On Dec 16, 2:57 pm, Peter Franks wrote:
Roger Coppock wrote:
November was 5th warmest on NASA's 128-year Northern Hemisphere
record.
...
Rank of the months of November
Year Temp C Anomaly Z score
2005 15.15 1.142 2.62
2004 15.06 1.052 2.41
...
2006 15.03 1.022 2.34
2007 14.96 0.952 2.18 --

An alternative way to look at the data:

We are in the fourth year of a November cooling trend.

You were given 128 years of data, and
you throw away 124 years and look at
only 4. In statistical term for this
is "bias."


Let's analyze this biased world of yøur's
attach(aframe)
aframe
Year TempC
1 2005 15.15
2 2004 15.06
3 2006 15.03
4 2007 14.96

fitted.model - lm(TempC ~ Year)
summary(fitted.model)

Call:
lm(formula = TempC ~ Year)

Residuals:
1 2 3 4
0.079 -0.053 0.001 -0.027

Coefficients:
Estimate Std. Error t value Pr(|t|)
(Intercept) 99.281000 62.717994 1.58297 0.25426
Year -0.042000 0.031273 -1.34301 0.31138

Residual standard error: 0.0699285 on 2 degrees of freedom
Multiple R-Squared: 0.474194, Adjusted R-squared: 0.21129
F-statistic: 1.80368 on 1 and 2 DF, p-value: 0.311383

You're only 31% sure that this trendline explains 21%
of the variance. GIVE IS A BREAK, and forget your
imaginary cooling trends, OK Peter!

Now let's restore all 128 years of data, and sanity.

=-=-=-=-=-=

November was 5th warmest on NASA's 128-year Northern Hemisphere
record.
Basic greenhouse gas physics from MIT.
http://www-paoc.mit.edu/labweb/notes/chap2.pdf

Global mean surface temperatures continue to rise.

These hemispherically averaged temperature data come from NASA:
http://data.giss.nasa.gov/gistemp/tabledata/NH.Ts.txt
They represent the results of tens of millions of readings
taken at thousands of stations covering the lands of the Northern
Hemisphere over the last 128 years. Yes, the data are corrected
for the urban heat island effect.

The Mean November temperature over the last 128 years is 14.008 C.
The Variance is 0.18994.
The Standard Deviation, or SIGMA, is 0.4358.

Rxy 0.69647 Rxy^2 0.4851
TEMP = 13.478271 + (0.008215 * (YEAR-1879))
Degrees of Freedom = 126 F = 118.690192
Confidence of nonzero correlation = approximately
0.9999999999999999999 (19 nines)
The month of Nov in the year 2007,
is linearly projected to be 14.530,
yet it was 14.96. - 1 SIGMA above projected.
The sum of the residuals is 33.95314

Exponential least squares fit:
TEMP = 13.482701 * e^(.0005852 * (YEAR-1879))
The sum of the residuals is 33.96516 - Now larger than linear.

Rank of the months of November
Year Temp C Anomaly Z score
2005 15.15 1.142 2.62
2004 15.06 1.052 2.41
2001 15.06 1.052 2.41
2006 15.03 1.022 2.34
2007 14.96 0.952 2.18 --
2002 14.89 0.882 2.02
2003 14.84 0.832 1.91
1999 14.69 0.682 1.56
1998 14.69 0.682 1.56
1995 14.66 0.652 1.50
1997 14.62 0.612 1.40
1990 14.61 0.602 1.38
1983 14.58 0.572 1.31
MEAN 14.008 0.000 0.00
1882 13.42 -0.588 -1.35
1907 13.40 -0.608 -1.40
1898 13.40 -0.608 -1.40
1902 13.36 -0.648 -1.49
1894 13.36 -0.648 -1.49
1889 13.35 -0.658 -1.51
1890 13.34 -0.668 -1.53
1887 13.34 -0.668 -1.53
1884 13.34 -0.668 -1.53
1883 13.33 -0.678 -1.56
1908 13.26 -0.748 -1.72
1892 13.24 -0.768 -1.76
1891 13.24 -0.768 -1.76
1880 13.06 -0.948 -2.18

The most recent 168 continuous months, or 14 years and 0 months,
on this NH.Ts.txt data set are all above the 1951-1980
data set norm of 14 C.
There are 1535 months of data on this data set:
-- 783 of them are at or above the norm.
-- 752 of them are below the norm.
This run of 168 months above the norm is the result of a warming
world. It is too large to occur by chance at any reasonable level
of confidence. A major volcano eruption, thermonuclear war, or
meteor impact could stop this warming trend for a couple of years,
otherwise expect it to continue.