Tropical cyclone (TC) genesis is one of the most complex problems unresolved in TC prediction today. The formation and development of a TC is influenced by both the large-scale environment and small scale moist convection as well as interaction of the multi-scale features. The lack of in situ observations in the tropics contributes to slow progress in both understanding and prediction of TC genesis. The goal of this study is to better understand factors controlling TC genesis and develop potential parameters for TC genesis prediction.

To determine the characteristics of TC genesis in numerical models and observations, the 5th Generation PSU-NCAR Mesoscale Model (MM5), the TRMM Microwave Imager (TMI) and the Geostationary Operational Environmental Satellite Infrared Imager (GOES-IR) data are used in the study. TC genesis (i.e. the formation of a tropical depression) is defined based on a multi-criteria method in this study as 1) a closed circulation at the surface with wind speeds > 25-30 knots and 2) a closed isobar of sea level pressure (SLP) with persistent rain near the center. The MM5 forecasts of Hurricanes Katrina and Ophelia (2005) were made during the Hurricane Rainband and Intensity Change Experiment (RAINEX) with various horizontal resolutions using nested grids. The outer most model domain output with 15-km resolution is used. The lateral and initial condition for MM5 are from the global models including the CMC, GFDL, GFS, and NOGAPS. The model forecasts center of minimum SLP, which is tracked using 3-5 day forecasts at 3-hourly intervals. Genesis forecasts are separated into three categories—hits, misses, and overestimates—depending on whether the model forecasted genesis within 12 hours of the best track (hit), more than 12 hours after the best track (miss), or more than 12 hours before the best track (overestimate).

Subsequently, the model forecasts are compared with rain rate (RR) and cluster element (CE) data from TMI and GOES-IR to examine the characteristics of convection in the pre-genesis environment. Statistics of RR from TMI are taken within a 12 X 12 degree box centered at the location of genesis for each of the four days prior to genesis. The RR data are then divided into six categories of cumulative RR from > 0 mm/hr to >12.5 mm/hr (representing intensity of convection). Statistics of the area and count of CE's, which are cold cloud tops (< 218 K) with areas greater than 50 km, from GOES-IR are taken within the same sampling box. CE data is useful in tracking the growth and coalescence of convective cells associated with TC genesis.

The MM5 RR forecasts show a persistent mesoscale convective event around the location of TC genesis over a 24h or longer period. An interesting finding is that TMI and GOES-IR observations indicate, in general, a dramatic increase in RR and CE areas around two days before genesis, suggesting a convective blowup. More quantitative analysis of the MM5 RR forecasts will be conducted to examine whether the model captures these trends depicted in the observations and if the timing of these strong convective events correlates with the timing of tropical genesis in the model and observations.