Comparative analysis of the role of domain size, horizontal resolution and initial conditions in the simulation of tropical heavy rainfall events

Abstract

Limited area models (LAMs) currently provide the only computationally feasible tool to simulate weather processes at very high spatial resolution of 1–2 km. Unlike in a global circulation model (GCM), however, the limited domains, and the associated artificial lateral boundaries, introduce additional uncertainties and errors into the simulation of LAMs. Domain size and grid spacing (horizontal resolution) both influence the spectrum of resolved scale and the nature of scale interaction in the model dynamics. A comparative, comprehensive and quantitative estimation of the relative role of domain size, grid spacing and initial condition in the simulation of mesoscale events is, however, lacking. This issue is investigated using a mesoscale model (MM5V3) with respect to three high impact weather (heavy rainfall) events in the tropics with a series of ensemble (five initial conditions) simulations. A high-resolution (10 km) benchmark simulation is used to establish the model's performance in simulating heavy rainfall events which occurred over the west coast of India during 26–27 July 2005. As the study's emphasis is on relative roles and not on the precise forecast, sensitivity simulations are carried out at coarser resolutions in view of the large number of simulations involved. Results show that, along with initial conditions and grid size, the size of the domain also significantly affects simulated quantities such as total rain and maximum rain. The results also show that domain size plays as important a role as that of grid spacing and initial condition in the simulation of high impact weather events. Copyright © 2011 Royal Meteorological Society