We will be starting off the semester with our first all NWT meeting of 2020 on Wednesday, January 15th at 2:30 pm in S228. For this all NWT meeting we will be hosting a talk from Dr. Haruko Wainwright from Lawrence Berkeley National Laboratory, followed by time for discussion. 

The zoom link to attend this meeting remotely is: 

https://cuboulder.zoom.us/j/104089123 

Please see below for her talk title and abstract:

Watershed Functional Zonation for Quantifying Watershed Organization and Functions Based on High-resolution Airborne Remote Sensing Data

While watersheds are recognized as the Earth’s key functional unit for assessing and managing water resources, developing a predictive understanding of how watersheds respond to climatic perturbations is challenging due to the complex nature of watersheds. This is particularly true in mountainous watersheds, where extreme lateral gradients in hydrogeology, biogeochemistry and vegetation often exist, and perturbations (such as early snowmelt) can lead to changes in process interactions that potentially affect downgradient water, nutrient, carbon and contaminant exports. The Watershed Function Scientific Focus Area project (watershed.lbl.gov), which is being carried out in the mountainous East River, CO headwaters catchment of the Upper Colorado River Basin, aims to develop a predictive understanding of watershed hydrobiogeochemical response to perturbations.

This talk focuses on the spatiotemporal characterization of the heterogeneous and multiscale fabric of watersheds. Recent advances in remote sensing have revolutionized the way we characterize watersheds from bedrock to canopy such as LiDAR and hyperspectral technologies for high-resolution geomorphology, and plant species/structure. In addition, airborne electromagnetic surveys provide subsurface structure and properties across watersheds that were previously characterized only through borehole data. By taking advantage of these data layers, we have explored several unsupervised/supervised machine learning techniques to gain quantitative understanding of watershed organization and functions. We hypothesize that (1) the co-evolution of watershed terrestrial systems creates co-variability among subsurface/surface spatial features (such as topographic, plant, snow and geological metrics), (2) we can reduce the parameter dimensionality by exploiting such co-variability, and (3) we can identify several representative landscapes – watershed functional zones – that capture distinct characteristics of those co-varied properties and associated watershed functions. Results show that unsupervised learning is powerful to identify the surface/subsurface co-variability such as bedrock fracturing and plant species over the watershed, and identify several key zones that capture watershed-scale heterogeneity. Supervised clustering results shows that the elevation, aspect and geology are the key controls on both drought sensitivity and nitrogen export, and that we can map watershed “functioning” zonation, and predict annual nitrogen export in unmeasured sub-catchments based on spatial features and peak snow water equivalent.