Hot Water: Integrated Modeling for Watershed Fire Impacts and in Stream Thermal Habitats

Abstract

The temperatures of water and soil are fundamental controls on biogeochemical and ecological processes, ranging from carbon cycling to fish health. Heat is transported through watersheds by flowing water in runoff, soils, groundwater, and streams, with the implication that major disturbances to watershed hydrology – for example, from wildfires – can dramatically alter thermal conditions, and thus biogeochemical and ecological conditions, within streams. Given the strength and complexity of these connections among water, heat, and biogeochemical activity in watersheds, our field is increasingly turning to model integration as a way to combine observations and derive new insights that span the aquatic-terrestrial interface. In this session we will discuss examples of model integration that bring together diverse processes, domains, and resolutions to answer questions about the drivers and consequences of thermal processes for watershed ecosystem function.

Wildfires cause significant but poorly understood impacts on water and carbon cycles, posing a growing threat due to the increasing frequency, spread, and severity of these disturbances. Integrated hydrobiogeochemical (HBGC) models are crucial for understanding watershed water and carbon cycling dynamics, providing both retrospective analyses and future projections. We use the modeling system ATS-PFLOTRAN, which couples flow and reactive transport processes, to understand how land use, hydrogeology, climate, and wildfire interact to control carbon and nitrogen cycling in a few watersheds across the Yakima River Basin, located in the Pacific Northwest region of the United States. Our models successfully captured the fire-induced changes in watershed hydrology resulting from vegetation loss and alterations in surface physical properties. The reduction in topsoil permeability following moderate to high-severity wildfires leads to increased peak streamflow, which is exacerbated by increasing precipitation rates. In contrast, low-severity wildfires do not significantly impact hydrological processes. Furthermore, our study highlights the importance of integrating heterogeneous terrestrial carbon inputs into watershed HBGC models to accurately simulate carbon and nitrogen cycling dynamics. A gap remains in representing the wildfire-induced changes in thermal processes, which have important implications for water temperatures in fire-impacted streams.

At the U.S. Geological Survey, we are developing a national machine-learning-based model specific to stream temperature prediction, in which thermal processes are implicitly learned rather than explicitly encoded. Our national model will provide equitable access to water temperature predictions for citizens across the contiguous U.S. (865,000 km total stream length); however, this model has relatively coarse resolution (median reach length 13.7 km). Therefore, we are also working on multi-scale methods that integrate the coarse national model with local data to simulate higher-resolution spatial detail (median reach length of 1.6 km) in both mainstems and smaller tributaries (up to 5,100,000 km total length). Between recent enhancements of the national model to predict daily minimum, maximum, and mean temperatures and their uncertainty, and the addition of multi-scale modeling to leverage coarse national and fine local information, we are moving from a basic prediction capability to the ability to predict stream thermal habitat with enough fidelity and resolution to support fish habitat management. Gaps remain in these models’ ability to capture thermal effects of processes such as groundwater discharge, reservoir management, and wildfires – which raises the opportunity to integrate our baseline model with more detailed representations of those processes, potentially gaining benefits from both the data-driven and process-based elements of such an approach.

Biographies

Dr. Xingyuan Chen is a senior Earth scientist at the Pacific Northwest National Laboratory. Her research focuses on understanding and predicting how watershed and river corridor systems respond to various anthropogenic and environmental disturbances. Her research has involved using high-performance process-based models and machine learning for data-model integration. She got her PhD in Civil and Environmental Engineering from University of California at Berkeley.

Dr. Alison Appling is a water data scientist with the U.S. Geological Survey (USGS). She uses a combination of statistical, theoretical, and knowledge-informed machine learning methods to model water quality in rivers and lakes. She is especially passionate about developing new, improved, and integrated modeling methods to understand patterns and drivers of water quality. She also manages multi-method water prediction projects at USGS, including data-driven and process-based approaches that span the water cycle, in support of national and regional water availability assessments. She got her PhD in Ecology from Duke University.