Systematically Modeling the Interactions among Multiple Indicators While Considering the Structure of a River Network

Multiple water quality-related indicators in surface
water involve
many interactions. However, there is still no global interactive landscape
considering the river network (up- and downstream relationship). Fuzzy
cognitive maps (FCMs) are a type of quantitative method that conducts
the time series prediction while considering the interaction among
different variables (including self-impact). Additionally, FCMs enable
the analysis of variable interactions across different locations,
such as the monitoring sites within a surface water network (e.g.,
river network). In this study, we utilized the global monitoring data
from 49 stations (sites) along the Qu River, Fu River, and Jialing
River, Hechuan, Chongqing, China (February 1, 2021, to March 9, 2024),
to construct a global map illustrating the interactions among the
indicators across all of these sites. The analyzed results provide
insights to infer the interaction between any pairs of variables and
predict the amount of variables in future time stamps. The interstation
and intrastation relationships were analyzed from three perspectives:
simple path, cycle, and degree derived from the FCM-produced graph.
Concrete interactions were quantified using edge weights in the graph
to uncover the causes of pollution and understand the hidden trends
in the data.