Abstract: The meteorological data assimilation process can be described as a procedure that uses observational data to improve the weather forecast produced by means of a mathematical model. Traditional methods include the Kalman filter. However, this method demands a heavy computational power. Recently, neural networks have been proposed as a new method for meteorological data assimilation by employing a multilayer perceptron network to emulate Kalman filtering at a lower computational cost. This papper presents a new schme for learning process for the multilayer perceptron network, giving a more stable behavior for the assimilated data. Numerical results are shown for the one-dimensional shallow water meteorological model.