Andrea Schumacher, CIRA/CSU, Fort Collins, CO Mark DeMaria and John Knaff, NOAA/NESDIS/StAR, Fort Collins, CO NCAR/NOAA/CSU Tropical Cyclone Workshop 16 November 2011 CIRA, Fort Collins, CO

 Maximum potential intensity (MPI)  Upper bound to TC intensity  Important quantity to know when predicting TC intensity  Approaches  Empirical  Theory  Modeling (not discussed here)

 Atlantic (DeMaria and Kaplan 1994)  Clim SST/Vmax sample from  Fit upper bound of Vmax (least squares, exponential)  A = 28.2, B = 55.8, C = ,T 0 = 30.0 °C  T in °C  V given in ms-1  E. Pacific (Whitney and Hobgood 1997)  Clim SST/Vmax sample from  Fit upper bound of Vmax (least squares, linear)  C 0 = ms -1, C 1 = 5.36 ms -1  SST in °C  EPMPI given in ms -1

 Two main thermodynamic theories  Carnot heat engine (Emanuel 1986; 1991) ▪ Acquire enthalpy from the ocean ▪ Convert to wind energy ▪ Function of sea surface and storm outflow temperature ▪ Calculates MPI as a maximum surface wind speed  Eye subsidence (Miller 1958 and Holland 1996) ▪ Requires SST, atmospheric sounding and surface pressure ▪ Calculates MPI as a minimum central pressure

 Current version of Statistical Hurricane Intensity Prediction Scheme (SHIPS) uses empirical MPI as a predictor  Pros: ▪ Availability (function of SST only)  Cons: ▪ Small sample set – 31 years ▪ Doesn’t include atmospheric variables ▪ Only uses SST – no ocean structure (e.g., OHC) ▪ Doesn’t always serve as true “upper bound” to Vmax…

 Sometimes, Vmax > theoretical MPI  Example: Hurricane Celia 2010

 Re-examine empirical formulas using SHIPS sample set – valid?  Examine Emanuel’s theoretical MPI (1995) over the SHIPS dataset  Identify when valid / not valid  Create a combined theoretical / empirical formula for MPI in SHIPS

10653 cases total Vmax > MPI for 30 cases (0.3%)  Good upper bound for sample cases total Vmax > MPI for 8 cases (0.1%)  Good upper bound for sample

First, try a few simple improvements... Vmax > MPI 12% of cases Vmax > MPI 16% of cases *Note: Atmospheric sounding comes from GFS model.

 Vmax includes asymmetric adjustment, MPI does not  For proper comparison, remove asymmetric component from storm motion (Schewerdt): Vmax(adjusted) = Vmax – 1.5*(TC speed) 0.63 TC Motion Symmetric Vortex Asymmetric Vortex

Vmax > MPI 10% of cases Vmax > MPI 14% of cases

 Emanuel MPI assumes 700-mb winds reduced by factor of 0.8 at surface (friction)  Franklin et al. (2002) used dropwinsonde data to show reduction factor is closer to 0.9  Multiplied Emanuel MPI by 0.9/0.8

Vmax > MPI 9% of cases Vmax > MPI 11% of cases

Observed Vmax Exceeds Theoretical MPI. Why? When?

“Violators” have…  Higher intensity, weakening  Lower MPI, decreasing  Lower SST, decreasing  Larger low-level temperature gradient  Lower OHC  Higher latitude Average (t=0)Avg 12-hr Change V > MPIMPI > VV > MPIMPI > V VMAX MPI EMPI RSST TGRD TADV T SHDC RHCN EPOS V V V20C TSPD TSPY LAT

 Empirical MPI provides good upper bound for intensity  SHIPS parameter based on Emanuel theoretical MPI improved by  Adding asymmetry adjustment due to storm motion  Using more appropriate reduction factor when converting 700-mb winds to surface

 Objectively identifying conditions where Emanuel MPI should be adjusted or replaced with empirical  Fast-moving TCs moving over cooler SST/OHC, lack of equilibrium  Extratropical cases (Atlantic)  Annular cases (E Pacific)  Creating combination empirical/theoretical MPI for testing in SHIPS  Parameterizing SST change under inner core  Decrease MPI