On the Predictability of Hurricane Track and Intensity Using an OSSE Framework

A regional OSSE framework was developed at NOAA/AOML and University of Miami and features a high-resolution regional nature run embedded within a lower-resolution global nature run. Simulated observations are generated and provided to a data assimilation scheme, which produces analyses for a high-resolution regional forecast model. Our nature run was generated using a high-resolution, 27-km regional domain with 9/3/1 km storm-following nests (WRF-ARW, v3.2.1) embedded in a low-resolution T511  ECMWF global nature run.  For data assimilation we use NOAA's GSI, which is a three-dimensional variational data assimilation scheme. The analyses are performed in the parent domain at 9-km resolution. Our forecast model is the 2014 operational Hurricane-WRF model. The parent domain has a 9 km grid spacing, while the single storm-following nest has a 3 km grid spacing. Data assimilation and model cycling were performed every 6 hours, with each analysis used to initialize a 5-day deterministic forecast, for a total of 16 cycles. A reference data assimilation cycle is performed using perfect data as observation sampled at 27 and 9 km as well as perfect lateral boundary conditions. This is followed by multiple experimental assimilations in which the initial conditions are degraded gradually in terms of resolution, data coverage, and variables assimilated. Additional experiments also include degraded lateral boundary conditions. The impacts from these degraded assimilation experiments on the forecast of track, intensity, and storm structure will be presented to discuss the predictability characteristics of the simulated hurricane.