Improved Rain Mapping by Graph Signal Processing Tools

In this paper, we apply, for the first time, the use of graph signal processing (GSP) to the emerging application of near-ground two-dimensional rainfall mapping based on signal level measurements from commercial microwave link (CML). Most existing rainfall reconstruction algorithms are based on spatial interpolation of assumed point rain measurements, arbitrarily representing each CML by a virtual rain gauge (VRG), located at the center of the CML. To demonstrate the usability of GSP tools in this setting, we challenge the arbitrary location of the VRG, looking for the best location of the VRG along the CML in some sense. Here, we show that using GSP tools enables the identification of alternative representative locations along the links that yield improved rainfall maps. The proposed approach is validated using real CML data, demonstrating its effectiveness in enhancing map accuracy.