Date: Wed Jan 18, 2006 6:14 am. By: Coby Beck
"LongmuirG" wrote in message
Coby Beck failed to live up to his reputation, writing about global temperature: This is a strange assumption. Why should the error band increase going back in time? Because the ink on the records fades?? Accurate thermometers have been around a long time, regular readings from hundreds of stations have been available since decades befor that record starts.
Think about normal scientific error bands. Start with what we mean by "global temperature" - you know, the temperature that is supposed to have increased by about 0.8 Celsius degree over the last 100 years? Is there a temperature gauge somewhere in the Sierra Club's palatial headquarters which reads the "global temperature" accurate to 0.01 Celsius?
I am not the right person to give you any detailed explanations of how this is done. I would like to repeat one of my questions to you that you snipped: why should anyone think that the fellows at GISS are less competent than you and unable to think of the things you can rattle off in a usenet post? Seriously, why would you yourself rely on your own speculations? If you had any serious concerns you would try to find out how it's done, which from the wild speculations below it is clear you have not. I don't think you understand statistics and how sampling is done and how many samples it takes for statistical conclusions. Again, I am not the right person to explain it to you.
I will take a few pot shots at you below where it is most obvious you are off base.
In my neighborhood, difference between daily low & high is often over 20 Celsius degrees. Difference between winter low & summer high is over 50 Celsius degrees -- and we are looking for year-on-year average global temperature differences measured in hundredths of a degree? Quite a challenge!
I don't believe that it is such a technological challenge to measure 1/100th of a degree celsius. But anyway, I think the high precision numbers you do sometimes see are results of the mathematics and as they come with ranges and error bars you are really just creating a strawman here. You are also getting confused by number of decimal places versus significant digits. If we have a measured change of .8oC over 100years this is .008oC per year, both figures are one significant digit. That does not mean anyone is claiming they measured 14.003oC one day and 14.011 the next.
To come up with a definitive annual average global temperature, we would first have to decide what we wanted to measure -- temperature at ground level? Or 1 meter above ground? Or 3 meters above ground?
You must also ensure that wind and rain and direct sunlight do not bias your readings. Yes, these things have been thought about and standards set. It is not however required that every station adhere to the same height standard as every other, only that they each be consistent. No one cares if the global average temperature at 2 metres is 14.50 or 14.51 the information of interest is the change over time. Consistency is the only requirement.
Then we would have to take many measurements and average them over time & location. Say we took a fairly coarse discretization -- one temperature reading per hour per square kilometer. Over a year, that would be roughly 8,800 readings at each of 510 million locations; 4.5 trillion temperature readings.
This is just ridiculous overkill. You know nothing about sampling data. Nor are hourly readings required for this analysis, just the daily high and low.
We would have to correct each reading for calibration, and then adjust them all to a common datum. (Temperature varies with elevation, of course, and it is about 9,200 meters from the lowest point to the highest point on the Earth's surface).
Again, irrelevant except where you can show a particular station has changed its altitude due to subsidence or geostatic rebound or whatever else might effect it.
Next we would need to develop an objective unbiased methodology to modify those adjusted readings for local effects -- some of those square kilometers are going to include things like volcanoes & steel mills. Then do the arithmetic and we have got our annual average global temperature for one year.
Why do you find this so daunting? But rejecting the ridiculous premise of one station every kilometer we can avoid volcanoes and most steel mills.
Of course, we don't actually measure 4.5 trillion real temperature readings per year, so all we can get is an approximation to the true annual average global temperature, incorporating many assumptions. How good an approximation? Great question! It would be nice to know, especially when we are trying to measure such tiny changes.
Why don't you read all about it here: http://data.giss.nasa.gov/gistemp/2005/
Follow the links, read the papers, ask the authors for clarifications. Or do you prefer FUD over facts?
Now, how do we come up with the global temperature for the year 1900? In 1900, horsepower meant exactly that. Sailing ships were still in use. No human being had even reached the South Pole, let alone measured its temperature. How do we confirm the calibration of old temperature records from 1900? Obviously, there are lots & lots & lots of assumptions in the global temperature for 1900.
You are again arguing from ignorance and assuming that the people who study these issues know less than you. Why would anyone believe that? Why would you believe that? There are indeed issues similar to what you describe. But they are problems that get solved not issues everyone obediently ignores so as not to anger their Greenpease masters.
If you are dealing with all manner of random errors it will not effect your trends, the real challenges are in identifying sources of bias (such as urban heat islands) and coming up with methods for correction. But these are not show stoppers, they are things that very smart people study and deal with.
It is fairly obvious that the uncertainty band for "annual average global temperature" gets wider as we go further back in time and have to rely on increasingly sparse, decreasingly reliable records. And if we take account of that increasing uncertainty, the alleged "global warming" trend may well disappear into the error bars.
Did you look at these? http://www.grida.no/climate/ipcc_tar/wg1/fig2-1.htm
The error range is way too small. Besides, that wishful thinking cuts both ways, why are you not warning us that the trend may actually be much larger?
Did you look at the borehole records? Less resolution but longer term and all modern-day direct measurements: http://www.ncdc.noaa.gov/paleo/globalwarming/pollack.html
Satellites, radio sondes, surface readings, borehole readings, ocean temperature measurements, world-wide glacial response, it all agrees. Aren't you tired yet? You must be because you ignored most of my post.
-- Coby Beck (remove #\Space "coby 101 @ bigpond . com")