This post was prompted by articles by Cliff Mass (Was Global Warming The Cause of the Great Northwest Heatwave? Science Says No) and Michael Kile (How Warmists Package Panic ). Perhaps a comparison of the above two maps should be included in that discussion.

To the activists who preach the dogma of Climate Change, its great virtue is that it precludes further explanation such as volcanic heating. as suggested by the above maps.

It is Theory of Everything. It accounts for the rabbit population explosion on Macquarie Island, and coral bleaching on the Great Barrier Reef. There are always many and varied climate model forecasts on hand to justify any unusual event and much busy-work and kudos to be gained from inter-comparisons. At the end of the day, we, the lay public, are told “the Science says”.

**But it isn’t science.**

In order for it to be science, model outcomes needs to be tested against real world statistics and those models with improbable outcomes must be modified or rejected. That this is not done is not an oversight, it is policy. The IPCC Third Assessment Report specifically dismisses the need for statistical testing when it states: “our evaluation process is not as clear cut as a simple search for ‘falsification’” (Section 8.2.2 on page 474).

Science requires that theories be tested against observations. Until that happens they are just ideas, formulas or software that may or may not emulate reality. Statistical inference was devised by Fisher and others in the 1930s as a method of organising numerous observations to test theories. The fundamentals are outlined here.

A statistic is a numerical property of a set of numbers called a *sample. *Examples are the *mean* (average), the *variance and *the *standard deviation. * Two sets of numbers in one-to-one correspondence have a *correlation coefficient*. If they are also ordered in time they are called *time series* and more statistics can be defined known as *regression coefficient*s. Conceptually every statistic has a “real” value termed the population or ensemble value which is estimated from the sample. Confidence limits show the probable range of the population value.

Here is the abstract of my paper *Statistical Testing of Climate Models* submitted to *Statistical Science* on 26/8/21.

**Two useful climate statistics were derived from observations, the climate sensitivity and the CO2 concentration sensitivity. Both were estimated using an autoregressive (ARX) method without making assumptions about the underlying physics. An impulse response sequence was found as the convolutional inverse of the prediction error filter coefficients of the autoregressive process. Sensitivity was then calculated as the sum of terms of the impulse response sequence. These estimates provide sample statistics with which to test numerical climate models. Using the method, the maximum likelihood climate sensitivity estimated from observed global average temperatures and measured CO2 concentrations is 2.3 deg C with 95 percent confidence limits of 1.9 deg C and 2.9 deg C. The estimated impulse response sequence of atmospheric carbon dioxide concentration as a function of emissions is exponential with a half-time of 43 years. The most probable value of concentration sensitivity to emissions is 1.4 parts per million/(Gt/year) with 95 percent confidence limits of 1.05 and 4.2 ppm/(Gt/year). The widely accepted hypothesis, that 10 to 20 percent of carbon emissions remain in the atmosphere indefinitely, can be rejected.**

A pre-print can be downloaded here: Reid2021

Climate map source: https://cliffmass.blogspot.com/2021/07/was-global-warming-cause-of-great.html .

Volcano map source: https://www.usgs.gov/news/which-us-volcanoes-pose-a-threat .