Classification of California streams using combined deductive and inductive approaches: Setting the foundation for analysis of hydrologic alteration

Abstract

Regional classification of streams is an early step in the Ecological Limits of Hydrologic Alteration framework. Many stream classifications are based on an inductive approach using hydrologic data from minimally disturbed basins, but this approach may underrepresent streams from heavily disturbed basins or sparsely gaged arid regions. An alternative is a deductive approach, using watershed climate, land use, and geomorphology to classify streams, but this approach may miss important hydrological characteristics of streams. We classified all stream reaches in California using both approaches. First, we used Bayesian and hierarchical clustering to classify reaches according to watershed characteristics. Streams were clustered into seven classes according to elevation, sedimentary rock, and winter precipitation. Permutation‐based analysis of variance and random forest analyses were used to determine which hydrologic variables best separate streams into their respective classes. Stream typology (i.e., the class that a stream reach is assigned to) is shaped mainly by patterns of high and mean flow behavior within the stream's landscape context. Additionally, random forest was used to determine which hydrologic variables best separate minimally disturbed reference streams from non‐reference streams in each of the seven classes. In contrast to stream typology, deviation from reference conditions is more difficult to detect and is largely defined by changes in low‐flow variables, average daily flow, and duration of flow. Our combined deductive/inductive approach allows us to estimate flow under minimally disturbed conditions based on the deductive analysis and compare to measured flow based on the inductive analysis in order to estimate hydrologic change.

Tools for managing hydrologic alteration on a regional scale: Setting targets to protect stream health

“…Indeed, this index, combined with the regional applicability of the hydrologic models, likely account for the small differences in targets we observed among hydrologic stream classes (classes from Pyne et al. , results not shown). Generating targets for other biological management endpoints, such as fish, benthic algae or riparian vegetation, may require additional analyses to control the influence of natural factors if predictive models or indices are unavailable.…”

A stream classification system for the conterminous United States

“…The above principles are a stark contrast to the many previous national-scale stream classification efforts, which have either classified discrete observations (e.g., stream monitoring points) 2 , used deductive approaches for grouping streams 10,24 , and/or classified singular, as opposed to multiple, habitat components, primarily hydrology 22 . While these approaches have enriched our understanding of stream function, they are limited in their ability to comprehensively represent the emergent properties of stream ecosystems and their habitat components across large regions 6,7 .…”

National classification of surface–groundwater interaction using random forest machine learning technique

“…The RF machine learning technique combines multiple decision trees into an ensemble to produce accurate classifications or regressions (Breiman, ; Cutler et al, ). This method has been widely used for variable classification and regression in the field of hydrology, for example, in stream classification (González‐Ferreras & Barquín, ; Pyne et al, ). In this study, the method (Breiman, ) was used to classify stream statuses (i.e., gain, loss, gain_loss, and uncertain; Section above) as a function of available catchment or stream characteristics (Section below and Table ).…”

“…Olden, Kennard, and Pusey () differentiated between inductive and deductive classification and developed a framework to depict the critical components of the classification process and associated statistical approaches used by each. By contrast, Pyne, Carlisle, Konrad, and Stein () used a combined regional‐scale inductive/deductive classification approach to categorize streams according to their climatic, hydrologic, morphological, and/or geomorphological characteristics and then utilized a random forest (RF) approach to determine which variables were the most effective classifiers. Brown, Lester, Versace, Fawcett, and Laurenson () went further and used combinations of nonhierarchical clustering, RF, and principal components analysis to classify streams in south‐eastern Australia.…”