On Apr 12, 1:49 am, Bill Ward wrote:
On Fri, 11 Apr 2008 12:55:40 -0700, John M. wrote:
On Apr 11, 9:15 pm, Bill Ward wrote:
On Thu, 10 Apr 2008 11:14:20 -0700, John M. wrote:
On Apr 10, 1:43 pm, "Paul E. Lehmann" wrote:
Bill Ward wrote:
On Wed, 09 Apr 2008 13:12:25 -0700, Roger Coppock wrote:

On Apr 9, 8:34 am, Bill Ward
wrote:
On Wed, 09 Apr 2008 07:04:24 -0700, matt_sykes wrote:
On 9 Apr, 10:24, Roger Coppock
wrote:
On Apr 8, 8:01 pm, Poetic Justice
-n-

Dog.com wrote:
Roger Coppock wrote:
March ties for 3rd warmest on NASA's
129-year record.

Why is NASA the official keeper of the temperature?

NASA Goddard Institute for Space Studies offers data, as do
several other
organizations. I use NASA's data because GISS corrects for
UHI using nighttime earth shine, artificial lighting, measured
from satellites. IMHO, this method is better than using census
data to locate urban areas.

That is a feeble way to adjust for UHI.

The ONLY way to adjust for UHI is to put a station in a rural
area near to the urban station to act as a control (being a
sicnetist you wold know this of course).

Mind you, one you have done that you might as well ignore the
urban station data since rural data is true surface temp.

And what happens when you do that? You get no warming trend.
RURAL STATIONS ACROSS THE GLOBE SHOW NO OVERALL TEMPERATURE
TREND.

Roger can't comprehend that bad data is worse than no data.
Unless you're trying to scare people, of course.

If you have better data, or a method for UHI correction, you are
more than
welcome to present them here. Until then the data presented above
are a better indication of reality than your fantasies.

Fantasies aren't science, whether they're mine or NASA's. That's
the problem with trying to "correct" bad data. If it's bad, it
can't be
used - it's a fantasy based on invalid assumptions. Averaging bad
data with with good data hides the problem, but doesn't fix it.

It is even worse than that. I suspect there is an analogy with wine
making. I knew of a winemaker who had a small amount of wine made
from under ripe grapes. He blended it (10%) with some good wine
90%). That small amount ruined the whole lot. A "Little BAD Goes a
LONG ways". I suspect the same is true of data. A little bad or
incorrect can have effects that are way beyond suspected results.

Such are the musings of a statistical illiterate. In actual practice,
the way to avoid problems from extraneous or erroneous data (ie. bad)
is to swamp it by using large sample sizes. If you measure the height
of 100 men taken at random, the mean height will be virtually
unchanged should one or two of them be giants or dwarves.

This is what happens in practice with temperature data. Very large
sample size ensures that the famous UHIs will have little effect on
the calculated global means.

Only if you can prove the assumption that all errors are symmetrically
distributed. Otherwise it's a fantasy. Or a hoax.

Ah. Another statistical illiterate crawls from the woodwork, to give us
his take on something he knows little or nothing about.

The manner in which the residuals are distributed is immaterial. It is
merely required that their distribution remains the same in repeat
samples.

So it doesn't matter to you if there are more errors showing increasing
warming than showing cooling? I think I see your problem. You are
reading "residuals", where I wrote "errors". Don't you know the
difference?

Generally speaking, statistical errors can never be quantified. I
simply corrected your sloppy language, as it was clear to what you
referred.