believe “The Anomaly Will Fix It 100%!”. Part of my goal with this series was to use a blatantly simple and clean anomaly process and then we STILL see that “Instrument change MATTERS”. And we did see that.

    Now, is that a deliberate lie and fraud, or just a stupid assumption that has gone unchallenged in the Church Of Climatology? I’ll leave that for others to sort out in “discovery”.

    What is absolutely clear to me, from looking at these graphs, is that the very early years with just one or two instruments can have quite broad ranging anomalies as any local excursion carries forward 100%. When they load up the dataset with thousands of extra thermometers in the “Baseline Period”, volatility compresses greatly. Why? Because when your thermometer field covers far more places each instrument is less correlated to the next and any excursion at one place gets muted and averages away in the bundle.

    In other words: Anomaly processing does NOT remove the volatility compression of larger numbers of thermometers.

    I’d further assert that “Volatility matters”. Especially to the low side. When Truckee (up near Lake Tahoe and Reno Nevada) takes a cold turn, it can drop 30 F. At the same time, here near the ocean, it can maybe drop 10 F. That “volatility to the downside” is removed when you decide to use “4 at the beach” in California and leave out the “high cold places”. The assertion that this is OK due to the “Reference Station Method” (Hansen about ’86?) is broken and needs to be overturned. This is a first step in that process.

    In the graphs we see this as the wide volatility range in the early years, the “waist” of very low volatility in the “Baseline Period” (where the assertion is that they added more thermometers to “improve the baseline” but IMHO it just taints the conclusions with a volatility error) then the expansion back to more volatility after the Baseline Period (but not as much as the very early years of just one thermometer).

    Where this breaks down in the places with only one thermometer for the whole period, so I need to take a look at places like Dijibouti with an eye to that. Basically I need to get down to the single thermometer records to demonstrate how volatility changes over time with just one station vs averages.

    What accounts for the pruning of “low going anomalies” with a linearly increasing loss over time after the baseline is anyone’s guess. Could be anything from QA algorithms that toss out anything too far different from “recent history” so gradually ratchets up, to human intervention, to concrete growth at airports and MMTS on a leash near the buildings. That’s a big “Dig Here!”.

  8. E.M.Smith says:

    @A C Osborn:

    Say you are a Government Employee trying to make retirement. You also want to visit a nice set of countries on the Government Dime by presenting papers. What do you choose:

    a) Prior folks did everything just right. All you need is my assistant to keep typing in the new data as it arrives. Lay me off now.

    b) It is critical to do further research into improving the data for better saving humanity, assuring my Post Doctoral funding, publishing some “landmark” papers and sending me to venues all over the world to make speeches (preferably nice places with 5 star accommodations and recreation).

    Now, once you chose “b” (was there any doubt?…) you simply MUST have some “work product” to demonstrate that you are working and to bolster your desire to “publish” (and attend conferences and…) so it is a simple imperative to “Diddle the data” every so often to “demonstrate your worth”. If there is nothing to correct or “improve” , YOU are not needed.

    @Bill In Oz:

    I think it is a mix, but do not know.

    Remember that many of the national BOMs are “True Believers”. Then remember that GHCN says this is “unadjusted” then tosses in the caveat that it is unadjusted by THEM but the upstream BOM can do whatever the want with it first.

    This means, for example, that a “Fellow Traveler” (perhaps after discussions at a “conference on Climate Change”) in, say, Australia might institute a “New & Improved” temperature QA and pre-processing step that homogenizes temperatures over a 400 km range. That right there will reduce volatility, and since we know volatility upward is limited by water effects and convection (thunderstorms and hurricanes start moving a LOT more heat once 85 F water is reached and convection over land becomes extreme at about 110 F) but low going volatility has not such things limiting it; then that process will reduce “low going excursions” more than high going. Now that is classified as “raw data” as it arrives in the GHCN, yet it isn’t.

    Then the GHCN construction uses a different set of data to “improve the baseline” by adding a bunch of historical data (by what processes from where with what QA that might reject extreme cold events as a claimed “error”…) and then the profile of GHCN differs over time compared to the recent Australia data and compared to the reality of old temperatures.

    So I could see a case where it is a Comedy Of Errors with folks making invalid assumptions and the errors just keep stacking up. I could also easily see a clever set of folks “choosing their errors wisely” for effect. There’s no data to disambiguate those two, really, other than a circumstantial case that “change always goes one way and that isn’t random error”. Then there is Noble Cause Corruption acting one way only.

    Were I to GUESS, I’d guess this way:

    1) There is SOME corruption and Noble Cause Corruption. We’ve seen that in their eMail and in Hanson’s constant hand wringing and bleating..

    2) Some of it is just stupid on steroids. Folks do NOT like to admit they are wrong. EVERYONE on the Warmista side repeats the mantra that “Instrument change doesn’t matter because we use anomaly processing”. Yet nobody seems to test it. It would shake the foundations. Yet every chemist knows that’s Bull Shit in calorimetry (and this IS a global calorimetry exercise, let there be no doubt).

    3) Some of it is politically driven / purchased. Say you are the meteorologist in SomeDinkyLand and you have been told your cut of the $200 Billion / year sucked out of the USA by the UN will be about $1 Billion. Your Presidential Commission On Climate Change representative asks you what you are doing to demonstrate the “harm” being done to SomeDinkyLand by Climate Change. Are you going to answer “Why, our temperatures have not risen, it’s a crock.”? Or are you going to say “Why look at this! By use of my new QA method that removes low errors, we are drowning in Climate Reparations Harm! Please sign here to authorize my trip to Paris to collect our share.”? Then that changed data rolls up hill into GHCN for further tuning.

    But those are just guesses based on human nature and the money / status involved.

  9. E.M.Smith says:

    BTW, I think that “find the several causes” approach explains things fairly easily. The “Baseline waist” is the result of ignoring the effect of averaging on volatility. The lowering of the baseline comes from the belief that anomaly process means particular instruments don’t matter; so when GHCN adds a load more instruments into the baseline to “improve the quality” it also has an effect that they interpret as “more accuracy” when it is really statistical artifact. Then for some places, like perhaps Dijibouti and other DinkyLand places where the “change” looks blatantly ham-handed; I’s suspect the local BOM was acting in their own self interest and “for effect” (unless it can be shown the data they sent up was clean and it was the GHCN process that changed it, then we know it is inside NOAA/NCDC/NCEI.

    So when I get time to go back and look at this in more detail for particular countries, I think it will start to show patterns that match different causal agents. IF it is disbursed over the whole set, it’s at NCEI. When it is just a few DinkyLands, it’s local to them. Etc. The confounder here is “Standards”. So when a new, improved pre-processing method (QA or homgenizing or…) is presented at a conference, it will go global despite being implemented by each locale individually. To detect that will require a literature search of “what was new” in each data set version.

    FWIW, I find it telling that we can spend, what was it, $25 Million on a Presidential Political Witch Hunt but nearly nothing on a Special Investigator for Climate Data Fraud… Give me a $25 Million budget, and a nice office in D.C. and subpoena power, and I would have this done in a year. Tops.

    Oh Well. Not going to happen. People pay for what they want, not for the truth. Very few people really want truth. (That’s why “How do I look?” is rarely answered with truth… At about 12 years old I tried constant truth and honesty for a while. The “experiment” was short lived… )

  10. Bill in Oz says:

    E M I just read all this. My only thought was : “Holy Sh*t”..It ain’t simple; it’s complicated !
    But surely Trump would enjoy providing $25 million to sort this out.

  11. mlr77062 says:

    E. M. I have examined the climate stations data for my home state of Iowa. Twenty three of them. I did this some ten years ago with v2 data. I downloaded both monthly and daily data. I mostly worked with daily data. As a retired engineer, I spent quite a bit of time looking at the raw daily data. I noticed that the most frequent error was reversing Tmax and Tmin data. Unfortunately at the time I did not keep statistics but frequently it would be in the 1-2 % area, enough to screw up monthly Tmax and Tmin averages. Also there were occasional key punch errors that I observed manually and confirmed by referring to the monthly station logs. I suspect the daily data remain mostly uncorrupted just because it is so much effort to “fudge” it.

    One thing I did in an Open Office spreadsheet was perform linear regressions for each day of the year and use the results to tabulate the extrapolations by decade to 2100. For the most part, the regressions had minimal R2 and reflected small cooling and warming trends, often overall cooling.

    If you are up to the effort, a QA program on the daily data could reveal a lot.

  12. wyzelli says:

    Just to in with my compliments on a lot of of detailed and comprehensive work.

  13. Pingback: W.O.O.D. – 24 May 2019 | Musings from the Chiefio

  14. E.M.Smith says:

    There is a point to emphasize. I put it in the description in the posting, but I think it may need some added emphasis.

    This posting is for the purpose of emphasizing where there can be or might be “issues”, it is not for the purpose of finding an actual “global climate” nor for the purpose of identifying EXACT changes in the data. It deliberately uses averages that are not the same for the two data sets.

    Since GHCN v4 is a longer data set than v3.3 so by using “all the data” for any given instrument to make the average, I’ve included end point “hotter” data in v4 than exists in v3.3 to the extent that v4 end data are, in fact, hotter.

    That of necessity would mean that “the past was cooled” but via the inclusion of the 2016, 2017 and 2018 data points in v4 IF THEY ARE WARMER.

    Now you can say that’s “lying with statistics” in that the actual data points from 1900 to 2000 might all be identical and the “cooling the past” is only an artifact of “cooking the present”. Does it really matter? Not if your purpose is to find out “how cooked are the data”, but yes if your purpose is to find “what data items are suspect”.

    So there’s another step I need to do before “pushing the idea” that “the past was cooled”. That is to pick a couple of countries where there’s a “cooling of the past”, and compute the v4 anomalies using ONLY the data up to and including 2015. Then make a comparison graph.

    At that point one of two things are likely to happen.

    a) More of “the past” ends up near zero change (those displaced flat lines at the start of some graphs) because the data items are in fact identical between the two data sets AND the more recent data points are going to be even higher (so those that are +2 and +3 C will be even “hottest evah!!!”)

    b) We have graphs that look more or less just like these and it points out that the data used ARE different, either from changes to data items or via the particular instruments in the set (and proving that just using “anomaly processing” does not eliminate impact from instrument change).

    In essence, this present approach is a forensic approach to make it obvious IF there is something to dig into, and point at WHERE to look a bit more. It is NOT for finding WHAT was done. The next step is to look more at the WHAT. So my phrasing of “cool the past” is something of a statement about WHAT. which makes it ‘less than scientific”. Because after doing the added comparison described above it may turn out that the WHAT is excessive (really excessive) hot data points in 2016, 2017 & 2018 and the “cooling the past” is more artifact than reality.

    In the GHCN (at least the older versions) there are many more records that end “in the past” than run all the way to the present. For that reason, most of the comparisons (done the way I did them) will be “apples to apples” as both thermometer data records will end in 2015 or before. Using GHCN v3 as an example, of the roughly 7600 instruments in the inventory only about 1200 survived to the end of the data set. That means at least 84% of the records end before 2015.

    Then in v4, there are a LOT more records (something like 27000?) so most of the records will not have a matching instrument in v3 data. So their “current data” from 2016, 2017 & 2018 can not cause a “warming of the past” via their average biasing that particular instrument record vs that same instrument’s v3 record.

    So between those two effects the vast bulk of the records will not suffer from this “cooling the past via a few hot data items in the recent v4 data” potential artifact.

    But in a small country, like Lithuania, it is more likely that they only have a couple of instrument records and that they match between v3 and v4, so the effect might be significant.

    And that’s part of why I’m saying it “needs more polish” before touting it at WUWT. I need to address that point (as it is an obvious “place to attack”) and measure the degree of the effect in some smaller places vs larger. Then also demonstrate that the only real effect (IF it is an issues) is to make the “way too hot now” look even more crazy..

    You can guarantee that someone will make a statistical attack and say something like 9x.xx% of the specific data items in v3.3 are identical to their values in v4; and then you end up in a “war of the statistics” that just sucks down a lot of time and nobody wins as almost nobody really understands statistics uses and limits. AND despite my pointing out this is a forensic exploration with known limits, I’ll be accused of “lying about changes”. I’d rather address that and bury it before the food fight…

    One of my major goals with this work was to demonstrate that you can NOT just wave around “used anomalies” and have issues of instrument change banished. With that goal in mind, having “used anomalies” if then the added data in v4 causes a “cooling of the past”, it DOES demonstrate that just using anomalies is not enough. That’s pointing at WHERE there are issues and that helps to emphasize that those “cooler baseline periods” found can very well be a problem even with “anomaly processing”.

    With that in mind, you can see that comparing v3.3 to v4 while using only v4 data up to the 2015 end point data of v3.3 in the v4 anomalies so they become more comparable to the v3.3 is a necessary “sanitation measure” to address the clear attack point from those who do not care that my goal was to show “anomaly processing” still can ‘have issues’ (and not to make a peer reviewed Climate Science analysis of individual data item changes).

    Yeah, I know, all this is way down in the weeds… but that is where the cockroaches live…

  15. p.g.sharrow says:

    @EMSmith your efforts to tease out the clues is beginning to yield fruit now that you can begin to manipulate the records. Now that you have blazed out the trail to how it can be done by anyone, perhaps others can add to the examination.of how this sausage was made.

    It appears to me that your impression that much of this was made by fellow travelers rather then a central figure. They all went to the same schools and studied the same books and justified the same conclusions and errors in judgment. It appears that only a few resisted the push to go along, After all, Everyone does it! so it must be the correct thing to do. If you want to do well in your field you can’t make waves.

    I am more in awe of the path you have followed rather then the destination that you have arrived at. A few hundred dollars and your abilities has created tools that others with millions at their disposal could not accomplish. HT for sure…pg

  16. E.M.Smith says:

    I just realized I put that prior comment on the Africa thread instead of the Europe thread.
    https://chiefio.wordpress.com/2019/06/03/ghcn-v3-3-vs-v4-anomaly-graphs-europe/
    Oh Well ;-)

    @P.G.:

    It is a personal quirk of mine. Partly from my Amish roots (never waste anything). Partly from British Monty Python kind of giggle sense (can I do this in such a way the other guy looks silly?). Partly from the instruction my folks gave me about life in The Great Depression (how do you make a life work with nearly nothing as resources?). Partly just a love of the Minimalist Game.

    So I’m ALWAYS trying to find the absolute minimum resources to do a given thing. The old saw about “I’ve been told to do more with less for so long I can now do anything with nothing!” isn’t just a joke to me, it’s a goal.

    So I bought an Odroid N2. Found out it “had issues” with the matplotlib, and went back to the Pi M3 HOWEVER: I also measured the performance level, realized the Pi M3 was more than fast enough for the graphing, and rather enjoyed doing all the graphs on a Raspberry Pi instead of Big Iron.

    I loved my time running a Supercomputer Center and effectively having a PC, Personal Cray. ;-) But what really makes me light up is stuff like turning a 4 hour data load into a 10 minute one by finding a “new technique”…

    As to why “Climate Science” is the way it is: IMHO they removed the competition element and inserted a “consensus” one. Publish OUR WAY or Perish. So I don’t give a tinkers damn about their status system, their peer review, their awards, accolades, self-abuse honors, or approval. I only care about learning what is real and true. If that makes me sand in their gears, then it is the diamond sand of real truth… ALL I do is search for understanding and truth. I want to die with the biggest hoard of understanding and universal truths I can scrape together. I have no time for the BS of “go along to get along” or the crap science of “that will never be published unless you toe the line and kiss the ring”

    So I am what I am.

    And if I can “make it go” on a $15 Orange Pi One, instead of an expensive Raspberry Pi M3 at $35, well, all the better… Maybe that will show some “little black kid” in Africa with money from collecting Pop Bottles that his dream of Doing Science is something he can start now, and doesn’t need to be left on the trash heap of poverty in a 3rd world backwater. ( I grew up in a 3,000 person rural backwater dreaming of being a Scientist, so this is from what I lived… my first science was done funded by Pop Bottles and gunny sacks of walnuts collected around town.)

    FWIW, at some point I am going to test these codes on the Orange Pi One. The only reason it might have an issue is the 512 MB of memory. It will swap a lot, but likely will be fast enough… I was going to try it on the Pentium 4 but I don’t have enough free disk space ;-)

    It is sort of like Haiku, but with “minimum syllables” as the rule, not 17… Or “Guess That Tune!…” Where you say “I bet I can make that database work on a Pi M2 with 512 MB of memory!!!” and then try to prove it… It can be a bit addicting ;-)

    Well, FWIW, the last bit to do on this series is the “top post” that ties it all up in a nice bow and has full instructions and codes for anyone who wants to replicate it / modify it / play with it. I think I’ll include a parts list and suggested prices too ;-) All up, I think I can make the minimal system run on about $75 of gear. $50 might be a “bridge too far”, but I’ll test it ;-)

  17. Bill in Oz says:

    E M , You are going to dig in Lithuania. Good ! But I think that the data for Malta, Iceland, Gibraltar & Guensey Island might also be worth digging into. Each of these places is small and probably only has one long term weather station….

  18. E.M.Smith says:

    Just a note that I did the work to compare v3.3 with v4 with identical “baseilne” length of all data up to 2015 inclusive. There’s a tiny change in the graphs but almost imperceptible
    . https://chiefio.wordpress.com/2019/06/09/ghcn-v3-3-vs-v4-baseline-end-2015/

    So my concerns here:
    https://chiefio.wordpress.com/2019/05/22/ghcn-v3-3-vs-v4-anomaly-graphs-africa/#comment-113019
    above are a nice hypothetical, but it doesn’t change the graphs enough to change any conclusions. The last couple of hot years in v4 are NOT the reason for the apparent “cooling of the past”… It is still there if you match both sets to “all data up through 2015” for the anomaly creation.

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

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.