Temporal dynamics of streamflow: application of complex networks

• DOI: 10.1186/s40562-018-0109-8
• Publication Date: 2018-04-04T05:17:09Z
• Authors: Xudong Han; Bellie Sivakumar; Fitsum M. Woldmeskel; Milena Guerra de ...
• Keywords: Degree distribution; Flood forecasting; Clustering coefficient

This study employs the concepts of complex networks to temporal dynamics streamflow, with emphasis on annual scale (i.e., year-to-year connections). The proposes a new approach construct streamflow network at scale. It uses daily data network, instead using (mean or accumulated) data. With this approach, each year serves as node in having time series values (not single value). Streamflow observed over period 151 years (October 1862–September 2013) from Mississippi River basin St. Louis, Missouri, USA are considered for implementation approach. properties investigated three network-based methods: degree centrality, clustering coefficient, and distribution. sensitivity results correlation threshold is also examined. suggest that (1) there only few very significant nodes (years) (degree centrality method); (2) not classical random graph, but may be small-world scale-free (clustering coefficient (3) exhibits combination exponential power-law distribution method). Based identification stretch (around 1950s–1990s) weak connections rest studied, influence dam construction (and other anthropogenic factors) evolution dynamics.