Fig. 2 Global sea level from tide gauges (red) and satellite altimeter data (blue,
with linear trend line).
This shows the HadCRUT3 temperature record since 1850
with linear trend lines, which all end in 2005, but which begin at different years, covering time periods of 150,100, 50 and 25 years.
Not exact matches
Kowatsch and Kämpfe have plotted the January data over the past 31 years along
with the computed
linear trend line:
(C) Mean of all records transformed to summer temperature anomaly relative to the 1961 — 1990 reference period,
with first - order
linear trend for all records through 1900 (green
line), the 400 - year - long Arctic - wide temperature index of Overpeck et al. (2)(blue curve; 10 - year means), and the 10 - year - mean Arctic temperature through 2008 (red
line).
We found that the observed temperature evolution since 1880 is only very poorly characterized by a
linear trend, so we used a non-
linear trend line (see Fig. 1 above) together
with Monte Carlo simulations.
Either we have a dynamical system as you claim in your paper, or the system is
linear as you infer
with your straight
trend line.
Clearly, the sea ice volume data plot is the single most important topic of discussion, yet in the article it is shown in Figure 1
with a poor vertical scale and amongst
linear trend lines which mislead and make the curve appear to be
linear and reach the zero point far out in the future.
(c) The global mean (80 ° N to 80 ° S) radiative signature of upper - tropospheric moistening is given by monthly time series of combinations of satellite brightness temperature anomalies (°C), relative to the period 1982 to 2004,
with the dashed
line showing the
linear trend of the key brightness temperature in °C per decade.
Since we mainly use it here to estimate post-1990
linear trends, any other filter
with this capability would have done just as well - or indeed simple
linear trend lines.
You could also question why to fit
linear model but since climatology seems obsessed
with «
trends» I though this was better than a single straight
line.
For this article, a statistically - significant global warming means that the
linear trend (slope of the
trend line) is likely greater than zero
with 95 % statistical confidence (i.e. the 95 % error bars do not include a possible 0.0 or negative temperature degree slope).
One of the «problems»
with the way climate data are handled is in the obsession
with applying
linear trend lines to non-
linear data.
You need to look at tools for identifying a periodic (or quasiperiodic) signal on top of a base
trend that is NOT
linear; because there's a heck of a lot more going on
with climate that you can capture on such scales
with one
line a sine wave.
A
linear regression
trend line with normal iid error gets to the heart of what Mandelbrot was questioning.
If I limited the test to forecasting the
trend only and use the
linear trend from 1969 — 1988 (0.15 per decade) as the null hypothesis to compare
with Hansen's
trend the skill of Hansen is — 1.76 i.e. just sticking a
line through the last 20 years is much more skillful.
We see the last half century,
with 95 % lower and upper bounds and
linear trend lines through a dual pass prime 13/11 filter (which should minimize distortions), and for comparison both Mauna Loa and the last
line, «plot / best - lower / from: 1960 /
trend / detrend: 0.9» in dark blue, far below the real global temperature curve, maintaining the slope Girma claims is the actual temperature
trend «unchanged in 160 years».