Fig. 3: The framework of YingLong model. | Communications Earth & Environment

Fig. 3: The framework of YingLong model.

From: An artificial intelligence-based limited area model for forecasting of surface meteorological variables

Fig. 3

Rolling forecasts of YingLong on the region ED through the process in Methods “Rolling forecast”. The prediction of Yinglong on the ED \({\hat{X}}_{t+i\Delta t}\left({{{\rm{ED}}}}\right)\) is divided into \({\hat{X}}_{t+i\Delta t}\left({{{{\rm{ED}}}}}_{{{{\rm{in}}}}}\right)\) on the inner area and \({\hat{X}}_{t+i\Delta t}\left({{{{\rm{ED}}}}}_{{{{\rm{LB}}}}}\right)\) on the lateral boundary region. Then boundary smoothing (Eq. (3)) is applied to \({\hat{X}}_{t+i\Delta t}\left({{{{\rm{ED}}}}}_{{{{\rm{LB}}}}}\right)\) and \({\hat{Y}}_{t+i\Delta t}\left({{{{\rm{ED}}}}}_{{{{\rm{LB}}}}}\right)\) (the forecast of Pangu on \({{{{\rm{ED}}}}}_{{{{\rm{LB}}}}}\)) to form \({X}_{t}\left({{{{\rm{ED}}}}}_{{{{\rm{LB}}}}}\right)\). Finally, \({\hat{X}}_{t+i\Delta t}\left({{{{\rm{ED}}}}}_{{{{\rm{in}}}}}\right)\) and \({X}_{t}\left({{{{\rm{ED}}}}}_{{{{\rm{LB}}}}}\right)\) are merged for the initial value of YingLong prediction \({\hat{X}}_{t+\left(i+1\right)\Delta t}\left({{{\rm{ED}}}}\right)\).

Back to article page