The research carries out an evaluation of the 3DVAR method with different options for the assimilation of reflectivity data, which are applied to the SisPI system with the purpose of determining which scheme presents the best results in the short-term numerical weather prediction. For this, data from 6 meteorological radars with coverage over a domain with 3km of spatial resolution are used, using the indirect method with (3DVAR-Rad) and without (3DVAR-NoRain) activate an option to also consider null-echoes of reflectivity without presence of precipitation. PrepBufr data from the GDAS system and radiance data obtained from different sensors on board polar orbiting satellites are also assimilated in all simulation domains. As a case study, the cold front that affected western of Cuba on December 10^th, 2018 is taken, which was preceded by a prefrontal depression that caused heavy rains on the north coast of Havana with significant accumulations in less than three hours. In the development of the work, the WRF model (4.1.2) is used with the WRFDA module (4.3), for which with a period of a month of SisPI simulations, covariance matrices of the background errors of each one of the domains with 27, 9, and 3 km spatial resolution, where in the last case this matrix is generated with the inclusion of hydrometeors as additional control variables. Also with the 3DVAR method, multiple outers loops are used, noting that the resulting cost functions show great similarity, although the 3DVAR-NoRain configuration tends to minimize this function in fewer iterations. The results obtained suggest that the highest Pearson correlation values are reached with the 3DVAR-NoRain configuration in an order of 0.4, while in the first 3 hours this configuration also shows the lowest MAE values compared to the INSMET synoptics stations network.