A discussion about flight level / sfc. wind reduction factor

[Chris_in_Tampa]

I wanted to talk about a feature I added to my recon decoder:
http://www.tropicalatlantic.com/recon/

While the following is Dean related, I only use it as an example storm since it is obviously relevant, but the discussion applies to all storms.

It relates to calculating a separate reduction factor for every line of an HDOB observation and then calculates the estimated 30 second sustained winds at the surface. I think it is a fairly accurate tool that shows some of the interesting wind profiles in the hurricane.

Take this observation from Hurricane Dean on August 20:

URNT15 KNHC 200354
AF304 1104A DEAN HDOB 32 20070820
034430 1737N 07859W 6969 02531 9298 +161 +153 109053 053 056 002 00
034500 1739N 07859W 6964 02545 9313 +157 +151 102073 095 098 000 00
034530 1740N 07859W 6967 02581 9383 +128 +128 101122 135 114 000 00
034600 1742N 07859W 6968 02646 9448 +130 +130 100140 142 117 000 00
034630 1744N 07859W 6966 02706 9501 +138 +119 100134 136 118 000 00
034700 1745N 07859W 6955 02764 9560 +128 +128 100132 133 125 000 00
034730 1747N 07859W 6968 02786 9600 +130 +130 098128 129 122 000 00
034800 1749N 07859W 6957 02835 9690 +088 +088 097140 142 104 020 00
034830 1750N 07859W 6973 02851 9728 +081 +081 095143 144 100 021 00
034900 1752N 07859W 6967 02886 9790 +067 +067 094135 136 092 018 03
034930 1753N 07859W 6966 02910 9810 +067 +067 096132 133 092 020 00
035000 1755N 07859W 6967 02934 9828 +074 +074 097126 127 085 009 00
035030 1756N 07859W 6966 02956 9847 +076 +076 096120 123 082 011 00
035100 1758N 07859W 6969 02972 9868 +076 +076 095115 116 079 010 00
035130 1800N 07859W 6966 02993 9892 +077 +077 095110 111 073 008 00
035200 1801N 07859W 6962 03011 9910 +067 +067 095105 106 071 010 05
035230 1803N 07859W 6981 02998 9922 +068 +068 097103 105 067 017 01
035300 1805N 07859W 6984 03001 9924 +069 +069 102101 103 065 016 01
035330 1806N 07859W 6963 03032 9925 +076 +076 104101 104 999 999 03
035400 1808N 07859W 6960 03043 9937 +073 +073 105100 100 999 999 03

You can copy the whole text and out it into my decoder. But even if you don't you can follow the logic below.

For what follows, refer to the colored areas above.

The NHC says this at 5AM EDT about Dean and these two particular HDOB observations:

"OBSERVATIONS FROM THE LAST AIR FORCE HURRICANE HUNTER MISSION AGAIN
INDICATED A CONCENTRIC EYEWALL STRUCTURE. THE PEAK FLIGHT-LEVEL
WIND SPEED WAS 145 KT WITH AN SFMR SURFACE WIND MEASUREMENT OF 125
KT IN THE NORTHEAST QUADRANT. A GPS DROPSONDE MEASURED A SURFACE
WIND SPEED OF 133 KT IN THE SAME QUADRANT...BUT BASED ON
LOWER-LAYER AVERAGES FROM THE SONDE...THIS PROBABLY DOES NOT QUITE
CORRESPOND TO A 1-MINUTE AVERAGE SURFACE WIND. BASED ON ALL OF
THESE DATA THE CURRENT INTENSITY IS SET AT 130 KT."

My decoder does the following calculation...

If the SFMR surface wind is available and is not suspect and the flight level winds are not suspect, it will calculate the estimated 30 second average surface wind.

It does it by first finding the appropriate reduction factor for that particular region of the storm, specifically for that observation, not taking into account the pressure level or height of the aircraft.

The SFMR peak (10s Avg.) surface wind is divided by the peak (10 sec. Avg.) flight level wind to get the appropriate reduction factor of the peak wind at flight level to the peak surface wind.

The next step is to simply multiply that reduction factor by the flight level (30 sec. Avg.) wind.

The result, what should be a fairly accurate estimate of the 30 second sustained surface winds.

Take a look at those two observations from above...

The green one:
Flight Level Wind (30 sec. Avg.): 132 knots (~ 151.8 mph)
Peak (10 sec. Avg.) Flight Level Wind: 133 knots (~ 152.9 mph)
SFMR Peak (10s Avg.) Sfc. Wind: 125 knots (~ 143.8 mph)
Estimated Surface Wind (30 sec. Avg.) Using Estimated Reduction Factor: 124.1 knots (~ 142.7 mph) - Category Four Hurricane
Peak Wind at Flight Level to Surface Reduction Factor: 94.0%

The red one:
Flight Level Wind (30 sec. Avg.): 143 knots (~ 164.4 mph)
Peak (10 sec. Avg.) Flight Level Wind: 144 knots (~ 165.6 mph)
SFMR Peak (10s Avg.) Sfc. Wind: 100 knots (~ 115.0 mph)
Estimated Surface Wind (30 sec. Avg.) Using Estimated Reduction Factor: 99.3 knots (~ 114.2 mph) - Category Three Hurricane
Peak Wind at Flight Level to Surface Reduction Factor: 69.4%

What we notice is that despite the higher flight level winds in the red one, those winds do not translate to the surface. The green one has winds that do very closely match the surface winds. By taking a look at just the highest flight level winds, we have missed where the strongest surface winds are.

---

What we learn about the entire set of HDOB observations in the message above is that the first observation soon after passing the eye uses a reduction factor of 105.7% where tropical storm force winds are felt on the border of the eye and eyewall. Perhaps this is due to a stadium effect where the winds happen to be calmer up high rather than lower where the stadium like eye meets the ocean at a curve. Throughout the observations above, the reduction factor eventually decreases to 63.1% by the end of the observations above. The reduction factor has a second peak in the most intense part of the eye where the winds translate 94% to the surface. If you decode the message I have above you will see a perfect curve of wind. The curve steepens fast as the plane travels from the calm eye and passes through the core of highest winds. The winds then begin to more slowly level out all the way back to tropical storm force. Calculations are not performed for any data that is suspect or of course missing.

I wanted to bring this all up because the reduction factor changes throughout the storm. The peak wind at flight level is not necessarily where the highest surface winds are. In the two observations we saw above, the flight level winds were 11 knots lower in the green observation, but it turns out the surface winds were about 25 knots higher.

Perhaps someone will find it interesting.

---

Here you can see an image of the estimated surface wind using the method I describe above. The observation above is graphed:



---

A more detailed image:



Just shows what the north quadrant wind field was like after passage over Jamaica.

0 - center of eye
8 - border of center/eye wall where major hurricane force winds start
14 - highest surface wind, 124 knots
19 - major hurricane force winds stop
35 - hurricane force winds stop
112 - hurricane hunters stop flying north

Note how the highest surface winds do not match the locations where there was the highest flight level winds.

[MiamiensisWx]

You're the best expert I have seen who deals with reconnaissance data! Your site and converter has aided many people on Atlantic tropical cyclones. I have no idea as to why this outstanding discussion received little attention. Personally, I think it clears up many issues with regards to the semantics behind these obs. Additionally, it addresses the physics involved in the flight-level conversions and discrepancies between various methods. The data compilation is outstanding and invaluable for me. Thanks for this very informative post! I think it should be "stickied" for the newer members and others, many of whom asked questions.

[WindRunner]

I have some questions for you here, Chris.

First of all, your calculations seem to be a little redundant . . . follow me here for a second.

You take SFMR winds (SFW) and divide by FL winds (FLW) to get a reduction ratio to the surface . . . and then you multiply that reduction ratio by the FLW to determine the surface winds. I believe that is correct . . . the problem being as follows:

Code: Select all

SFW          FLW
---    *     ---      =     SFW
FLW           1


You will ALWAYS get a number close to the SFMR estimate, though slightly off because of rounding issues.

The only thing I see this taking care of is providing a fairly accurate surface estimate for areas with suspect SFMR readings . . . though, because of the method, it will not be able to show if there was a higher max wind than measured by the SFMR . . .
I think if you plot the actual measured SFMR winds on a graph like your second one, you will find that the graphs come out quite similar.


If I'm not following you correctly, please tell me . . . it seems like an interesting idea, I just don't think you have the calculations quite right . . .

[Chris_in_Tampa]

Thanks MiamiensisWx. I have a lot of other features I am working on as well. I eventually hope to work on a dropsonde and vortex decoder. Ultimately I would like to have a network link in Google Earth that plots all recon data of all types in real time, but if I do ever get that done, it will be at least next season.

To answer your question WindRunner, my site does the calculation a bit differently.

My site gets the reduction factor from the peak 10 second average flight level wind and the peak 10 second average SFMR surface level wind.

After that I multiply that reduction factor by the sustained 30 second average flight level wind to get the 30 second estimated sustained 30 second average surface winds.

I am comparing peak winds to peak winds rather than the 30 second sustained flight level winds and the peak SMFR peak surface winds. By comparing peak wind to peak wind I should get a more accurate estimate of the sustained surface winds since I am comparing the highest 10 second wind in the HDOB recording interval at both the surface and flight level.

For example I will show an example from an observation last night that was not suspect.

---

Observations:

Flight Level Wind (30 sec. Avg.): 114 knots (~ 131.1 mph)
Peak (10 sec. Avg.) Flight Level Wind: 121 knots (~ 139.1 mph)
SFMR Peak (10s Avg.) Sfc. Wind: 110 knots (~ 126.5 mph)

---

Calculations:

First...

SFMR Peak (10s Avg.) Sfc. Wind / Peak (10 sec. Avg.) Flight Level Wind = Peak Wind at Flight Level to Est. Surface Reduction Factor

110 / 121 = .909 (90.9% reduction factor)

Second...

Flight Level Wind (30 sec. Avg.) x Peak Wind at Flight Level to Est. Surface Reduction Factor = Estimated Surface Wind (30 sec. Avg.) Using Estimated Reduction Factor

114 x .909 = 103.6 knots

---

Result:

The SFMR instrument indicated a peak 10 second wind of 110 knots (~ 126.5 mph) but the estimated 30 second average surface wind might be closer to about 103.6 knots (~ 119.2 mph).

---

So the surface wind estimate will always be less than, or in a rare case equal to, the SFMR peak surface wind because my calculation gets the 30 second sustained surface wind.

This does of course mean that the SFMR winds have to be accurate. If the data is suspect, I now display it, but I include a red asterisk to denote suspect data. However, the NHC noted yesterday in the 5AM discussion the following:

"DEAN MADE LANDFALL ON THE EAST COAST OF THE YUCATAN PENINSULA NEAR
THE CRUISE SHIP PORT OF COSTA MAYA AROUND 0830 UTC...AND THE EYE IS
NOW JUST INLAND. OBSERVATIONS FROM AN AIR FORCE HURRICANE HUNTER
PLANE INDICATE THAT THE HURRICANE WAS INTENSIFYING RIGHT UP TO
LANDFALL. A PEAK FLIGHT-LEVEL WIND OF 165 KT WAS MEASURED JUST
NORTH OF THE EYE. MAXIMUM SURFACE WINDS FROM THE SFMR WERE 124
KT...BUT IT IS HIGHLY LIKELY THAT THE MAXIMUM SURFACE WIND SPEED
WAS NOT REPORTED BY THE SFMR INSTRUMENT. A GPS DROPSONDE IN THE
NORTHERN EYEWALL MEASURED A WIND SPEED OF 178 KT AVERAGED OVER THE
LOWEST 150 METERS OF THE SOUNDING. BASED ON THE DROPSONDE AND THE
FLIGHT-LEVEL WINDS...THE INTENSITY IS SET AT 145 KT. A DROPSONDE
IN THE EYE MEASURED A CENTRAL PRESSURE OF 906 MB JUST PRIOR TO
LANDFALL."

So if the SFMR instrument under reported winds or they are saying that it missed the highest surface winds perhaps, then my decoder will under report the highest estimated sustained winds. But that takes analysis and other tools from the NHC to determine. (It would be nice if it were this easy to determine the actual estimated wind speed.) I am going to add this SFMR note to my decoder.

I also want to note that the HDOB message contains 30 second sustained wind readings, not 1 minute sustained, so my decoder, if SFMR instruments are absolutely correct, may actually read slightly high at some point. I could average the sustained winds of two observations in the message, but I think people can eye ball it just as easy to see what the 1 minute sustained might possible be. (Add two estimated surface wind observations in two 30 second intervals right next to each other and divide by 2)

[WindRunner]

Ahh, OK, that makes more sense. You get cconversion factors from comparing the peak wind readings and translate that to the 30-sec sustained and use the conversion factor on that group . . . simple, but very useful! Great idea, and thanks for explaining it to me!