Causality in Water Systems

Projections of near-term climate change and human population growth suggest that freshwater resource allocations for humans and ecosystems will be increasingly at odds—with concomitant risk to people, economies, and ecosystems. Large, existing water infrastructure is progressively viewed as a solution to reduce human vulnerability to climate variability. However, we lack robust methods for balancing human and ecosystem needs via water infrastructure operations in the face of uncertainty. Here we propose that a better understanding of causal pathways and feedbacks within water systems—the causality and feedbacks between hydroclimatic conditions, human and ecosystem needs, and reservoir operations—could lead to more sustainable management outcomes when dynamic, competing needs for water exist. This is because feedbacks between hydroclimate, human consumptive and non-consumptive uses, and ecosystem needs are common; yet, these interactions are rarely considered in models examining sustainable water infrastructure management. This interdisciplinary working group has two objectives: First, we will evaluate the relative merits and limitations of a wide range of statistical and mathematical techniques in detecting causal networks from time-series data. Our comparison will include novel methods such as convergent cross-mapping, which can capture the feedback loops and non-linearities that are often present in complex socio-environmental systems. Second, we will apply these methods to the Lower Colorado River, an over-allocated system with abundant empirical data on hydroclimate, human demand for water (municipal use, irrigation), hydropower production, downstream river ecosystem integrity, and regulatory constraints. We anticipate that this application will show how a better understanding of causal pathways and interdependencies within a water system can lead to a higher chance of satisfying multiple objectives, and to a better quantification of risk on water security for society and ecosystems. Thus, the inferences drawn here will be relevant to other socio-environmental problems with complex dynamics and with time-series data on their potentially-interacting components.