Here are some more statistical fiddling with the global temps.

Interestingly the fiddling ALWAYS seems to INFLATE the amount of global warming.

Coincidence?

Here is the data for station

http://data.giss.nasa.gov/cgi-bin/gistemp/gistemp_station.py?id=222241430002&data_set=1&num_neighbors=1

In this case its quite clear that the choice of what years to use as the base
temps makes the early 21st century appear warmer than normal.

Why?

Well if you take the first ten complete years of the station record (1938-1947) as the base period, then the anomaly for the last four complete years on record (2004-6 & 2008) drops from +1.06 C to a much lower +0.18 C.

Clearly warming over a 50 year period of 0.18 is nothing to get worried about, hence the fiddling with the base period to accentuate the apparent warming.

But that’s not all.

The other fiddling is even more problematic.

Notice I mentioned COMPLETE years.

Turns out 2003 & 2007 weren’t complete.
Each was missing one month of data (999 in the record)

What is problematic is GISS still computes the ANNUAL record for a station with missing months.

So, what did GISS use for the missing month?

In 2003, March was missing.

GISS substituted a VERY WARM value of -22.3 for the month.

But, here are the March values on the years before and after 2003

-26.3
999.9
-24.6
-25.5
999.9
999.9 <==2003
-25.9
-23.5
-26

Or an average March temp of -25.3

Significantly there are NO months anywhere near as warm as the -22.3 value used by GISS

Using GISS’s VERY WARM but fictional temperature for March inflates the ANNUAL FIGURE (the only one that really matters) by a SIGNIFICANT +0.25 C.

Then there is 2007, which is missing December.

Once again GISS substituted a VERY WARM December value of -30.5

Here are the actual Decembers before and after 2007 though:

-31.5
-42.4
-32.7
-30.5
999.9 <== 2007
-38.6

The average for these five years is -35.1

But, using GISS’s VERY WARM but again fictional temperature for December
inflates the ANNUAL FIGURE by a SIGNIFICANT +0.38 C.

This is Global Warming GISS style.

Take Kasalinsk here

In the middle of the wierd global hot spot.

Its anomaly this month was a +4.81 C, reporting a average March Temp of 3.5 C when the GISS 1951-1980 base period averaged -1.31 C

A look at the station data quickly shows that the Anomaly in this area is
accentuated in the Jan-Feb-Mar time frame.

But, if you compare the temps AFTER the period where the station was offline (~1990 – ~1996) to the decade just BEFORE it went offline (~1980 – 1990) what you find is the average Anomaly for Jan-Feb-March has gone up a more reasonable amount of +0.44 C.

But

After the step change these same months report an average anomaly of +4.60 C.

Clearly this is not due to CLIMATIC CHANGES.

Clearly there is a LOCAL issue.

To include this faulty data in the Global Average is clearly bad science.

Of course it just might be that Hansen spends so much time running around
drumming up protests to support his beliefs that he has no time to attend to the actual science.

Arthur

The numbers will never stop changing. It is a fundamental part of how GIStemp works. The most recent few years are used to re-write past years as some kind of fictional “correction”. Since each year we have a new year, the “correction” value changes… (And if you are wondering why an equipment change in the last 6 years ought to rewrite temperatures in 1919, well, so am I…)

a_set=1&num_neighbors=1

Arthur

See here to see a graph of the temps:

http://data.giss.nasa.gov/cgi-bin/gistemp/do_nmap.py?year_last=2009&month_last=0\
3&sat=4&sst=0&type=anoms&mean_gen=03&year1=2009&year2=2009&base1=1951&base2=1980\
&radius=1200&pol=reg

Note that 4-8 degree C HOT SPOT that affects an area almost 20 times its size because of the extreme amount of warming compared to the global average.

BUT

Then go to the station data in this area:

See anything STRANGE?

http://data.giss.nasa.gov/cgi-bin/gistemp/gistemp_station.py?id=211358490000&dat\
a_set=1&num_neighbors=1

http://data.giss.nasa.gov/cgi-bin/gistemp/gistemp_station.py?id=231386960000&dat\
a_set=1&num_neighbors=1

http://data.giss.nasa.gov/cgi-bin/gistemp/gistemp_station.py?id=211381980000&dat\
a_set=1&num_neighbors=1

http://data.giss.nasa.gov/cgi-bin/gistemp/gistemp_station.py?id=208407450002&dat\
a_set=1&num_neighbors=1

http://data.giss.nasa.gov/cgi-bin/gistemp/gistemp_station.py?id=231389270000&dat\
a_set=1&num_neighbors=1

http://data.giss.nasa.gov/cgi-bin/gistemp/gistemp_station.py?id=231386180000&dat\
a_set=1&num_neighbors=1

GISS is BS

Arthur

Mark

Mark

“I am not sure what .01F equals in C, but it is more than .001 C for certain.”

Right. The conversion isn’t linear though, it depends on the specific temperature.

For example:

.01° C = 32.018° F

1° C = 33.8° F

15° C = 59° F

20° C = 68° F

25° C = 77° F

Larry

Thanks Larry. The person who answered my question is Anthony Watts. I specifically asked him because he is a meteorologist and sells weather instruments from this site.

Anthony’s answer tells me that the graph is nonsense because there is no way they have the resolution from the measurements to know that the temperature has increased from .018 C to .020 C. They cant do better than watching it move in .01 F increments (I am not sure what .01F equals in C, but it is more than .001 C for certain.)

I agree with you completely Larry, that the calibrated resolution is not going to be maintained in the site environment and Anthony has done a remarkable job documenting just how poor are most of the sites.

If people like Hanson and Gore weren’t so able to do us serious economic damage by getting leaders to seriously believe them and act on what they say, we could all just move along because when it comes to AGW, now called “climate change”, there really is nothing to see here.

woodNfish (09:55:17) :

Anthony,

Okay, I’m looking at the chart with Celsius divided into thousands and what I want to know, since you are a meteorologist AND you sell instruments, is just how accurate are the instruments?

My reason for asking is in another blog a while back a fellow who claimed to calibrate temperature instruments said they were only accurate to within +/- 3 degrees.

What is the truth?

REPLY: The mercury thermometers used in Stevenson Screens are accurate to 0.1 degree F in the USA, but the observer rounds to the nearest whole digit, thus 57.6 degrees becomes 58 degrees when recorded in the B91 logbook sent to the National Climatic Data Center. Ditto for the new MMTS electronic instruments. – Anthony

The other issue is that there are several elements to accuracy in measurements.

Take for example precision measurement of machine components.

You have a micrometer that is certified to be able to measure to 0.0001 inches. A machinist takes that tool and measures a bearing journal and announces with authority that the journal is x.xxxx inches in diameter.

Implicit in that measurement is :

The physical precision of the micrometer — was it physically marked accurately so that when it reads a measurement of 1.0000 inches the ends of the micrometer spindle and anvil (the measuring surfaces) are actually 1.0000 inches apart?

Then you have observational issues — is the machinist actually reading the markings correctly? Was he standing on his head in a poorly lit location holding the micrometer in an awkward position when he took the measurement? Did he accidentally bump the spindle as he pulled the micrometer out to look at it and read the markings? Did he actually “see” what the indicator marks were displaying or make a mental error reading the markings incorrectly?

Then you also have measurement conditions to consider. Machine part precision measurements are temperature dependent, both the temperature of the part being measured, and the temperature of the measurement device effect the accuracy. In quality control environments they do precision measurements in a climate controlled room so parts do not “grow” or “shrink” due to thermal expansion.

Add to those elements you have the physical care with which the measurement was made. Did the machinist wipe any dust or grit off the part being measured and checked to be sure the anvil and spindle of the micrometer were clean before he took the measurement? Did he hold the micrometer square and perpendicular with the surfaces being measured? Did he use excessive force tightening the micrometer and spring its frame slightly so that it indicated a measurement smaller than the actual part size?

The same sort of considerations apply to temperature measurements.

Is the thermometer accurately marked so that when its body is at x temperature it reads x temperature?
Did the observer, actually correctly read the temperature indicated on the thermometer?
Did the observer, make an error of procedure by exposing the thermometer to direct sunlight during observation?
Is the enclosure properly sited so that the temperature of the thermometer body is actually representative of the local atmospheric temperature?
Does this particular observer have a conscious or unconscious bias to their observations? Do they tend to round up more often than round down when the measurement is near xx.05 degrees?

In short measurements in the field are never as accurate as the certified accuracy of the measuring instrument. The above effects always degrade the precision, but the precision can never be higher than the lowest precision element in the process.

If the physical precision of the thermometer is accuracy to 0.01 degree.
The ability of the observer to read the actual indication might be accurate to 1/2 a division (if the observer is careful). The precision of the enclosure due to sighting might have a bias of + or – anywhere from 0.5 to 2.5 deg compared to the actual air temperature due to siting issues.

Maybe one of the math wizards can weigh in here but my gut reaction is that for in the case of the Stephenson screen enclosures based on the survey info so far gathered we should not consider temperature measurements using manual measurements methods to be accurate to more than perhaps 2 degrees from actual temperature, perhaps 3-4 degrees in the poorly sited locations.

Even digital systems and satellite systems will have these sorts of error build up from multiple possible system errors.

I find it absurd that they presume to measure temperatures to the accuracy that they do. Even with certified measurement equipment in carefully controlled environments and cautious careful observers errors still crop up. To assume they can “calculate” changes that are a small fraction of the precision of the measurements is silly.

In answer to the original posting, that person might be quoting you the “real world” precision for typical users, and not the physical limits of the thermometers themselves.

Larry