The Predictability of Atmospheric Conditions Relevant for Tropical Cyclone Activity on Subseasonal Timescales

The Model for Prediction Across Scales (MPAS) with 15 km uniform grid spacing is used to explore potential opportunity to predict TC activity out to 5 weeks. First, an MPAS climatology for September TC activity is established. Next, the simulations resulting in the most and least active month are analyzed in further detail to understand why those model simulations predicted an active or inactive September. The total number of TCs, TC days, accumulated cyclone energy (ACE), and the track density are each evaluated relative to observations. It is shown that MPAS simulations can reproduce the year-to-year variability in North Atlantic TC activity for the month of September and have skill in predicting an active or inactive month, largely related to ENSO. Finally, a series of identical twin experiments are conducted to examine the intrinsic predictability of both TC activity and large-scale conditions relevant to TC activity. It is shown that errors of weekly averaged fields relevant for TC activity, such as vertical wind shear and atmospheric moisture, have predictability beyond the classic 2-week weather predictability limit.