Evaluating WRF Model Boundary Layer Structure Using Radar Remote Sensing Data to Understand Hurricane Intensification

The main objective of the current research is to understand the dynamics of hurricane intensification in observations and numerical models for applications related to forecasting. The boundary layer is a critical region of the hurricane where we need to characterize the sub-grid scale fluxes of mass, momentum and energy. The boundary layer is also connected to the formation of deep convection, which is the source of energy input to the system. All turbulent fluxes are parameterized in mesoscale models, and they need to be evaluated and improved to enable improved understanding and forecasting. The WRF model is an open source, community based numerical weather prediction system. We have conducted WRF simulations of Hurricane Dorian (2019) with 1 - 2 km horizontal and 150 m vertical spacing to evaluate the boundary layer and convective structure during intensification using two different boundary layer schemes. The evaluation is performed primarily with radar remote sensing measurements from the Imaging Wind and Rain Airborne Profiler (IWRAP) that have horizontal and vertical spacing of 125 m and 30 m, respectively.
Two modes of variability are present in nonlinear dynamical systems such as hurricanes: deterministic and stochastic. To evaluate the deterministic mode, individual cross sections of the boundary layer and convective structure are compared between the model and observations. The comparisons were obtained after filtering the radar remote sensing data to the minimum resolvable wavelength scale of the model. To evaluate the stochastic mode, statistical analysis in terms of scatter plots, bias, correlation functions and modal decomposition analysis is performed. The stochastic mode of variability showed that when averaging over space and time, the WRF model underpredicts the boundary layer wind speed by ~ 12 m/s with a standard deviation of 9.5 m/s. This result provides further evidence that the WRF model is overly dissipative by one to two hurricane categories. The convective structure is also dissipative with a highly tilted structure and weak updrafts. The deterministic results agreed with the stochastic findings and reveal details of the structural differences. The next step is to examine the combined numerical and physical diffusion effects and their contribution to the bias.