Numerical Simulations of Hurricane Harvey with Data Assimilation and Cloud Initiation

A momentous hurricane, Harvey (2017), evolved over the Gulf of Mexico and greatly impacted Texas and Louisiana with strong winds and historical rainfall amounts from 26 to 30 August 2017. The hurricane is studied with emphasis on its microphysical characteristics and their influences on the hurricane evolution. In this paper, a sequence of high-resolution mesoscale simulations using an advanced research version of the Weather Research and Forecasting (WRF ARW) model and a three-dimensional variational (3D-Var) technique with the NCEP Gridpoint Statistical Interpolation (GSI) system are performed. Data assimilation experiments are conducted with conventional observations, airborne tail Doppler radar (TDR) and ground-based WSR-88D radial velocities and reflectivities. It is found that the accurate simulations of Harvey’s track, structure and rainfall before and after landfall are very sensitive to the types of observations integrated. Specifically, the assimilation of radar reflectivities provides the most significant improvements in track and intensity forecasts against the control simulation. Cloud analysis with reflectivity data, along with a digital filter initiation (DFI) scheme is then conducted for Harvey’s evolution period to examine the microphysical characteristics and other controlling factors that cause the massive rainfall and cloud characteristics.

Further diagnoses are made with a hybrid ensemble Kalman filter–three-dimensional variational (3D-Var/EnKF) scheme with assimilation of conventional, radar radial velocity and reflectivity observations. The impact of integration of these observations on the numerical simulations of Harvey’s track and subsequent precipitation are assessed and compared against the simulation results with 3D-Var and DFI techniques.