Marks Exploration 1 - Hurricanes and The Golden Ratio - Non - Annotated
Marks Exploration 1 - Hurricanes and The Golden Ratio - Non - Annotated
The aim of the exploration is to determine whether there is a correlation between the similarity of
the pattern of a hurricane to the golden ratio and the amount of damage the storm causes
(measured in US Dollars). The reason I chose this topic for my exploration is because I have always
been interested in the concept of the golden ratio, after having heard about its use in art and
architecture, making me curious as to how this worked. As I continued to do some research into the
subject matter, I soon realized the extent of its presence in nature. One of the areas was within the
patterns formed by hurricanes. While it is unlikely for pattern to achieve the golden ratio exactly, as it
is an irrational number, it is possible to find a similarity. This made me wonder whether there was
any correlation between the damage caused by a storm and how much the pattern of a storm
There are many hurricane-prone areas across the world that can often suffer from great damage to
their infrastructure due to these storms. If we can find a correlation between the damage caused by
previous hurricanes that have passed and how much the pattern of the hurricane differs from the
golden ratio, it could be used to help prepare these areas for the damage that they will endure before
First I will find the height and width of 24 hurricanes using images found online, and determine the
height : width ratio along with the amount of damage the hurricanes caused in dollar using the
greatest horizontal and vertical deviations to form the ratio. I will then be using subtraction in order to
calculate the deviation of each of these height to width ratios from the golden ratio ( 1.618) through
subtraction to find the deviation. The ratio is rounded to 4 significant figures so that once the
deviation is found, the values can be graphed on a realistic scale. Following this I will calculate the
PMCC value and determine whether there is a correlation between the variables and use hypothesis
testing to determine whether there is enough statistical evidence to support my hypothesis. Lastly, I
will use a regression line to determine how the independent variable (deviation from golden ratio)
2
Data Collection:
Data Table:
The first step is to collect data on height, width and damage caused and calculate the ratio. I
used a digital ruler in order to measure the two furthest points vertically and horizontally
perpenclicular to this from images of the hurricanes. The height was then divided by the width
1033]
Michael 10.6 15.9 0.667 25 [ CITATION Mor19 \I 1033]
Florence 5.87 5.95 0.987 24 [ CITATION Dun19 \I 1033]
Humberto 9.26 15.9 0.579 0.025
Wilma 8.89 13.5 0.659 22.3
Rita 15.6 12.3 1.27 18.5
Irene 6.61 5.24 1.26 15.8 [ CITATION Fre12 \I 1033]
Matthew 15.6 15.8 0.987 6
Ivan 6.61 5.03 1.31 27
Ike 5.45 6.56 0.831 7.4
Epsilon 6.85 5.24 1.31 4.3
Gustav 11.1 14.2 0.782 6.9 [ CITATION Ama19 \I 1033]
Camille 15.6 12.1 1.29 9.9
Jeanne 10.9 10.5 1.04 7 [ CITATION Kid20 \I 1033]
Allen 6.09 5.58 1.09 0.3[ CITATION Fre \I 1033]
Frances 6.67 3.86 1.73 10.1
3
Paloma 5.56 6.46 0.861 0.4544
Charley 13.4 11.6 1.16 20
Stan 9.13 9.05 1.01 3.96
Calculation of deviation:
Subtraction is now used to calculate the deviation between the hurricane ratio and the golden
4
Scatter Graph:
A scatter graph can be used in order to show how closely correlated the data points are. The
damage caused can be plotted against the y-axis while the deviation can be plotted against the
•
•
-so-, t----+----1-----+----+---+---+---
The graph shows the results plotted showing a relationship between the two variables. From
the graph it can be seen that the overall trend of the results shows that the smaller the
deviation from the golden ratio, the higher the cost of damage was, showing an overall
negative correlation. For example, Hurricane Sandy bad a deviation of 0.198 and damage
worth $ 70 billion. Hurricane Andrew had a larger deviation than Hurricane Sandy of 0.278
and smaller cost of damage worth $61.9 billion. Some of the points however show small
deviation and less damage which could explain why the results do not show perfect
correlation. An example of one of these data points could be for Hurricane Frances where
deviation was relatively small (0.112) and so was the damage ($10.1 billion).
5
Linear Regression:
The linear regression can be calculated as follows to determine the line of best fit of the
scatter graph.
y = 23.78
x = 13.72
SSx= 1.71
6
Y intercept= y-bx = 39.6
y= -27.78x + 39.6
From this deduced equation, the data can be interpolated and extrapolated. Interpolation
involves using the equation to estimate the value of data within the range of the data set.
For example, in order to find the amount of damage caused of a hurricane which deviates
from the golden ratio by 0.76, it would need to be substituted into the equation as follows.
-27.79(0.76) +39.6 = y
y= $ 18.5 billion
Hence, theoretically the damage caused would be approximately 18.5 billion dollars.
Extrapolation on the other hand uses the linear regression equation to estimate the value
for data outside the range of the data set. So for example the value for a hurricane that
deviates from the golden ratio by 1.2 can be determined using extrapolation.
y= 6.252
As per the equation, the data shows that if the x value (deviation) is 1.2, it would cause
Calculation of PMCC:
PMCC can determine whether there is a correlation between the two variables. I have used
this calculation in order to help determine the strength of the correlation between the
deviation of the ratios and the damage it caused, if any, as per the aim. PMCC can be
7
X;-X
l
l
l2L ( y;-y)2
r= l
Ll
.JI
I (x;-x)(y;-y)
l
y )
0.298 81 -0.274 57.282 0.075 3281.192 -15.671
0.87 50 0.298 26.282 0.089 690.727 7.843
0.931 3 0.359 -20.718 0.129 429.248 -7.447
0.278 61.9 -0.294 38.182 0.086 1457.842 -11.21
0.198 70 -0.374 46.282 0.14 2141.995 -17.29
0.618 94.4 0.046 70.682 0.002 4995.902 3.281
0.951 25 0.379 1.282 0.144 1.643 0.486
0.631 24 0.059 0.282 0.004 0.079 0.017
1.039 0.025 0.467 -23.693 0.218 561.373 -11.075
0.959 22.3 0.387 -1.418 0.15 2.012 -0.549
0.348 18.5 -0.224 -5.218 0.05 27.231 1.167
0.358 15.8 -0.214 -7.918 0.046 62.7 1.691
0.631 6 0.059 -17.718 0.004 313.938 -1.053
0.308 27 -0.264 3.282 0.069 10.77 -0.865
0.787 7.4 0.215 -16.318 0.046 266.287 -3.515
0.308 4.3 -0.264 -19.418 0.069 377.071 5.118
0.836 6.9 0.264 -16.818 0.07 282.855 -4.447
0.328 9.9 -0.244 -13.818 0.059 190.946 3.366
0.578 7 0.006 -16.718 0 279.502 -0.107
0.528 0.3 -0.044 -23.418 0.002 548.417 1.021
0.112 10.1 -0.46 -13.618 0.211 185.458 6.259
0.757 0.4544 0.185 -23.264 0.034 541.209 -4.314
0.458 20 -0.114 -3.718 0.013 13.826 0.422
0.608 3.96 0.036 -19.758 0.001 390.391 -0.72
Total = 1.713 Total= !fatal=
17052.613 -47.591
8
-47.59 f = -0.28
r= ✓17052.6 X 1.713
The negative r value indicates that there is a negative correlation between the two variables.
This means that as the deviation from the ratio decreases the amount of damage caused
increase. However, this correlation of only -0.28 appears to be weak as it is closer to 0 (no
Hypothesis Testing:
Hypothesis testing can use the r value calculated for the PMCC in order to determine whether
there is likely to be an overall general trend in the entire population. In order to determine
whether the PMCC of an entire population is greater than or less than zero, a two-tailed test
needs to be used. This is so that it can be determined whether the results from this exploration
can be considered applicable to determining the destructivity for other hurricanes in future to
Ho = There will be no correlation between deviation from the golden ratio and the damage
caused
Ho: p=O
H 1 : As deviation from the golden ration increases, damage caused will decrease
Sample size : 24
9
Prodlld fflO!llflll coeffidt'DI
Le,d :ample
0.10 0.05 0.025 0.01 0.005 Le,l'I
0. (JOO 0.9000 0.9500 09800 0990() -I
06"70 0. 05-1 o., 7 3 0.93-13 0.95 7 5
06084 0729, 0.8114 0.88:?2 0.9172 6
05509 0.66� 07545 08129 0. 7-15 7
0.5067 0.621S 0 �067 0.78S7 0.81-H 8
0-1716 O.S822 0.66<..l0.749 0.79TT 9
0.-142 0.5-194 0.6319 07155 0.7646 10
04187 0.5214 0.M.21 06X51 07'4 II
U.39 I 0.-1973 0.5760 06 I 0.7079 12
0 J802 04762 0.5529 06339 0.6Rl:S ll
0 16-16 04�75 O.S324 0.6120 0.661-1 14
0.3507 0.4409 0.51-IO 05923 0.6411 IS
Ol\81 0 4259 0.-$971 0.5H2 06226 16
0.3271 0.-112-1 0.-111210.55"7 0.605-5 17
0.3170 0.4000 0.-1683 0.5425 0.5li97 18
I 0.1077
om.2
0 187
031 l
0.4555
0-1-138
o s2 ·s
05155
0.S7SI
0.5614
19
:?O
U 2914 0.16.�7 04129 0�.\4 0.5-lli7 21
II 2841 0159 0-1�7 04921 05'6ll 12
02774 0.3515 0413' 04815 0.5256 23
0.2711 o.3.rn, O-IO-l4 0-1716 0.5151 .N
026H OJJ65 01961 04622 0 5052 25
0259 0,:1297 018 2 045� 0.495 26
0.2S46 01:m 01809 0-I-ISI 04869 27
0 2497 01171 03739 04'72 0471,5 28
02-151 o ms 01(173 04297 0.470:S 29
02407 0.3061 0.1610 04226 0.4629 JO
02070 0.2638 0.3120 0."1665 0-1026 40
01841 021H 027 7 012 I 0.3610 �o
01678 021-1-1 0 25-12 om1 O.HOI 60
01550 019 2 OHS:? 0 :?T'6 0. l(l(,O 70
0.144 0.1852 02199 O . .?S97 0..! 6-1 0
0136-1 01745 0.2072 0..:'449 0..!701 90
0.1292 0.16S-I 0.1966 0.2324 0.2565 100
Using the table of critical values, the r value for a significance level of 5% on each tail with a
sample size of 24 is r= ± 03438 therefore making the critical region r>0.3438 and r<-0.3438.
-0.28 does not fall into this region so Ho must be accepted and H 1 must be rejected.
Subsequently, from this data, that at a 10% overall level of significance there is no corre1avofl
between the similarity of the pattern to the golden ratio and the amount of damage it caused.
This therefore means that you cannot rely on the linear regression model found previously.
10
Exponential Model:
It can be said that linear regression is relatively unreliable as the assumption that the
correlation between the variables is linear must be made. Hence, on the below graph the data
the
Damagi, caused (in<i<NMil§ffi9n between Deviation from
Golden Ration and Damage Caused
100 Y.= 45.262e.19x
90
80 R2 = 0.1575
70 e
60 •
•
.. .
50
40
30
•••••••··········· · •
.
20 · · · --.. .. •
..
.......
10 • •
........
............ .....tt.tl..............
.
0.2 0.4 0.6 0.8 1 1.2
The r2 value for this graph was found using an online calculator producing a value of 0.1575,
giving an r value of 0.39686 showing a much stronger correlation between the variables than
Conclusion:
Limitations:
The data sample focused only upon the correlation between the pattern and damage when in
reality there are likely other factors that would also affect the amount of damage caused that
were not taken into consideration such as the infrastructure of the land. If the affected area
was better designed for natural disasters the amount of damage would be less. Furthermore,
another potential extraneous variable that may have led to an inaccuracy of results was that
11
while collecting data, the amount of damage caused in dollars was measured according to the
value the dollar held at the time of impact. As currency fluctuates, the value for the dollar in
one year may have varied to that of another, enough to impact the results. This lack of
standardization would mean that the currency would not have been a reliable grounds for
comparison of damage.
In future when exploring this topic, a more holistic view should be taken and the degree of
impact of variables such as the aforementioned ones should be taken into account. Factors
such as such number of injuries and fatalities should also be taken into consider as damage
Conclusion:
To conclude a correlation between the damage caused and the deviation of a hurricane pattern
from the golden ratio can be found. The PMCC provides a negative r value of -0.28
suggesting that they are inversely proportional however the hypothesis testing shows that the
results are not significant and that this correlation is not generalizable to all other storms. This
lack of significance could be due to the many other factors that were not considered during
this exploration leading to a limited reliability in the results found. However, as seen with the
exponential model of the graph, while there is correlation between the data it is not
necessarily linear meaning that the variables do not change at the same rate.
Bibliography
12