Total Pageviews

Thursday, January 23, 2020

Hurricane Counts Use Data Mining for Deliberate Deception

The Pielke graph below,
on hurricane landfalls, 
has been used here.

A climate alarmist
has turned it into 
bad news by the use
of data mining.










The gray bars 
are the strong 
hurricanes,
Category 3+.

If you data mine,
you could state
that Category 3+ 
landfalling 
hurricanes
have been in 
a rising trend
since the 1970's,
when sorted 
by decade:

Category 3+ 
Landfalling 
Hurricanes
 Sorted by 
  Decade:
  1970s = 38 
  1980s = 41  
  1990s = 55
  2000s = 58
  2010s = 60

The conclusion 
would be a rising
trend for Cat 3+
in the past 
50 years.

I had previously
stated there was
no obvious hurricane 
trend ( Category 1 - 5 ).

That's still true.


(1)
Some of the Cat 3+ 
increase could be 
attributed to better 
measurement, 
sensing and 
detecting.

Wind speeds
can slow 
very quickly
just before
as a hurricane 
makes landfall.

Wind speeds 
drop drastically 
as a hurricane 
crosses a coast, 
where there are 
more observers.


(2)
Data sorted by decade
could be deceptive too.

Any arbitrary division 
of years can produce
meaningless patterns.

Throw dice fifty times, 
divide the results
into groups of ten,
and we are likely
to see meaningless
"patterns".


(3)
The best
explanation 
is that the 1970s 
starting point 
for the first chart, 
happened to be 
a quieter period 
for overall tropical 
cyclones / hurricanes.

In fact,
major hurricanes 
were just as frequent 
in the 1950s and 1960s 
as they are now. 

There's a 
cyclical pattern, 
and no obvious trend 
since the 1950s.

Data prior 
to the 1950s
are questionable.

The cyclical 
pattern
is linked 
into the 
Atlantic 
Multidecadal 
Oscillation. 

NOAA describe its effect:

We have been 
in the AMO 
warm phase 
since the 
mid 1990s, 
just as 
we were 
in the 1950s. 

In between, 
the AMO was 
in its cold phase,



Starting
in the 1970s, 
and just 
looking at 
Category 3+
landfalling 
hurricanes,
is data mining,
that leads 
to a different 
conclusion
than looking 
   at ALL 
good data 
since 1950, 
looking at 
    ALL 
landfalling 
hurricanes.

In the 
North Atlantic,
and the Western 
North Pacific, 
good data
are available 
since 1950.

Hurricane 
landfalls 
in those 
two basins 
accounted 
for about 68% 
of all global
landfalls from 
1970 to 2019.

The chart below 
shows a 10-year 
moving average 
of tropical cyclone 
global landfalls 
from 1950 to 2019. 

The numbers 
since 1970 
are identical to 
the top graph. 

Numbers from 
1950 to 1969, 
are based on 
actual landfalls 
for the N Atlantic 
and NW Pacific, 
added to the average 
for the rest of the world
between 1970 and 2019. 

   The dotted line 
is major hurricanes.








To be more specific,
( and more confusing )
    1970 to 2019 
     is based on 
observations, and 
uses data identical 
to the first chart. 

The black line shows 
all tropical cyclone 
landfalls of at least 
hurricane strength 
      and the 
dashed black line 
shows all landfalls 
of tropical cyclones 
        of major 
hurricane strength. 

For 1950 to 1969 
      there are
actual data on landfalls 
in the North Atlantic 
and Northwest Pacific, 
added to the average 
number of landfalls 
from the rest 
of the world ,
    based on 
observations from 
1970 to 2019 
(there is no trend 
over the period). 

The grey line and 
dashed line represent 
an estimate of total 
hurricane-strength 
        and major 
hurricane-strength 
tropical cyclone landfalls 
from 1950 to 1969. 

The dotted lines represent 
plus and minus one standard 
deviation from this average.