The Research Of Water Substance Advection Problem In GRAPES Model

How to better simulate the advection and distribution of water substance has an important significance for the improvement of the numerical weather prediction, especially for precipitation. With the rapid development of work-station, the resolution of numerical model is increasingly sophisticated, it is possible to calculate the cloud explicitly, so it bring forward a higher request for the computation of water substance advection, it must be high-order accuracy, conservation, and shape preserving. At the same time, the direct use of higher-order central difference method to solve the water substances advection problem will cause a lot of numerical dispersion, and lead the appearance of negative water vapor. The advection process should not destroy the original spatial features of the scalar variables, should not create new negative values. On the other hand, even if the scheme guarantees the monotonic of the advection process, when there is latent heat of condensation occurs, because the interpolation to get the semi-Lagrangian upstream point, the region of cloud edge would produce ostensible cloud water, resulting in unreasonable phase-change process.To solve these problem, this paper study a new water substance advection scheme and it's application into GRAPES model; then, study the amendment scheme to phase change and latent heat, saw about the affect of this scheme to the simulation of convective cloud structure and precipitation forecast in GRAPES model. Mainly to amend the simulation of water substance advection, improve the effect of precipitation in the forecast.This paper divides into two sections. In the first section, verified the water substance advection scheme CSLR, compared with the old scheme QMSL, can do a better job in high-order accuracy, conservation, and shape preserving; then added the CSLR scheme into GRAPES_meso and GRAPES_global, through practical forecast experiments show that, with CSLR scheme, the model can better simulate the water substance advection, improved the effect of precipitation prediction. In the second section, through ideal tests show that, verified the phase-change amendment scheme NLSL scheme can effectively limit the ostensible phase change in cloud edge region generated by interpolation, then added the NLSL scheme into GRAPES_meso, through practical forecast experiments show that, with NLSL scheme, the model can better simulate the water substance advection and distribution, and the central precipitation area of rain bands is more comply with observation.