Fraunhofer IDMT
Abstract:
The most common method for measuring precipitation is based on the collection of rain in a defined container,
where the amount of water accumulated is read off after a time interval. A major disadvantage of this method is that
the rainfall amounts measured in this way always represent an average over time, and do not provide data on the
course of precipitation with high temporal resolution. Since time immemorial, meteorological elements have been described
colloquially by their generated sounds. The aim of this study is the implementation of an acoustic rain sensor. For this purpose,
acoustic data were generated in a laboratory setup by a simplified sprinkler system, with which a machine learning
algorithm was trained.