Section 4.4: Conclusions
Hence
SD(t)SMR(t) + D(t)M
Z2
C(t) = aR(t) + bD(t)
where
a =
b =
71
(4.8)
(4.9)
(4.10)
Equation (4.8) shows that C(t) has an R(t) component and a component
determined by the spatial-mean of the D field. If, therefore, D(t) has a
trend, this will produce a trend in C(t) independent of the behaviour of
R(t). It may be, therefore, that C(t) teIls us little more than D(t), in which
case C(t) would be of little value as a fingerprint test statistic.
The reason C(t) (or, probably, any uncentered statistic) does not succeed
as a multivariate test statistic is because it does not separate the effect of the
changing mean (which, for D(t), is a strongly time-varying quantity) from
the true pattern similarity.
R(t) is now calculated using the observed data and the five sets of model
data discussed earlier. Pattern correlations were calculated using the gridpoint data (after excluding grid-points with missing observed data and the
corresponding model grid-points), defined by (4.1-4.4). The results for the
five models are shown in Figures 4.9 (annual), 4.10 (seasonal for the UKMO
model) and 4.11 (seasonal for the GFDL model). In none of the five models
is there any increase of R(t) with time. In other words, there appears to
be no increasing expression of any of the five model signals of the enhanced
greenhouse effect in the observational record.
4.5 Conclusions
Considerable care must be taken when dealing with any type of climatic
data. Several sources of potential error can easily impair climatic time series, rendering some analyses of doubtful value. The most serious, and often
ignored, source of error is due to the change, over the course of time, in the
environment around astation, particularly due to urbanization.
Using carfully checked temperature data from over 2000 land-based stations and about 50 million marine observations it has been shown that the
average "daily-mean" temperature of the world has risen by about 0.6°C over
the last 140 years. The warming has been uneven in both time and space.
Much of the warming occurred in the Northern Hemisphere before 1940 and
Hence
SD(t)SMR(t) + D(t)M
Z2
C(t) = aR(t) + bD(t)
where
a =
b =
71
(4.8)
(4.9)
(4.10)
Equation (4.8) shows that C(t) has an R(t) component and a component
determined by the spatial-mean of the D field. If, therefore, D(t) has a
trend, this will produce a trend in C(t) independent of the behaviour of
R(t). It may be, therefore, that C(t) teIls us little more than D(t), in which
case C(t) would be of little value as a fingerprint test statistic.
The reason C(t) (or, probably, any uncentered statistic) does not succeed
as a multivariate test statistic is because it does not separate the effect of the
changing mean (which, for D(t), is a strongly time-varying quantity) from
the true pattern similarity.
R(t) is now calculated using the observed data and the five sets of model
data discussed earlier. Pattern correlations were calculated using the gridpoint data (after excluding grid-points with missing observed data and the
corresponding model grid-points), defined by (4.1-4.4). The results for the
five models are shown in Figures 4.9 (annual), 4.10 (seasonal for the UKMO
model) and 4.11 (seasonal for the GFDL model). In none of the five models
is there any increase of R(t) with time. In other words, there appears to
be no increasing expression of any of the five model signals of the enhanced
greenhouse effect in the observational record.
4.5 Conclusions
Considerable care must be taken when dealing with any type of climatic
data. Several sources of potential error can easily impair climatic time series, rendering some analyses of doubtful value. The most serious, and often
ignored, source of error is due to the change, over the course of time, in the
environment around astation, particularly due to urbanization.
Using carfully checked temperature data from over 2000 land-based stations and about 50 million marine observations it has been shown that the
average "daily-mean" temperature of the world has risen by about 0.6°C over
the last 140 years. The warming has been uneven in both time and space.
Much of the warming occurred in the Northern Hemisphere before 1940 and