The impact of topography filter processing and horizontal diffusion on precipitation prediction in numerical model
数值模式中地形滤波处理及水平扩散对降雨预报的影响

In high-resolution modeling,there is a problem of the prediction of unrealistic precipitation field in mountainous regions with deep valleys and 'singular' mountain tops.To solve the problem,digital filtering and horizontal diffusion were used in this paper to process topography and computational noise,respectively.The filter consists of different one-dimensional high-order low-pass implicit tangent filters and can selectively remove different scale noises that result from the change of topographical gradient.Horizontal diffusion was used to control small-scale noise that results from numerical dispersion,nonlinear instability,and discontinuous physical processing,etc.by adding a linear forth-order monotonic horizontal diffusion with orographic flux limiting to the prediction equation.Our tests show that the phenomenon is eliminated in which extremely large amount of precipitation occurs directly over mountain tops and at the same time no precipitation occurs in deep valleys and at mountain lee-sides by using both topography filter processing and horizontal diffusion.Therefore,the precipitation patterns appear to be more realistic.