Modelling the re-intensification of tropical storm Erin (2007) over Oklahoma: understanding the key role of downdraft formulation

摘要

This article reports on the inland re-intensification of tropical storm (TS) Erin (2007). In this research, the physical processes that resulted in the re-intensification of TS Erin over Oklahoma, USA, on 19 August 2007 was determined and a sensitivity study on microphysics, planetary boundary layer and convective parameterisation schemes was performed in the mesoscale modelling system, MM5. Also, we diagnosed and explained the remarkable difference between model behaviour of the original Kain–Fritsch 1 (KF1) scheme and its revised counterpart (KF2). The numerical results showed only modest sensitivity to the selected microphysics schemes – the relatively simple ‘Simple Ice’ and the advanced Reisner-Graupel. We found a relatively high sensitivity to the selected boundary layer parameterisation. Enhanced mixing in the medium range forecast (MRF) scheme leads to a relatively small convective available potential energy (CAPE), a deeper boundary layer and a lower dew point temperature, thus to a relatively stable environment. Therefore, MRF forecasts less precipitation (up to 150 mm) than the local mixing scheme, ETA. Model results appeared most sensitive to the selected convection schemes, that is, Grell, KF1 and KF2. With Grell and KF1, Erin intensifies and produces intense precipitation, but its structure remains close to a mesoscale convective system (MCS) or squall line rather than of the observed tropical cyclone. Both schemes also simulate the most intense precipitation too far south (100 km) compared to observations. On the contrary, KF2 underestimates precipitation, but the track of the convection, the precipitation and the pressure distribution are relatively close to radar and field observations. A sensitivity study reveals that the downdraft formulation is critical to modelling TS Erin's dynamics. Within tropical cyclogenesis, the mid-level relative humidity (RH) is generally very high, resulting in very small downdrafts. KF2 generates hardly any downdrafts due to its dependence on mid-level RH. However, KF1 and Grell generate much stronger downdrafts because they both relate the downdraft mass flux (DMF) to vertical wind shear. This larger DMF then completely alters TS Erin's dynamics. A modified version of the Grell scheme with KF2 RH-dependent downdraft formulation improved the simulation considerably, with better resemblance to observations on trajectory, radar reflectivity and system structure.