Microwave Imager TC Applications Jeff Hawkins Kim Richardson

Microwave Imager TC Applications Jeff Hawkins, Kim Richardson, Mindy Surratt, Tom Lee, Rich Bankert, Joe Turk 2, Charles Sampson, Jeremy Solbrig, Arunas Kuciauskas, John Kent 3, Naval Research Laboratory, Monterey, CA 2 Jet Propulsion Laboratory, Pasadena, CA 3 Science Applications Inc. International, Monterey, CA International Workshop on Satellite Analysis of Tropical Cyclones (IWSATC) Honolulu, HI April 15, 2010 1

Apparent LLCC True LLCC Exposed LLCC Sheared Convection 2

Good central mass Remove doubt Large rainfree eye Rainband structure 3

http: //www. nrlmry. navy. mil/TC. html Scatterometer & Cloud. Sat Vis/IR imagery suite Storm Basins & Names Microwave imager/sounder product suite Automated Tropical Cyclone Forecasting (ATCF) System warning graphic Latest 1 -km Visible/IR imagery (GEO/LEO) 30 minute MTSAT refresh with AVHRR/OLS as available FNMOC: https: //www. fnmoc. navy. mil/tcweb/cgi-bin/tc_home. cgi 4

Microwave Imager Temporal Changes 512 km Katrina 5

Microwave Imager Temporal Changes 14 deg Tropical Cyclone Yasi 6

Microwave Imager Temporal Changes 512 KM Concentric Mode 1: Evolve into one larger diameter eye 7

Satellite – Aircraft Comparisons NOAA 43 14301520 Z 10, 000’ NW North Flt level wind TMI Tb TMI 85 H 9 -22 -05 1443 Z 8

COMET Module – TC Microwave http: //meted. ucar. edu/npoess/ocean_winds/index. htm TC specific focus with microwave sensors Courtesy: Tom Lee 9

Microwave Imager Training http: //www. nrlmry. navy. mil/training-bin/training. cgi 10

Automating TC Intensity? ? ? Huge-Intense Wrapping? Shear - Weak Small-Intense Annular - Intense Wrapping 11

Microwave Imager TC Intensity Estimation Microwave imager data provides structural characteristics not always found in typical Vis/IR imagery. GOES Visible SSM/I 85 GHz Image feature extraction Feature selection to reduce redundant and irrelevant features Segmented 85 GHz Courtesy: Rich Bankert Machine Learning Application Leave-One(TC)-Out Cross Validation Training and Testing Atlantic Basin Data Set **319 samples from 60 TC’s** RMSE: 13. 1 kts 12

Microwave Imager Top Impact “Features” 1. Symmetry measure – based on the 2 -deg radius gradient vector angles relation to center 2. Average % encirclement of 1 -km wide rings within 1 -deg radius area (pixels < 253 K) 3. Difference in Tb of warmest center pixel and coldest surrounding pixel 4. Summation of pixel Tb in the SE quadrant of the “eye” region 5. Average of the maximum Tb on each ring in the 1 -degree radius area 6. Area coverage of pixels with less than 228 K Tb 7. Number of 1 -km rings (within 1 -deg radius) with at least 33% pixels with < 253 K Tb 8. Average % encirclement of the 1 -km wide rings within 1 -2 -degree radius (pixels < 253 K) 9. Maximum summation of pixel Tb values along a 1 -km ring for pixels with < 228 K Tb 10. Number of 1 -km rings (within 1 -deg radius) with at least 50% pixels with Tb < 253 K 13

Ritchie: Gradient Vector Feature Artificial Vortex Gradient Detail IR Image Gradient Detail 14

TC Intensity – Microwave Imager Automated Microwave Imager TC Intensity Estimates 319 Atlantic basin samples [1995 -2005] RMSE (kts) Features computed from SSM/I 85 GHz channel data • Cubist (machine learning tool) Cross Validation (CCV) - all original features 21. 0 20. 0 19. 0 18. 0 17. 0 16. 0 15. 0 • CCV - new feature set (modified original, new features added) • Manual. CCV - feature selection (remove redundant and irrelevant features) • estimation (Jeff) • Manual estimation (Jeff)added to total set, re-do feature selection • CCV - new features • CCV - 277 samples - high shear samples removed, re-do feature selection (Jeff RMSE – 16. 7 kts) 14. 0 13. 0 • CCV - 319 samples – add gradient vector (Ritchie) feature 12. 0 TIMELINE 15

Problematic TC Structure Cases SSM/I 85 GHz images (BT) High Shear Center Convection Dying Inner Eye 16

TC intensity – Microwave Imager (2006 -2009 Data Sets) Adding Near Real-Time Data Sets from NRL-TC Web Page Includes AMSR-E, TMI, SSMIS in addition to original SSM/I 17

TC intensity – Microwave Imager (Adding 37 GHz Data) Tropical Cyclone Yasi (11 P) 85 GHz H-pol 37 GHz H-pol Typhoon Sinlaku (15 W) 37 GHz H-pol 85 GHz H-pol 18

Microwave Channel Intercalibration 85 GHz H-pol F-15 SSM/I 0723 Z 89 GHz H-pol Aqua AMSR-E 0332 Z 91 GHz H-pol F-17 SSMIS 0750 Z Colder Tb at higher frequencies imply stronger storm: Incorrect 19

Passive Microwave Channel Intercalibration Process Hurricane Bonnie TB Simulations TMI 85 GHz H 8 -10 K Differences In Deep Convection TMI 85 GHz H - AMSR-E 89 GHz H Huge potential to create issues: - Qualitative assessments - Quantitative analyses (intensity alg) 20

Passive Microwave Imager Missions SSM/I Satellite sensors soon to cease functioning TRMM TMI AMSR-E AMSR WINDSAT SSMIS FY-3 MWRI Russia MTVZA Megha Tropiques MADRAS GCOM AMSR GPM DWSS-1 YEAR 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 Launches Primary mission Extended mission Future March 2011 Hawkins 21