180 likes | 191 Views
This paper discusses the transition of the HWRF (Hurricane Weather Research Forecasting) model from version 2 to version 3 in the EMC A Joint Hurricane Testbed (JHT) program. It focuses on improvements in intensity forecasts as well as upgrades to the land model and landfall prediction.
E N D
Transitioning HWRF upgrades into operations at EMC A Joint Hurricane Testbed (JHT) Program Robert E. Tuleya*, Yihua Wu, VijayTallapragada, Young Kwon, Hyun-Sook Kim,Zhan Zhang, Qingfu Liu,J. O’Connor 65th Interdepartmental Hurricane Conference Miami Fl March 2011 *NOAA Visiting Scientist @CCPO/ODU
JHT project task areas* • Improve HWRF intensity forecasts transition surface flux formulation from V2 to V3 • Trouble shoot and diagnose HWRF problems transition surface flux formulation from V2 to V3 • Upgrade land model and landfall prediction transition to NOAH LSM *½ time effort
Transition from HWRF 2010 (V2) Operational Sfc Fluxes to Alternative znot Formulation(V3)
HWRF 2010 V2 surface fluxes based on observation 2010 Results mixed: ~good track but low intensity bias HEXOS data (1996, Decosmo et al) CBLAST data (2007) HWRF 2010 V2 used constant CH at high winds
Similarity relationship for surface layer exchange coefficients under neutral condition Stress ~ u*2 ~ Cdumum Enthalpy flux ~ u*.q*~ Ch um ( qm – qs ) • HWRF 2010 prod sfc flux formulation uses Cd & Ch with low level wind cut-off above which there is no stability dependence • Given Cd & Ch from HWRF 2010 operational code(Kwon), solve for zo and zot for neutral conditions as function of low level wind • Use function form of zo and zot in MO formulation of GFDL surface flux formulationAssume stability dependence can be important in some cases • Method allows for roughness to be changed later based on more physical basis—e.g. wave coupling, current interaction, etc. • Alternative znot formulation used in V3-R2
Possible differences in results between HWRF prod and znot formulation • Znot formulation does not exactly match CD & CH of Kwon. Algebraic fit of znot for enthalpy and momemtum • Znot formulation has stability dependence of GFDL(HWRF 2009) sfc flux routine
HWRF 2010,V3-R2 (neutral)CD,CHCH/CD ratio HWRF V3 HWRF 2010 CH/CD ratio significantly lower than GFDL & HWRF 2009
H210 test with znot formulation based on H210 CH~1x10-3 above 10m/s V3 znot HWRF 2010 *** Note track nearly identical
Red: Oper. HWRF V2.0 Cyan: HWRFV3.2 w/POM Benchmarking HWRFV3.2–POM three season testing, ATLANTIC 2008-2009-2010. HWRFV3.2 produced nearly identical results compared to operational 2010 HWRF. HWRFV3.2 produced slightly less bias compared to operational 2010 HWRF. Consistent with slightly greater values of CH/CD for HWRFV3.2 vs 2010 HWRF
Summary of Znot formulation method • Relatively small difference between operational HWRF V2 and HWRF V3 znot formulation method • Some small improvement in reduction of low intensity bias? • Thermal znot can be reformulated to account for low intensity bias?? (next talk)
The NOAH LSM Issues in HWRF ~150 historic 2008 Atlantic cases ~280 2010 Atlantic cases run in parallel
Track NOAH LSM test cases 2010 AL01-AL19 2008 cases Track: Noah LSM ~20nm worse @96h Why??? Intensity: Noah LSM slightly better up to 96h (not shown)
Example of Alex (2010)two problems areas Hot spots in LST in parent & nest domains Apparent lateral BC problems
Hot spots (parent grid) in Noah LSM NOAH LSM GFDL slab LST > 330K 280<LST<310K
Lateral BC problems (nest): V3R2 makes problems worse HWRFV2 NOAH LSM V3R2 NOAH LSM Apparent lateral BC problems with NOAH LSM (H210) LSThot spot >500K
Apply patch to fix Tsfc values along perimeter after LSM call, hot spots reduced/removed
Hot Spots controlled by • patching LST from one point inside • changing from binary to netcdf!! • turning off gravity wave drag!! Issues and Solutions • Tsfc treated differently than other prognostic variables (e.g. u,v,T,R ???) • Apparent lateral BC problems remain for LST and other surface land parameters and variables in nest domain • Fix LSM issues in V3.2 and run in parallel for 2011