# Is Global Warming Making Atlantic Hurricanes Worse?

*MY ESOTERIC likes to think of himself as a bit of a polymath with degrees in Statistics, Accounting, Computer Science, & Operations Research*

## Atlantic Hurricanes

The Atlantic hurricane season lasts from June until November each year. The busiest months are August, September, and October. That said, I remember hurricanes happening in April, May, December, and January in years past.

The main driver of hurricanes is ocean temperature and low-pressure areas. The warmer the ocean, the more frequent and powerful the storms can be. Other factors, principally the El Niño and La Niña effects in the Pacific Ocean, when present, can enhance or kill hurricanes. For a fuller explanation of how hurricanes form, read here.

This article is concerned with the ocean temperature aspect of hurricane formation. All data is sourced from here.

## Heating the Oceans

As the global air temperature rises, due to global warming or any other reason that may cause global temperatures to increase, so does the water of the oceans. The mid-Atlantic Ocean, the Gulf of Mexico, and the Caribbean are no exceptions.

As the water temperatures rise so does the probability they will spawn tropical storms and hurricanes. In addition to producing more storms, warmer and warmer water will produce stronger and stronger storms.

The question then is, is there evidence that Atlantic hurricanes are getting more frequent and worse. Yes, there is as you will see in the next several sections.

There are two ways, along with several variations, to determine if the pattern of Atlantic hurricanes is indeed changing over time. One is to study variations over time to see if there is a trend. The other is to take statistics (mean, standard deviation, etc) at various points in time and then to test to see if they are different from each other. We will do both in this article.

## Atlantic Hurricanes: 1850 - 2018

We will start with method one, looking at the complete time series of hurricane data to see what we see. I don't plot each year because there is way too much variability. For example, in 1914 there was one tropical storm and no hurricanes; but in 2005, there were 28 tropical storms, 8 minor hurricanes, and 7 major hurricanes; a total of 43, the most to-date.

So to avoid this confusing set of highs and lows, there are several techniques to "smooth" the data but still retain its essence. One of those is called "moving averages". What one does is take the average of the first so many data points and that becomes your first data point. You then drop the first point, add the next point, and take a new average; the second data point. Do it again and again until you have gone through all of the data.

I this case, I chose to use a 20-year moving average because weather moves in slow cycles. And a 20 year period is long enough to smooth out yearly and seasonal variations.

The chart below depicts a 168 year period of Atlantic storms using the 20-year moving average method (which is why it starts in 1869, There are ten lines on it representing, from top to bottom:

- Total Storms
- Total Tropical Storms
- Total Hurricanes
- Total Minor Hurricanes (1, 2)
- Total Major Hurricanes (3, 4, 5)

The dashed lines with each one are a linear trend line and the numbers represent the equation that best fits the trend and the "R-squared" value tells you the goodness of fit or how much of the deviations are explained by the equation.

## Read More From Soapboxie

Here is what this chart tells us.

### Total Storms

This is the sum of the total tropical storms (not including tropical storms that later turned into hurricanes) and the total hurricanes. Even without the help of the trend line, it is easy to see that the total number of Atlantic storms is increasing over time. Specifically, there is, on average, 0.0679 additional storms each year. How good is that prediction? Look at the R-squared value of 0.77. Any value above 0.7 means, as a rule, the equation is valid. Taken together then, this is good evidence that the number of storms is increasing over time due to global warming.

### Total Tropical Storms

Now we look in more detail behind the increasing number of storms. Again we see a clear indication that storm activity is increasing, at least at the tropical storm level. Here, the number of tropical storms is increasing, on average, 0.0507 tropical storms per year. How confident are we? The 0.81 R-squared tells us we are very confident.

### Total Hurricanes

Hurricanes are measured as Category 1, 2, 3, 4, and 5 right now. Each increase roughly represents an order of magnitude increase in the damage and devastation caused by these storms. In fact, they are thinking about adding a category 6 to the mix because storms are starting to get that violent. While still terrible, CAT 1 and 2 hurricanes are considered "minor" while 3, 4, and 5 are considered "major". This line represents all 5 categories.

In this case, we see that the total number of hurricanes is increasing at a rate of 0.0171 hurricanes per year. But how confident are we? Not as much at a 0.49 R-squared. But, there is a reason for that as we will see.

### Total Minor Hurricanes

With Minor Hurricanes, things change. Part of the prediction from weather scientists is that the proportion of minor to major storms will change with the weight going to major storms. In other words, as the earth heats up, we will see more frequent and more violent hurricanes as time goes on. Indeed, that is what the data is starting to show.

For minor hurricanes, we actually see the number of is possibly __decreasing__. They are decreasing at a rate of -0.0005 storms per year. Unfortunately, with a R-squared of 0.0012, you can't take that prediction to the bank. With an R-squared that low, the best estimate is simply the average of storms over that period which is 3.56 minor hurricanes per year.

### Major Hurricanes

Where minor hurricanes don't tell you much, major hurricanes speaks volumes. Here we see that the number of major hurricanes is increasing at a rate of 0..0176 major hurricanes a year. And, we are extremely confident that is true because the R-squared is a whopping 0.91!

Now we can see why total hurricanes has a low R-squared because it represents the combination of those from both the minor and major hurricane data. We can also see it is the increasing number of major hurricanes that is driving the overall increase in hurricanes. That is not good for people but good evidence that global warming is driving more and more devastation.

The chart below is a clearer picture of what we just discussed.

## Testing Statistical Means For Global Warming

I said there were two methods of testing the hypothesis that Atlantic Storms are increasing due to global warming. We just covered the first, and the picture was clear. Now let's look at the second.

In this case, I am going to divide up the data into two large groups of about equal years - 1850 - 1934 and 1935 - 2018. Then we will calculate the mean and standard deviation of each group. With that, we will test to see if the means of each group are NOT EQUAL, meaning there is something going on. On the other hand, if they are EQUAL,or the second mean is smaller than the first then there is no support for global warming based on increasing Atlantic storm activity. We will consider each of our five demographics.

### Total Storms

Group 1 (85) Mean - 12.08 Storms per Year

Group 1 (85) SD - 5.6

Group 2 (84) Mean - 18.4 Storms per Year

Group 2 (84) SD - 6.56

On the face of it, it certainly looks like those means are different with Group 2 being much bigger - but are they statistically? Using a standard statistical test between two means, the t-test, we find that the probability that those two means are different is greater than 0.9999. Said another way, there is a 99% chance (at a confidence level of 95%) that the number of storms in Group 2 (1935 - 2018) is larger than Group 1 (1850 - 1935); a very strong indicator of global warming.

### Total Tropical Storms

Let's do the same thing for tropical storms.

Group 1 (85) Mean - 7.26 Storms per Year

Group 1 (85) SD - 3.64

Group 2 (84) Mean - 11.89 Storms per Year

Group 2 (84) SD - 4.04

Again, on the face of it, it certainly looks like those means are different with Group 2 being much bigger . Still using the t-test, we find that the probability that those two means are different is greater than **0.9999**. Said another way, there is a **99%** chance (at a confidence level of 95%) that the number of storms in Group 2 (1935 - 2018) is larger than Group 1 (1850 - 1935); a very strong indicator of global warming.

### Total Hurricanes

Let's do the same thing for tropical storms.

Group 1 (85) Mean - 4.82 Storms per Year

Group 1 (85) SD - 2.36

Group 2 (84) Mean - 6.51 Storms per Year

Group 2 (84) SD - 2.87

Again, on the face of it, it certainly looks like those means are different with Group 2 being much bigger . Still using the t-test, we find that the probability that those two means are different is greater than **0.9999**. Said another way, there is a **99%** chance (at a confidence level of 95%) that the number of storms in Group 2 (1935 - 2018) is larger than Group 1 (1850 - 1935); a very strong indicator of global warming.

### Minor Hurricanes

Let's do the same thing for tropical storms.

Group 1 (85) Mean - 3.47 Storms per Year

Group 1 (85) SD - 1.97

Group 2 (84) Mean - 3.65 Storms per Year

Group 2 (84) SD - 1.75

Again, on the face of it, it does not look like those means are different with Group 2 being much bigger . Still using the t-test, we find that the probability that those two means are different is only **0.4699**. Said another way, there is a **47%** chance (at a confidence level of 95%) that the number of storms in Group 2 (1935 - 2018) is larger than Group 1 (1850 - 1935); not an indicator of global warming at all. This is similar to the result we found in the first analysis for minor hurricanes.

### Major Hurricanes

Let's do the same thing for tropical storms.

Group 1 (85) Mean - 1.35 Storms per Year

Group 1 (85) SD - 1.39

Group 2 (84) Mean - 2.86 Storms per Year

Group 2 (84) SD - 2.01

Again, on the face of it, it certainly looks like those means are different with Group 2 being much bigger . Still using the t-test, we find that the probability that those two means are different is greater than **0.9999**. Said another way, there is a **99%** chance (at a confidence level of 95%) that the number of storms in Group 2 (1935 - 2018) is larger than Group 1 (1850 - 1935); a very strong indicator of global warming.

## Conclusion: Global Warming is Real

It should now be obvious, even to the deepest skeptic who doesn't have an agenda that, at least based on increased activity of Atlantic storms, that global warming isn't real. All you have to do is believe that:

- Atlantic storms are related to water temperature
- That the warmer the water, everything else being equal, you get more storms
- That the warmer the water, everything else being equal, you get more violent storms
- That the warmer the air, the warmer the surface water
- That all scientists agree to the above

If you believe those things, then the data shows that Atlantic storms are increasing in number and ferocity and therefore it is global warming that is driving it.

*This content is accurate and true to the best of the author’s knowledge and is not meant to substitute for formal and individualized advice from a qualified professional.*

**© 2019 Scott Belford**