Total Pageviews

Monday, June 1, 2020

Why models go wrong -- add selection bias to the reasons

There are no real
climate change 
computer models.

A real
climate change
model must be 
based on a 
climate change
physics model.

That physics model
has to accurately
describe all 
the causes of 
climate change.

A precise physics 
model would be
the foundation 
for a real climate 
change model, 
that would at least 
have a chance 
of making accurate 
climate predictions.

Such a physics 
model does 
NOT yet exist, 
so the physics
of climate change 
has to be guessed.

Meanwhile,
the current 
"climate models" 
are based on 
personal opinions
about climate 
change physics 
from the modelers.

That's why the
current models 
are so wrong,
on average,
over-predicting
global warming 
by 2x to 3x.

The big surprises:
-- Modelers, 
on average,
continue to use 
an old CO2 versus
average temperature
"formula" from the 
1979 Charney Report.

-- Modelers ignore the 
fact that the 1979 formula
consistently over-predicts 
global warming.

-- The mass media ignores
the fact that the modelers 
consistently over-predict 
global warming, and

-- Many new models, 
for the next IPCC Report, 
are predicting even faster
global warming  -- 
which is very likely 
to increase the 
already large gap 
between predictions 
and reality !



Some day, 
maybe during
our lifetimes, 
an accurate
climate change
physics model 
may be developed 
to explain EXACTLY
what causes climate 
change.

That physics model
will NOT be claiming
the carbon dioxide level 
is the global average 
temperature
"control knob".

Reasons:
We already know
CO2 levels rose
between 1940 
and 2020, but 
that time span had 
two significant periods 
with no global warming, 
even while CO2 levels 
were increasing:

-- 1940 to 1975, and 
-- 2003 to mid-2015,

That's a total of 37 years,
out of the past 80 years, 
without the expected
global warming trend,

43 years of the 80 years 
did have a warming trend:
-- 1975 to 2003, and
-- Mid-2015 to 2020

Note: 
Smarmy government
bureaucrats have been 
slowly "adjusting" the 
1940 to 1975 period 
to show less global
cooling, and one 
compilation of the 
global average 
temperature no
longer shows any 
cooling !

Those 80 years
are also a very tiny 
percentage of Earth's 
4.5 billion year history,
to be used for jumping 
to any conclusions
about the effects 
of man made CO2 !


In addition, studies of 
Antarctica ice cores 
found that changes
of the global temperature,
in the past 500,000 years,
happened BEFORE changes
in CO2 levels -- not after !

Explanation:
Some variables
caused oceans 
to gradually warm, 
and that warming 
caused the oceans 
to release CO2 
into the atmosphere
hundreds of years
LATER.

A warmer ocean can not 
hold as much dissolved 
CO2 as a cooler ocean can.

Think of a cold carbonated
soda left outdoors on a
hot day -- it will will gradually
lose its carbonation,  as it
warms, and goes "flat".



The COVID-19 models 
grossly overstated 
the virus mortality rate.

Those models 
seemed to follow 
the tradition of 
climate computer
models.

Since the 1970s,
every climate model 
has exaggerated 
actual global warming,
except for one
Russian model.

The COVID-19 
models were 
even worse
than climate
models !



Why do models 
make such bad 
predictions ?

And why do the 
worst model predictions
get so much attention ?

Bad predictions
come from:

-- The general inability 
to predict the future,
unless a regular cycle
is discovered, such as
the 11-year sunspot cycle.

-- A lack of accurate data.

-- Guesses and biased
personal opinions about 
climate change physics.


The attention comes from
a model selection bias:
-- Models with scarier 
predictions get the 
most attention ! 

Perhaps because
humans evolved
with an internal
self-defense
mechanism
to focus on 
the worst risk.

So modelers are 
encouraged to make
scary predictions. 

Such as rapid future 
temperatures rises,
and rapid future
CO2 level growth.

Even "better" 
is predicting 
the end of our
planet as we 
now know it !

In twelve years !

Those predictions
use CO2 and global
average temperature
growth rates that are
too fast, versus our 
past 325+ years of 
actual experience 
with global warming.



Below is a chart that 
compares increases 
in atmospheric CO2,
which should have been
easy to predict, to the 
CO2 forecasts from 
models issued regularly 
since the 1970s: 








Since 1980, 
every model 
has over-predicted 
the amount of CO2 
that would accumulate 
in the atmosphere, 
as a result of fossil 
fuel burning. 

Reality is the 
dark black line. 

Model projections 
are the colored lines, 
rising too quickly 
above the black line. 

After about 1980, 
every model but two 
predicted more CO2 
accumulation in the 
atmosphere than 
was observed.

One of those two
models only ran 
from 1970 to 2000, 
and the other model
-- the green line --
went flat after 2000,
by assuming that
CO2 emissions 
would stop rising 
after 2000.

The same model, 
when run under 
the assumption 
CO2 emissions 
would continue 
to rise after 2000,
would have also
over-predicted
CO2 accumulation.



The initial distribution 
of unbiased models 
should be random
errors.

After comparing the 
model predictions
against actual results
over the next 
decades or two, and 
revising the models
to better match reality, 
future climate predictions 
should become more 
accurate.

But that's not done
with climate models !

Climate models 
have been making
the same mistake,
in the same direction, 
for 50 years running.

Those are not 
random errors 
-- there's obvious bias
to grossly exaggerate
global warming !

Models that suggest 
runaway temperature 
increases, are the 
models that allow
policymakers 
to demand ever-more 
stringent policies 
to reduce CO2 
accumulation.

Politicians get 
a "coming crisis"
they can 'fight'.

It's not real, but that
does not matter if 
enough people believe
a climate crisis is coming.

Scientists who make
the "right" predictions, 
get government grants 
and salaries, along with 
permanent job security. 

Most climate scientists
depend on federal or 
state government
funding.


The problem is NOT
in our computers, 
but with leftist politicians,
who pay for climate 
predictions that
scare people, and
then use an imaginary
coming climate crisis 
to control the scared
people !

Remember when 
cigarette companies 
used to pay for 
the "science" they 
wanted to "prove"
cigarettes were safe ?



If the leftist politicians
wanted to have accurate
climate predictions,
they would reward 
government bureaucrat
scientists for making 
accurate predictions.

And the mass media would 
publish articles about how
accurate the predictions
were, rather than ignoring
the fact that past predictions
consistently over-predicted
global warming by 2x to 3x !