Classifying Hydrologic Regimes of the Amazon

As the largest watershed in the world, the Amazon River basin contains a vast diversity of habitats and accompanying hydrologic regimes. Further understanding the spatial distribution of streamflow regimes across the Amazon can inform river management and conservation, especially in areas with limited or inconsistent streamflow monitoring. This study compares inductive and deductive approaches to classify flow regimes across the Amazon River basin using an unprecedented compilation of streamflow records from Bolivia, Brazil, Colombia, Ecuador, and Peru. Inductive classification schemes use attributes of streamflow data to categorize river reaches into similar classes, while deductive classification approaches use environmental attributes to predict regions with similar flow regimes. In this study, inductive classification was accomplished through principal components analysis and k-means hierarchical clustering of 73 hydrologic indicators for 369 stations. These results were compared with deductive classifications produced by the Global Rivers Classification (GloRiC) database, where stations were assigned to a series of hydrologic, geomorphic and physioclimate classifications. The first two principal component axes represented 29% and 18% of the variance, with IHA parameters related to magnitude and duration as strong determinants of flow regime, however no discernable groupings were apparent from this analysis. K-means clustering grouped streamflow records into six distinct hydrologic classes, which were shaped by magnitude of flow and seasonality. The resulting inductive classes aligned strongly with the deductive classifications produced by GloRiC. Exceptions were explained by within group variation, underscoring the value of nested clusters when exploring streamflow across multiple scales. These results highlight the diversity of flow regimes across the Amazon and provide a framework for studying hydrology in the context of changing climate, land use, and human-induced hydrologic alteration. The methodology developed here provides a rigorous, data-driven approach for classifying flow regimes based on observed data. When coupled with local knowledge and expertise, these classes can be used to support hydrologically and ecologically sound conservation practices.