Data Assimilation using neural networks applied to the CPTEC-INPE global atmospheric circulation model

Abstract

The description of the most physical phenomena is based on differential equations. But such representation has modeling error. For the operational prediction systems, a strategy to deal with the uncertanties in the model is to add some information from the real world to the mathematical model. This adicional information consists of observations (measurement values) from the modeled phenomena. However, the observed data might be carefully inserted, in order to avoid the degradation of the prediction. Techniques for data assimilation are tools for effective combination two source of data: observation and physical-mathematic model, for computing the analysis. The analysis is the initial condition used in the prediction computer model. This is an essential procedure for, as an example, operational weather forecasting. The goal of the present research project is to employ the artificial neural networks as a data assimilation method applied to the General Atmospheric Circulation Model from the CPTEC-INPE. The neural networks should compute a good analysis, enhancing the computational performance.