1. Accounting for Variations in TC SizeBy
John Knaff

2. Idealized Structure
From Holland (1987)

3. Problem Many TC applications make a one-size assumption
This potentially leads to biases as a function of size
e.g. Dvorak Technique
e.g., inland decay
e.g., surface to flight-level wind reductions
Scaling by the radius of maximum (RMW) wind is limited to near-core applications and the RMW is very difficult to accurately estimate without an eye structure or aircraft observations.

4. Examples of TC Size Variations
Very Large
Katrina (2005), 08/28/18UTC, 155 kt
Very Small Felix (2007), 09/03/03UTC, 150 kt

5. Proposal Objectively scale the tropical cyclone Separate
By infrared (IR) measures of TC size
Separate
Inner core/Eyewall&inner Rainband region
Outer Rainband /Environmental Interface Region
Determine if such scaling improves statistical relationships for
Intensity forecasting
Intensity diagnostics
Data assimilation (possibly)

6. Review IR-Based TC Metric (R5)
Create Azimuthal
Averages
V500= ∗ sin |φ |−1.350∗PC ∗PC ∗PC3
V500 ≡ 850-hPa mean tangential wind at r= 500 km induced by the TC
R5=( R5 + V500−V500c ∗ 500 V500c−V1000c ),
R5 ≡ the radius of where the TC wind field is indistinguishable from the background flow in a climatological environment.
Latitude +
First 3
Principle Components

7. Using Observed R5 to Spatially Scale TCs
R5 Scaling Factor: 𝐹 𝑅5 = 𝑅5 𝑅5 𝑐 ( 𝑣 𝑚 ) 𝐹 𝑅5 ≡𝑆𝑐𝑎𝑙𝑖𝑛𝑔 𝐹𝑎𝑐𝑡𝑜𝑟 𝑅5 𝑐 ≡𝑚𝑒𝑎𝑛 𝑅5 𝑎𝑠 𝑎 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑖𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 𝑣 𝑚 ≡𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒 𝑜𝑓 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑤𝑖𝑛𝑑𝑠
R5 as a function of intensity

8. Scaling Methodology 𝑟 𝑠𝑐𝑎𝑙𝑒𝑑 = 𝑟 𝐹 𝑅5
𝑟 𝑠𝑐𝑎𝑙𝑒𝑑 = 𝑟 𝐹 𝑅5
Small (Large) storms have 𝑟 𝑠𝑐𝑎𝑙𝑒𝑑 values that are greater (less than) 𝑟.

9. Scaling via R5
Very Large
Very Small

10. IR Ships/RII Predictors
IR00: Predictors from GOES data (not time dependent). The 17 values in this record are as follows: 1) Time (hr*10) of the GOES image, relative to this case 2) Average GOES ch 4 brightness temp (deg C *10), r=0-200 km 3) Stan. Dev. of GOES BT (deg C*10), r=0-200 km 4) Same as 2) for r= km 5) Same as 3) for r= km 6) Percent area r= km of GOES ch 4 BT < -10 C 7) Same as 6 for BT < -20 C 8) Same as 6 for BT < -30 C 9) Same as 6 for BT < -40 C 10) Same as 6 for BT < -50 C 11) Same as 6 for BT < -60 C 12) max BT from 0 to 30 km radius (deg C*10) 13) avg BT from 0 to 30 km radius (deg C*10) 14) radius of max BT (km) 15) min BT from 20 to 120 km radius (deg C*10) 16) avg BT from 20 to 120 km radius (deg C*10) 17) radius of min BT (km)

11. Straight Scaling Dependent Results
Atlantic
East Pacific
SHIPS (22 predictors)
Degradation
RII
Improvement
LGEM (19 predictors)
SHIPS
Degradation
RII
LGEM (19 predictors)
Work in progress

12. Next Step: Lightning Predictors in RII
Inner-core, especially eyewall lightning is often associated with
Nearing peak intensity
Strong vertical wind shear
Increased rainband lightning is often associated with
Intensification
Note GOES-R will have a continuous lightning mapper

13. Next Step: Address Intensity Biases
Dvorak Size-Related biases
Microwave Sounder–based biases
Similar biases exist with small TCs being underestimated and Larger TC being overestimated

14. Next Step: Application to Pressure-Wind Relationship Applied to Vmax estimates
Current
Proposed
Courtney & Knaff (2009) for pressure estimate
Relies on ATCF/TC vitals
R34 (Size)
POCI (environmental P)
Motion
Latitude
* Sometimes very subjective and/or wrong.
Courtney & Knaff (2009) for pressure estimate
ATCF/TC vitals
POCI
Motion
Latitude
V500 estimate from IR image.