Community Detection in Very High-Resolution Meteorological Networks

• Published in: IEEE geoscience and remote sensing letters Vol. 17; no. 11; pp. 2007 - 2010
• Main Authors: Ceron, Wilson, Santos, Leonardo B. L., Neto, Giovanni Dolif, Quiles, Marcos G., Candido, Onofre A.
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01-11-2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Clustering methods; Community structure; complex networks; Correlation; Disaster management; Ecological studies; Embedding; Emergency preparedness; Geographic information systems; Geographical distribution; Geographical locations; Graphs; High resolution; Information systems; Land cover; Land surface; Land use; Meteorological data; Meteorological networks; meteorological radar; Meteorology; Networks; Nodes; Properties; Radar; Radar data; Radar remote sensing; Rain; Rainfall; Rainfall data; Resolution; Risk management; Risk reduction; Spatial discrimination; Spatial resolution; Surface topography; Surface treatment; Sustainable development; Topology; Urban planning
• DOI: 10.1109/LGRS.2019.2955508
• EISSN: 1558-0571
• ISSN: 1545-598X

Several complex dynamical systems are embedded in geographical space. Geographical data have proven its importance in several domains. For instance, the formation and scrutiny of climate networks have emerged as a new research topic in environmental literature. However, there is still a lack of investigations of scenarios with very high spatial resolution, such as those considering meteorological data. Recently, a new concept, named (geo)graphs, was proposed. (Geo)graphs are graphs, or networks, in which the nodes have an assigned geographical location. Besides embedding nodes into space, these graphs are readily manipulated with a geographical information system, and, thus, represent a suitable tool for dealing with very high-resolution scenarios, such as meteorological data. In this context, here, we apply a (geo)graph approach to model a radar-derived rainfall data set. We represent the nodes as a point-type shapefile and the edges as a line-type shapefile, which are standard file types in geoinformatics. After, we analyze the topological properties of a family of (geo)graphs considering distinct thresholds. The analysis of these networks reveals a spatially well-defined community structure, which, interestingly, is consistent with topographical/altimetric and land use/land cover data. These results show the relation between geographical properties and the topological structure of the network might be applied to different ecological studies, from sustainable development to urban planning and disaster risk reduction.