# Modified HSI better reflecting IKE values, but easy enough for forecasting

By: RTSplayer , 3:33 PM GMT on September 06, 2012

I played around with HSI and developed a modified system for ranking, which more closely reflects IKE on a linear scale of 1 to N, with no limit, ( so far, comparing ratios of this system closely resembles comparing ratios for IKE values.) This is a multiplicative index, so both AREA and intensity are accounted for.

These numbers do NOT reflect rainfall or additional surge or wave setup due to variations in movement speed, because that is too complicated to put in an index that would be useful for "generalized" forecasting.

How this works:

First:

Follow the guidelines for the HSI on Impact Weather.

Impact Weather HSI page

Second:
Since HSI calculate "size" as the RADIUS of winds at each category, I found this to be lacking, and it shows in some of the damage calculations.

I found that RELATIVE AREA is what we need. Therefore the "size" score should be squared.

8 would become 64.
13 would become 169.
15 would become 225.
etc

Third:
Because this is a large number, I want to re-normalize it to a value between 0 and 1, you'll see why later. so divide the result above by 25^2.

225/625 = 0.36

This becomes a normalized, linear value representing "area" with a maximum score of "1" if the HSI "size" value was 25.

Fourth:
Remove the "top cap" from the intensity part of HSI (this only effects a few storms), and recalculate the value if needed.

90kts has a rank of 9, for example.
180kts has a rank of 36.

Multiply The intensity value from HSI by the result of step 3.

All decimals should be kept until the final step.

0.36 * 9 = 3.24

Fifth:

Set value to 1 if less than 1.

0.2 -> 1

round UP to nearest tenth otherwise.

3.24 -> 3.3

That is your new "Absolute Severity Index" value.

Shorthand formula:

ASI = ((HSI Size^2)/625) * (HSI Intensity)

Here are some example values for classic storms:

Leslie(now): 1 (rounded up from 0.2)
Michael(now): 1.2

References:

Dennis(05): 1 (rounded up from 0.69)

Gustav(LA): 1.8
Isaac: 2
Andrew(Fl): 2.4

Andrew(LA): 3.1 (Yes, hit low population area here.)
Gustav(Cu): 3.7
Rita(LA/TX): 4.1
Ike(TX): 4.7

Camille(LA/MS): 9.5
Katrina(LA/MS): 11.1
Wilma(Fl): 12

Hugo: 14.8
Betsy: 15
Carla: 17

Wilma(Mx): 19

Katrina: 21.2 - peak over water

Wilma: 28.8 - Official peak over water

Wilma: >= 36 - theoretical Peak over water
Typhoon Tip: >= 36

These values give the "real" linear relative damage potential, given the exact same track and movement speed. The relative ratios are not identical to IKE in all cases, but are very close for most storm vs storm comparisons...

Michael actually has larger wind fields at important benchmarks than does Leslie, in spite of appearances on satellite alone.

In general, if Katrina, Wilma, and Camille all hit the same place under the same conditions, you'd expect them to do about the same damage, with Wilma and Katrina being marginally more powerful than Camille.

Katrina would be expected to do about 3.6 times as much damage as Andrew if it hits the exact same location at their respective landfall intensities, and this is close to reality.

The Levee failures amplified Katrina's dollar value somewhat, and the death toll significantly compared to these storms.

Because I save decimals at every step (and redo intensity calculations using the nearest 1kt,) I can get this value very close.

Cat: Min/Avg/Max value by category:

TS: 1/ 1.0/ 1
c1: 1/ 1.0/ 2.8
c2: 1/ 2.9/ 10.2
c3: 1/ 4.1/ 14.2
c4: 1/ 5.9/ 20.2
c5: 1/ 8.0/ 36 or more

You can easily see now why some "low category" storms are actually more destructive than storms significantly higher than them.

Over all, Andrew ranks less than the maximum category 1 storm, or even the average category 2 storm, it just happened to hit a more populated area with higher land values.

I am not intendeding to plagiarize the HSI.

I simply recognized an "easy" fix to some flaws with it, and have presented this as an alternative, or "fix" for those flaws.

Comments will take a few seconds to appear.