Rainfall monitoring networks are key elements in the development of warnings and predictive models for communities at risk of flooding during high intensity rainfall events. Currently, most of these networks send the rainfall measurement to a data center in real time using wireless communication protocols, thus avoiding travel to the measurement site. An application case of this is an Early Warning System (EWS) for pluvial flash floods developed in Barranquilla, which used the GPRS protocol to send real-time rain gauge measurement data to a web server for further processing; however, this protocol has high energy consumption and high maintenance costs. In the present work, an evaluation of three low power wireless communication protocols, Zigbee, LoRaWAN and Sigfox, is carried out to determine which one is the most suitable to be applied to an early warning system that monitors rainfall in the city of Barranquilla. The stipulated evaluation metrics are the characteristics of the link profile, energy consumption and costs of the devices for the implementation of the system. To perform the evaluation, a wireless sensor network was designed and characterized, which was analyzed using the free software tool Radio Mobile; this tool allowed determining the performance for each communication protocol, taking into account the measurement points implemented with the GPRS network of the previous SAT. The final results of the simulation show the LoRaWAN protocol as a viable alternative with good performance.