Development of a moisture scheme for the explicit numerical simulation of moist convection

Abstract

The eastern part of South Africa is a summer rainfall region, and most of its rainfall is of convective origin. Convective rainfall is often associated with hail, lightning and tornadoes – all of which can damage property and cause casualties, especially within such a densely populated region. Improved forecasting of convective systems can lead to better early warming systems which will in turn reduce casualties and damage to movable properties. Atmospheric models that assume there is a balance of forces in the vertical direction in the atmosphere are called hydrostatic models, and have been used successfully to make operational forecasts since the 1950s at low resolutions. These models cannot explicitly simulate moist convection, but apply a procedure called cumulus parameterisation to statistically represent the effects of convection on the atmosphere. These parameterisation schemes are thought to be an important source of uncertainty in both numerical weather prediction and climate simulations. The development of nonhydrostatic atmospheric models that can explicitly simulate the dynamics of atmospheric convection has been ongoing since the 1960s. These models have been utilised largely for research purposes, as their application to operational weather forecasting and climate simulation was hindered by computational restrictions. However, with the advent of ever faster computers, the operational numerical integration of weather prediction models at spatial resolutions beyond the hydrostatic limit, has become a reality. This has led to a renewed and worldwide effort to develop nonhydrostatic models. A nonhydrostatic sigma coordinate model is currently being developed at the CSIR for purposes of simulating weather at spatial resolutions where the hydrostatic approximation is not valid (i.e. horisontal resolutions higher than about 10 km). Note that hydrostatic atmospheric models – applied at relatively low spatial resolutions where convection cannot be resolved – are likely to be applied for decades to come in the computationally expensive study field of climate simulation (and for numerical weather prediction over large regions). The cumulus parameterisation schemes, that statistically represent the effects of moist convection and convective rainfall on the larger environment, should therefore continue to be improved for use in hydrostatic models