Applications of JPSS Imagers and Sounders to Tropical Cyclone Track and Intensity Forecasting Team Lead: Mark DeMaria NOAA/NESDIS/STAR Fort Collins, CO Team Members: Galina Chirokova, Robert DeMaria, Steve Miller, CIRA/CSU Jack Beven , NOAA/NHC Chris Velden, Tony Wimmers, UW/CIMSS Briefing to the JPSS Program Office January 25th, 2013

Project Overview Two basic methods for improving tropical cyclone forecasts with S-NPP Assimilate data in numerical forecast models F. Weng and J. Li JPSS projects Improve analysis and statistical post-processing forecast products This project focusses on method 2 Multi-spectral center fix algorithm ATMS, CrISS, VIIRS Maximum intensity estimation method Generalization of operational AMSU methods Improve statistical-dynamical intensity forecasts using ATMS/CrISS retrievals

1. Multi-Spectral Center Fix Algorithm Aircraft reconnaissance only available for about 30% of Atlantic tropical storm forecasts Center fix is usually the first step in the forecast process Accurate center estimate impacts all downstream forecasts Better satellite intensity estimates Better numerical model forecasts

Operational Center Fix Methods Center Location = surface center Center of circulation Lowest sea-level pressure Visible and IR methods – Dvorak Eye Parallax Distinct and inferred center with shear pattern and low-level clouds Spiral bands and curved cloud lines Wedge method Using animation Low-level cloud motions Deep layer cloud motions Ignore cirrus layer cloud motions Mid-level centers tilted from surface center Using microwave images Thick cirrus clouds in visible and IR images obscure features below, used for center location Thick cirrus clouds in microwave images are more transparent, and the microwave images may often provide better views of features, for improved center locations Using 3.9-micrometer images at night Proxy for visible CIMSS developed automated ARCHER method to fit spiral patterns to microwave imagery

New JPSS Center Fix Algorithm Start with ATMS (and maybe CrISS) T,q retrievals Use hydrostatic and nonlinear balance equations to diagnose geopotential height and wind fields Estimate center from Z and wind fields Refine using VIIRS IR, vis and DNB imagery Initial testing with AMSU retrievals and AVHRR data ATMS/VIIRS dataset also being collected MIRS ATMS retrievals from K. Garrett

Hydrostatic Balance Ptop Ztop R = ideal gas constant dp/dz = -ρg p = pressure z = height p= ρRTv ρ = density P Z g = gravity  dp/p = -(g/RTv)dz Tv = virtual temperature Ptop Ztop R = ideal gas constant Z = Z(x,y,P) Given T, RH retrieval, Tv can be used to provide geopotential height (Z) on pressure levels

Pressure-Wind Relationships Hydrostatic integration and ideal gas law give gZ = Φ(x,y,P) Approximate form of horizontal momentum equations provides horizontal wind estimates Symmetric flow – gradient wind V2/r + fV = ∂Φ/∂r Asymmetric flow – Nonlinear balance equation from soundings, u,v = horizontal components of non-divergent wind

Initial Test of Sounder Center Fix Algorithm Use AMSU temperature retrievals 2006-2011 Atlantic Sample 2021 Cases Hydrostatic/nonlinear balance Z, winds Simple machine learning algorithms tested LDA, QDA Provides estimate of most likely center location Provides estimate of the importance of predictors Next step is refinement with vis, IR, DNB

Tropical Storm Gordon Example Aug 17 2012 16 UTC 925 hPa geopotential height 925 hPa nonlinear balance winds AVHRR Visible AVHRR IR Window Channel

Contributions to Center Fix Algorithm

VIIRS Imagery and Sounder Z

ATMS and AMSU Temperature Retrievals for Hurricane Sandy AMSU ATMS

2. Operational AMSU Tropical Cyclone Intensity Estimation Products CIRA AMSU intensity and wind structure estimation Hydrostatic integration of AMSU soundings to give Pmin and Vmax Statistical bias correction Also provides radii of 34, 50 and 64 kt winds Transitioned to NCEP operations in 2005 CIMSS AMSU intensity estimation Uses 4 channels sensitive to upper level warm core Eye size parameter (IR data or ATCF) to account for resolution variations Run in real time at CIMSS, provided to NHC in real time

Conversion of AMSU Algorithms to ATMS CIRA Supported by PSDI/NDE project K. Garrett providing MIRS ATMS retrievals to statistically adjust CIRA AMSU algorithm CIMSS Supported by JPSS-PGRR project Radiative transfer model being used to adjust warm core-intensity relationship

Evaluation of Min Pressure-Warm Core Relationships

3. Operational Atlantic Intensity Forecast Model Errors (2008-2012) HWRF, GFDL are regional coupled ocean/atmosphere models DSHIPS, LGEM are statistical-dynamical models

Logistic Growth Equation Model (LGEM) dV/dt = V - (V/Vmpi)nV (A) (B) Term A: Growth term, related to shear, structure, etc Term B: Upper limit on growth as storm approaches its maximum potential intensity (Vmpi) LGEM Parameters: (t) Growth rate (from shear, instability, etc)  MPI relaxation rate (constant) Vmpi(t) MPI (from SST and sounding) n “Steepness” parameter (constant) LGEM might be improved by estimating Vmpi(0) and instability contribution to (0) from ATMS/CrIS soundings.

Maximum Potential Intensity Theory (Emanuel 1988, Bister and Emanuel 1998) Ts, To, k*, k can be estimated from the SST and a sounding. Ck/CD= specified ratio of surface exchange coefficients

Maximum Potential Intensity Estimation in Irene’s Environment

Summary Satellite T/RH soundings used to estimate tropical cyclone wind field using hydrostatic and nonlinear balance ATMS soundings better resolve warm core due to increased horizontal resolution and wider swath than AMSU Multispectral center fixing algorithm combines ATMS and VIIRS input CIMSS and CIRA AMSU intensity algorithms being transitioned to ATMS ATMS/CrISS soundings have potential to improve statistical dynamical intensity forecast models VIIRS imagery to be demonstrated in 2013 NHC Proving Ground