Let be the empirical variance estimated using tsbootstrap for
the GISTEMP series on its common support with HadCRUT4.
You can take out data points you don't like, you can apply whatever correction factors you want (such as the one that Nasa's
GISTEMP series uses to compensate for the dearth of measuring stations across the Arctic), and you can therefore end up with a temperature curve that might look a little different: but don't say it can't be done, because it can.
In the US,
the GISTEMP series comes via the NASA Goddard Institute for Space Sciences (GISS), while the National Oceanic and Atmospheric Administration (NOAA) creates the MLOST record.
Not exact matches
2011 was the 9th warmest year in the
GISTEMP global temperature
series.
A linear extrapolation over 1880 - 1980 of the
GISTEMP meteorological station
series gives a temperature anomaly of 0.2 K in 2012, much lower than the lowest curve drawn.
As it is the
GISTEMP time
series looks equivalent to a model scenario.
The
series is only 42 data points long (equating to 126 years), so is hardly robust, however, it may be a useful predictor of future temps since it is
gistemp that lags the SST.
A linear extrapolation over 1880 - 1980 of the
GISTEMP meteorological station
series gives a temperature anomaly of 0.2 K in 2012, much lower than the lowest curve drawn.
Coverage bias estimates are shown for both HadCRUT versions using the
GISTEMP land - ocean
series and the UAH
series to provide the temperature maps.
(Note: For strict validity the anomaly baseline period of the
GISTEMP map
series was first adjusted to match the CRU data.)
Figure 5: Various best estimate global temperature climate model predictions evaluated in the «Lessons from Past Climate Predictions»
series vs.
GISTEMP (red).
FWIW, having been through the
GIStemp code, I would not call the output from
GIStemp a «temperature
series».
We can see that the long term trend shows a similar steady rise in all four
series since about 1980, albeit slightly steeper in the two interpolated
series, NASA
GISTEMP and Cowtan and Way.
The main point however is that those two
series (Cowtan and Way and NASA
GISTEMP) continue to run warmer than the two non-interpolated
series when evaluated over the full 20 - year period.
The value of the variance for the process noise in the above was arbitrarily chosen to be the same as the empirically estimated observational variance of the observations in the separate cases of the HadCRUT4
series and
GISTEMP.
For the HadCRUT4 and
GISTEMP temperature anomaly
series, was estimated using a Politis - Romano stationary bootstrap (because the
series data are interdependent) giving.
In particular, the model was changed to have a bivariate
series, the first component being from HadCRUT4, the second from
GISTEMP.
Let be the empirical variance estimated using tsbootstrap for the HadCRUT4
series on its common support with
GISTEMP.
It is not clear whether the time
series such as
GISTEMP actually correct for this.
In fact, were I doing a «
GIStemp like» temperature
series, I'd do it with Highs and Lows kept through the whole thing.
There are three main global land / ocean surface temperature
series, produced by NOAA's National Climate Data Center (NCDC), NASA's Goddard Institute for Space Studies (
GISTemp), and the UK's Hadley Center (HadCRUT).
The major land
series used are CRUTEM4 (the land component of HadCRUT4), NCDC,
GISTemp, and Berkeley Earth.
The
GISTEMP monthly global temperature
series was used for all temperature data.
Being already out of date (and hard to update) is a drawback IF the point was to create an ongoing time
series, an alternative to
GISTEMP, HadCRU and NCDC.
For example, I doubt that worldwide, the monthly
GISTemp, HadCRUT or other
series will be wildly anomalous as the great heat in the US / Central Canada is well balanced by cool areas elsewhere — Newfoundland where I live being one of them: — LRB -.
In the
GISTEMP index, the tables of zonal, global, hemispheric means are computed by combining the 100 subbox
series for each box of the equal area grid, then combining those to get 8 zonal mean
series, finally from those we get the Northern (23.6 - 90ºN), Southern and tropical means, always using the same method.
The
GISTEMP analysis recalculates consistent temperature anomaly
series from 1880 to the present for a regularly spaced array of virtual stations covering the whole globe.