Informing River Corridor Transport Modeling by Harnessing Community Data and Physics-Aware Machine Learning
Abstract
We are developing a data-driven and machine learning-facilitated transformation of how we craft one-dimensional models for solute transport in rivers. After testing state-of-the-art 1-D models with hundreds of tracer tests performed worldwide, research has consistently shown that our current solute transport theory in river corridors misbehaves categorically, and that we are missing some fundamental mechanisms controlling the exchange of solutes with the hyporheic zone. We propose to address this challenge by creating a public database of solute tracer experiments, which will be used to perform data analytics, multimodel analyses of temporal moments, and physics-aware machine learning. This work will help to assimilate tracer data and hydrogeomorphic measurements objectively to learn mechanistic, constitutive relationships representing unknown parameters and regimes that can inform and validate the governing equations of a newly proposed transport theory. In this presentation, we will share our progress towards developing the initial version of this global tracer test database, which includes hundreds of experiments, and our view for increasing synergy and broadening participation in river corridor research by improving the database through the incorporation of ICON (integrated, coordinated, open, and networked) and FAIR (findable, accessible, interoperable, and reusable) principles. We will discuss opportunities for public participation in the database design, its availability, and how new experimental work will expand and enrich the wealth of information contained, allowing inter-site and multi-scale comparisons of solute transport processes in river corridors.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFM.H35A..01T