The Idealized Tests And Real Case Studies On One-Way And Two-Way Nesting In Limited Area Model

Numerical weather prediction models are developing to a direction which the models can depict and forecast more precisely of the movement of the atmosphere in the future. To achieve this goal, Increasing the resolving power is a necessity but not sufficiency. Because of the limit of computer power, the models cannot run at a high resolution in the whole area, so nesting model is still needed to get high resolution in partial domain which can improve the ability of forecast and simulation. The nesting technology consists of one-way nesting and two-way nesting. In this article, my main job is to study the model self-nesting technology and find out the problems while using this technology, and also to prove the superiority of two-way nesting.In the first part of the article, we design some nesting experiments using the linear shallow water equations to find the problems while using the one-way nesting and two-way nesting. The main problems are the boundary forcing and the impact after the two-way nesting feedback, also we have to find out how to get through these problems. First we do the nesting experiments using one dimensional linear shallow water equations. The experiments are made up of a two domains nesting and a three domains nesting. The center of the nesting area is coincided. We study the change of a wave packet after it moving out of the inner nesting area with the influence of the lateral boundary. Also we do the nesting experiments using two dimensional linear shallow water equation and study the influence of nesting lateral boundary and two-way nesting feedback. In these experiments, as to the wave reflection of lateral boundary, we found that two-way nesting have less fake noise reflection than one-way nesting, also we found that the sponge boundary condition can generate less wave reflection than the interpolation boundary condition. When the wave length is small, the coarser grid cannot resolve the waves well and the inner nesting area cannot get a suitable boundary from the coarser grid interpolation which may give rise to a severe fake wave reflection in the inner domain. We can use a filtered sponge boundary condition to get rid of the short wave problem. We found that there are complicated interactions between the multiple dimensional grids when we doing the three domains nesting experiments. When there are noisy reflections in the parent domain then the child domain also generate noisy reflections especially when the real wave had move out of the child area. The noisy reflection can also be feedback to the parent domain when we using the two-way nesting. The advantage of two-way nesting is that the coarse and nested grid solutions are allowed to interact with one another. In our experiment we found that the wave moves faster and more accurate after using the two-way nesting.Then we do a series of real simulation using the WRF model to make a comparison of one-way nesting and two-way nesting. The comparison experiments are the same parameter setting except for the nesting methods. We found that when the weather system moves slow or still, the simulation results of one-way nesting and two-way nesting are basically the same, for example the cloudy-rainy weather for several days in the middle and lower reaches of Chang Jiang River. When the weather system moves fast, the results seem to make difference between one-way nesting and two-way nesting. In our simulation of systems like typhoon and the Beijing heavy rain of721, we found that two-way nesting performs better than one-way nesting. The reason two-way nesting performs better we think is that the two-way nesting can feedback the high resolution information that the finer grid captured to the parent domain and this help to improve the simulation results of parent domain, and the improved parent domain can give a better later boundary condition to its child which inverse help to improve the accuracy of simulation.