Global high-resolution models require excessive resources for long-term climate integration. To relax the resource limitation problem and to obtain high-resolution climate results, downscaling coarse resolution results to regional scale by the high-resolution models over a limited time range is an efficient way to achieve the goals. To regional climate study, high resolution comes along with more details of surface conditions, which enhances regional scale features, though there are only coarse resolution initial condition and large-scale nesting. In this study we will conduct an efficient way to nest large-scale results from coarse resolution to high-resolution global models. Numerical experiments are operated on NCEP Global Forecast System (GFS) and semi-Lagrangian scheme is an option to be used to save integration time. The results from IPCC AR4 and/or AR5 are selected as the driving large-scale forcing to the NCEP GFS atmospheric high-resolution model. The regional-scale prediction and projection from this global downscaling system by high-resolution global model will be demonstrated and illustrated in details.