Xu Lu- Oct 21

School of Meteorology (Defense) Improving High-Resolution Tropical Cyclone Prediction using a Cycled, GSI-Based Hybrid Ensemble-Variational Data Assimilation System for HWRF with Vortex Scale Observations Xu Lu Monday, October 21st 2:00 pm/NWC 5600   A GSI-based, continuously cycled, dual-resolution hybrid EnKF-Var DA system is developed for the HWRF Model.  The newly

Start

October 21, 2019 - 2:00 pm

End

October 21, 2019 - 3:00 pm

Address

120 David L Boren Blvd, Norman, OK 73072   View map

School of Meteorology (Defense)

Improving High-Resolution Tropical Cyclone Prediction using a Cycled, GSI-Based Hybrid Ensemble-Variational Data Assimilation System for HWRF with Vortex Scale Observations

Xu Lu

Monday, October 21st

2:00 pm/NWC 5600

 

A GSI-based, continuously cycled, dual-resolution hybrid EnKF-Var DA system is developed for the HWRF Model.  The newly developed DA system is then used in this dissertation to address a few scientific and technical challenges in assimilating the vortex-scale observations to improve the numerical prediction of TCs.

 

The first part first describes the newly developed data assimilation system and then addresses how various data assimilation configurations impact the vortex scale observation assimilation and the subsequent prediction.  It is found that dual-resolution hybrid DA improves the analyzed storm structure and short-term Vmax and MSLP forecasts compared to coarser, single-resolution hybrid DA.  Additionally, applying Vortex Relocation (VR) and Vortex Modification (VM) on the control background before DA improves the analyzed storm, overall track, RMW, MSLP, and Vmax forecasts.  Further applying VR on the ensemble background improves the analyzed storm and forecast biases for MSLP and Vmax.  Also, using 4DEnVar to assimilate TDR data improves the analyzed storm and short-term MSLP and Vmax forecasts compared to 3DEnVar.  Finally, a diagnostic on why advanced DA can improve the TC intensity forecast for Edouard (2014) is provided.

 

In the second part, using the newly developed DA system, the deficiency of the numerical model physics was discussed.  Although the DA produces realistic 3D analyses to initialize the model, persistent Vmax spin-down is found during the rapid intensification of hurricane Patricia (2015).  Diagnostics reveal that the spin-down issue is likely attributed to the deficient HWRF model physics which are unable to maintain the realistic 3D structures from the DA analysis.  The horizontal diffusion is too strong to maintain the realistically observed vertical oscillation of radial wind near the eyewall region.  The vertical diffusion profile cannot produce a sufficiently strong secondary circulation connecting the realistically elevated upper-level outflow produced in the DA analysis.  Further investigations with different model physics parameterizations demonstrate that spin-down can be alleviated by modifying model physics.  Additional experiments show that the peak simulated intensity and RI rate can be further improved by increasing the model resolution.

 

The last part of the dissertation explores the relative impact of various vortex scale observations on the analysis and prediction of hurricane Patricia (2015).  Results show that (1) assimilating the operational observations slightly improves the structural analysis and forecast, but overestimates the TC movement speed.  (2) Solely assimilating single-level wind observations from either CIMSS AMV or Stepped Frequency Microwave Radiometer (SFMR) primarily improve the analyses near the observation levels.  (3) Assimilating flight-level observations produces better inner-core structures and the improvements last longer in forecasts.  (4) Assimilating 3D observations like Tail Doppler Radar (TDR) or High Definition Sounding System (HDSS) dropsonde observations can produce even better structural analyses than the aforementioned experiments, and the structural improvements are better maintained during forecasts.  (5) HDSS_Only outperforms TDR_Only with additionally better analyzed and predicted outflow thermodynamic inner-core structures.   (6) Combining all observations complementarily improves all aspects of the TC structure in both the analysis and forecast.  Unlike the structural improvements which last the entire forecast, the intensity forecast improvement is usually found for the first several hours and for the timing of peak intensity.