Machine learning used for missing data in rainfall databases

In Mexico, the National Water Commission (CONAGUA) is responsible for monitoring climate variables through rainfall networks, therefore, in this dependency of the Government of Mexico, information on rainfall, temperature, wind direction is available. among others. However, it has been detected that sometimes missing data is presented, due to the operation of the systems and the lack of personnel responsible for data collection. To provide a solution to this lack of data, traditional hydrological methods have been implemented, such as the rationaldeductive method (MRD). However, thanks to technological development, new alternatives and proposals have been sought to obtain optimal rain predictions. Based on what has been previously described, this research focuses on the estimation of missing data of maximum 24- hour rainfall in some weather stations in the Zacatecas highlands region, this estimation is carried out by means of the MRD and by the proposed model of artificial neural networks (RNA), taking the mean square error (MSE) as a test.