In addition to the classic methods of data analysis, the researchers used artificial intelligence (AI) to filter out the rain signal from the noisy measured values. "Other factors such as wind or the sun can also lead to slight attenuation of the signal. With the help of our AI, we were able to detect when an attenuation is caused by rain," says Julius Polz, another scientist in the research group. In the meantime, the researchers have trained the AI so that they can do without calibration using traditional methods of rain measurement. This makes it suitable for use in regions without significant rainfall measurements that might be suitable for training the AI, for example more remote regions of the world.
For the general weather conditions prevailing in Germany, the method does not work equally well in all seasons. Accuracy suffers especially in winter, because hail and sleetfall cause an above-average attenuation of the signals; snow cannot be taken into account at all with this method.
Several projects are currently underway by the researchers to measure rain using radio relay systems, with the focus on Germany, in cooperation with the Deutscher Wetterdienst and the Saxony State Office for the Environment. In the course of the summer, further projects will start in the Czech Republic and in Burkina Faso, where a nationwide survey of microwave links in Africa is to be established for the first time.
More information: https://www.klima-umwelt.kit.edu/english/index.php