Modeling Variability in the Earth System: From Rainfall, to River Networks, to Landscape Processes
This research group focuses on a wide range of environmental problems including inverse problems for precipitation estimation from space, stochastic theories of water/sediment transport and landscape evolution, river delta network topology and dynamics for vulnerability assessment, and river morphodynamics from satellite imagery.
Current work focuses on new and innovative formalisms for inverse estimation problems (downscaling, data fusion, retrieval, and data assimilation) of precipitation using multi-sensor, multi-scale measurements from space. New methodologies for quantifying the stochastic nature of bedload sediment transport using multi-scale analysis and dynamical system theory are studied as well. The researchers seek to understand the relations between near-bed turbulence, riverbed morphodynamics, and sediment transport using experimental, theoretical, and numerical approaches. Landscape reorganization under climate change is also studied using both controlled laboratory experiments and conceptual modeling. Dynamical frameworks for the analysis of river meandering dynamics, inferring process from form, and quantifying response to perturbations are further studied in the group. They explore quantitative frameworks for studying river delta topology and dynamics based on graph-theoretic approaches, where deltaic systems are represented by rooted directed acyclic graphs. They also work on developing metrics that capture unique physical, topological, and dynamical aspects of delta networks with the ultimate objective that deltas can be compared and contrasted and also analyzed for relative vulnerability (or resilience) to change. At the basin scale, the researchers investigate network-based frameworks for identifying potential synchronizations and amplifications of sediment delivery to basin outlets and also for identifying hotspots of fluvial geomorphic change based on dynamic connectivity. They also analyze the transport and mixing of water particles traveling in a river basin based on a stochastic Lagrangian formulation of transport to evaluate the residence time distributions, which are fundamental catchment descriptors blending key information about storage, geochemistry, flow pathways, and sources of water into a coherent mathematical framework.
The group has recently begun a campaign to map and analyze river morphodynamics in South America using Landsat imagery. In order to map migration rates for thousands of kilometers of major rivers, they have developed a nearly-automated process that extracts binary water masks from tropical rivers via an in-house supervised classifier that exploits all bands available from Landsat. A toolbox has been developed to process these binary masks to extract river banklines, centerlines, and widths at the pixel (30 m) scale to quantify migration dynamics within the most active channels in the world at an annual timescale for more than three decades. Tools for geomorphologic analysis have been developed that compute migration rates, analyze in-channel bar morphologies, and automatically delineate individual bends through time. These tools are scalable and parallelizable; a proof-of-concept was successfully performed on 1,600 km of the Ucayali River (roughly 50 Landsat scenes). Expanding this study to the entire Amazon region will require computationally intensive processing, hence having access to powerful computational resources and parallel computational capacity is essential for the efficient implementation of these researches.
Return to this PI's main page.