Atlantic Hurricanes Trend

Climate alarmists claim that Atlantic hurricanes will increase in frequency and intensity due to emission of carbon dioxide. Is this true?

NOAA provides the data (HURDAT2) we need to examine this claim. Let’s first look at the frequency of hurricanes:

Hurricane Occurrences per Year

Their first claim has some evidence, but let’s give this some thought: is measuring the frequency really sensible? Wouldn’t it make more sense to measure the amount of time the Atlantic spends in hurricane mode? Yes, I think that is a better measure.

Hours of Hurricanes per Year

The amount of hours of hurricanes per year shows absolutely no trend!

What about their second claim: Is intensity increasing?

We can figure out hurricane intensity using a hurricane’s lowest pressure as a proxy. The lower the pressure the more intense the storm.

Here is all the hurricanes and their lowest pressure value:

Hurricane #’s Lowest Pressure

There is absolutely no trend in hurricane intensity in nearly 170 years!

Clearly, climate alarmists are wrong in regard to Atlantic hurricanes.

That’s all. Enjoy 🙂 -Zoe


# Zoe Phin, 2021/05/20
# File:
# Run: source; require; download; freqplot; hoursplot; presplot
# Output: freq.png, hours.png, pres.png

require() { sudo apt-get install -y gnuplot; }

download() { wget -cO atl.csv ""; }

cma() {
    cut -c6- | tr '\n' ' ' | awk -vp=$1 '{
    for (i=0;i<p/2;i++) print ""
    for (i=p/2;i<=NF-p/2;i++) { s=0
        for (j=i-p/2; j<=i+p/2; j++) s+=$j/(p+1)
        printf "%.4f\n", s

freqplot() {
    cat atl.csv | tr -d '\n' | sed 's/AL[0-9]/\nAL/g' | grep HU | awk '{
        print substr($1,4,4)}' | uniq -c | awk '{ 
            print $2" "$1 }' | tee freq.csv | cma 10 > freq.cma

    paste -d ' ' freq.csv freq.cma > freq.plt
    echo "set term png size 740,480; set mytics 2
    set key outside top center horizontal
    set grid; set xrange [1850:2020]
    plot 'freq.plt' u 1:2 t 'Hurricanes' w l lt 3 lw 2,'' u 1:3 t '10yr CMA' w l lt 6 lw 4
    " | gnuplot > freq.png

hoursplot() {
    awk -F, '$3=="  " && $4==" HU" { 
        print substr($1,1,4) }' atl.csv | uniq -c | awk '{ 
            print $2" "$1*6 }' | tee hours.csv | cma 10 > hours.cma

    paste -d ' ' hours.csv hours.cma > hours.plt
    echo "set term png size 740,480; set mytics 5
    set key outside top left horizontal
    set grid; set xrange [1850:2020]
    plot 'hours.plt' u 1:2 t 'Hurricane Hours' w l lt 3 lw 2,'' u 1:3 t '10yr CMA' w l lt 6 lw 4
    " | gnuplot > hours.png

presplot() {
    cat atl.csv | awk -F, 'NF==4 { print $1 } NF>4 && $3=="  " && $4==" HU" && $8!=" -999" { print $8
        }' | tr -d '\n' | sed 's/AL/\nAL/g' | awk 'NF>1{print}' | awk '{
            min=9999; for (i=2;i<=NF;i++) if ($i < min) min=$i; printf "%04d %d\n", NR, min
        }' | tee pres.csv | cma 50 > pres.cma

    paste -d ' ' pres.csv pres.cma > pres.plt

    echo "set term png size 740,480; set mytics 2
    set key outside top left horizontal
    set grid; set yrange [1005:880]
    set xrange [0:611]; set xlabel 'Hurricane #'
    set xtics nomirror; set x2tics ('1853' 1, '1900' 73, '1950' 182, '(Year)' 305, '2000' 466,'2019' 606)
    plot 'pres.plt' u 1:2 t 'Lowest Pressure' w l lt 3 lw 2,'' u 1:3 t '50 Hurricanes CMA' w l lt 6 lw 4
    " | gnuplot > pres.png

Published by Zoe Phin

32 thoughts on “Atlantic Hurricanes Trend

  1. Of course there are more Hurricanes Reported now, they didn’t have Satellites back before the 1980s.
    Now every Storm gets rated a hurricane by satellite measurements, but not by Ground based instruments.

    Liked by 1 person

    1. Satellite imagery started in 1960

      However the images were not archived until 1978. Ryan Maue highlighted this paper from 2020 the other day (Google translated from French)

      “This research on the North Atlantic attempts to establish a link between the activity of extreme hurricanes (categories 4 and 5) and global warming over the period 1945-2018. The intensity of the hurricanes from 1945 to 1977, estimated from aerial reconnaissance, comes from the Hurdat database at the National Hurricane Center in Miami. That of the hurricanes from 1978 to 2018 was estimated with the Dvorak technique from satellite imagery (beginning of the archives in 1978). Three indicators were studied: the number of extreme hurricanes, their proportion in all hurricanes (categories 1 to 5) and the absolute intensity of hurricanes per decade. Over the period 1945-2018, there is no upward trend in the number of extreme hurricanes. The proportion of the latter in all hurricanes increases sharply with the positive phases of the Atlantic Multidecadal Oscillation (AMO), and it decreases significantly with the negative phases of the AMO. The proportion doubles during La Niña episodes compared to El Niño phases of the Southern Pacific Oscillation. There has been no increase in the absolute intensity of hurricanes per decade. It is still difficult to detect a possible signal of global warming on hurricanes due to natural variations at different time scales. There has been no increase in the absolute intensity of hurricanes per decade. It is still difficult to detect a possible signal of global warming on hurricanes due to natural variations at different time scales. There has been no increase in the absolute intensity of hurricanes per decade. It is still difficult to detect a possible signal of global warming on hurricanes due to natural variations at different timescales.”

      Liked by 1 person

  2. Great post as usual, thanks Zoe. Would it make sense to plot an indicator that combines duration and intensity, sort of “energy” trend of the hurricanes. For instance by simply multiplying the two graphs….

    Liked by 1 person

      1. Ryan Maue has been performing this calculation — he calls it Accumulated Cyclone Energy — since his days as a graduate student at Florida State.

        He has ACE data going back at least 4 decades for all the cyclonic activity in at least 4 oceanic areas: Atlantic, Indian, W. Pacific, E. Pacific.

        Liked by 1 person

  3. Thanks for the analysis, very comprehensive and compelling ! However we seem to be at peak climate zealotry at the moment and all reason has been ditched.

    Liked by 1 person

  4. Before hurricane hunter airplanes, before satellites, before RADAR, before storm chasers and all the other technologies and people looking for storms, it was possible for land and ocean cyclonic storm to brew up and blow out without anyone ever knowing they’d happened. Hurricanes and cyclones at sea, if they didn’t run over an inhabited island or a ship, nobody knew the storm was there.

    What would be an interesting but time consuming project is to apply those rules of non-observability to the modern hurricane and cyclone record. If a storm was only observed by remote sensing, and the only people who encountered it directly specifically went out there to do so, take it out of the list.

    When the modern era is winnowed down to the sort of storms that would’ve garnered a report back in the day, then you have a stick to measure the amount of storms that may have happened in the past but nobody saw.

    Liked by 1 person

  5. Zoe,
    Pinging here from a dying, and rather too negative thread on a favorite climate skepticism website. This should give you a little more id info on me.
    I’m still impressed by the approach of hurricane*time and atmposheric pressure for this phenomenon. I accept your caveats as given on the other site. I think It is a good measure, but I’m not equipped to falsify it. I’m still hoping that you get constructive feedback.
    AKA DK_


    1. The only valid criticism I can think of: the resolution is limited to 6 hours. All hurricane*time is a multiple of 6 hours. Hurricanes obviously don’t work on such a rigid schedule.

      Thanks, Dave.


      1. I won’t reiterate my reply from that other web site. I think the measurement you’ve propose is valid. Rather than criticism, I should have said something along the lines of a need for scientific validation or falsification (if I was edumacated properly, I’d probably know how to put it). There’s plenty of people more qualified than I am that should be able to tell us we are right, or wrong and why. Personally, I think you should have naming rights and credit, but what do I know?
        I’ll take an autographed photo from your Nobel acceptance, please. Not the avatar one.


      2. So I had a think about the 6-hour interval. A hurricane is either below a standard atmospheric pressure or not, so it is like a switch for our purposes. “Off” when above the standard, and “on” when the eye falls below that pressure. It seems to me if we know at (24-hour clock) 00, 06, 12, and 18 hours whether a hurricane is “on” or “off”, we know that there can be as much as (full hour) 5 hours when a hurricane was “on” and not yet reported. We know that there may be as much as 5 hours when a hurricane was “on” until say 1700 hrs, and yet “off” when we checked again at 1800 hrs.
        Is it true that we know our minimum “on” time, but our maximum “on” time is as much as 10 whole hours later? Where have I messed up so far? I have more, but this seems like a good chunk.


        1. Let’s think of 3 abstract hurricanes:

          1st lasted from 17:59 to 18:01 (2 min)
          2nd lasted from 15:00 to 19:00 (4 hours)
          3rd lasted from 12:01 to 11:59 (nearly 12 hrs)

          Since we have observations only at 6 hour intervals, I would record 6 hours for all of them, due to entry at hour 18:00.

          This would be wrong. But there’s no more data!

          But for non-abstract hurricanes, which tend to last for several days, the error is reduced.

          I think you understand it.


  6. Back to 6-hours then. How much does it change the chart if we randomly add 0, 5 or 10 hours to the time for all the sample set hurricanes. Does the trend look wildly different?

    How much does the trend vary if we weight shorter hurricanes over longer ones-Leaving the long ones alone, but adding as much as 10 hours to the shortest?

    I think what I’m trying to kludge around with is the question “How much does lack of precision matter to the trend?” I’m fumbling a bit, but is it developing a confidence level for our the trend line overall?


    1. That’s a lot of work. If we add truly random time, it will be a normal distribution and shouldn’t matter – which is the thought I had in mind, and why I didn’t stop doing this.

      But what if not simple random distribution? What if there is time favortism? I don’t know. Can’t know. And so I leave it be.


      1. Random distribution is what I think I was after. Agreed. I was tending more toward, “how bad could it be?” But that is guessing.
        Thanks for bearing with me.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: