The measurement of rainfall and its prediction at short lead times (often referred to as nowcasting) have important implications in hydrometeorology, and improving their accuracy may have an impact on various human activities. Indeed, the importance of nowcasting is related to its application in preventing or reducing fatalities, injuries, and property damage caused by weather hazards, and in improving management strategies in agriculture, transportation, and industry. Currently the most used tools to estimate rainfall intensity are rain gauges and weather radars. Rain gauges have relatively high precision, but poor spatial representativity because of the high spatial and temporal variability of rainfall. Unlike rain gauges, weather radars monitor rainfall fields over large areas with high spatial and temporal resolutions, but through indirect measurements, with various sources of error. In addition, a complementary approach for rainfall measurement has been recently proposed, based on the relationship between path-averaged rain rate and attenuation affecting microwave links used in telecommunication networks when rainfall occurs. The main objective of this work is to improve the use of different sources of information for rain rate estimates and short lead-time prediction. Firstly, a new tool to identify faulty rain gauges using surrounding microwave links is proposed. Secondly, data from rain gauges, radars, and microwave links are combined to improve rain rate estimates. Here, assimilation techniques are used to retrieve the rain rate with greater accuracy at a high spatial resolution. Finally, a novel data assimilation framework is developed to generate accurate nowcasting (short lead-time rain predictions) using the three types of sensors, while incorporating the uncertainty associated with the different sources of information. While each sensor has its specific limitation, it is shown that significantly improved rain rate estimates can be obtained by combining the rain rates retrieved by different sensors after correctly parametrizing the respective errors. This statistical framework can also incorporate other sources of information, once the associated errors are known. The framework delivers reliable and accurate rain rate estimates and short lead-time forecasts at the ground level, with the associate uncertainty, as required by hydrological models.