Current Research
LakeView: New understanding of lake water quality through integrated Earth observing systems
When we look at the surface of a lake, we might note the clarity or color and then wonder why it looks that way. Stuff (organisms, particulate matter, etc.) in the lake reflects and absorbs light at different wavelengths, resulting in what we interpret as color with our human eyes. For example, the chlorophyll pigment in phytoplankton reflects green wavelengths. To look at an entire water body, we can mount a hyperspectral instrument on a plane. "Spectral” refers to recording reflectance or absorption at different wavelengths, and “hyper” means we get these data at high resolution so that we can differentiate slightly different colors that might correspond to different “stuff” in the water. How we interpret the hyperspectral data is reliant on information we are collecting from fieldwork. By getting out on the water and measuring things that produce color at the same time the hyperspectral instrument takes its images, we can then build algorithms that convert spectral information into, for example, approximations of phytoplankton abundance, which tells us a lot about water quality and lake processes. This project aspires to unlock the power of upcoming hyperspectral satellite missions; those instruments will give us so much data across space and time, but we need to be prepared to interpret and apply it.
|
Ecology Knowledge-Guided Machine Learning
Machine learning is a powerful way to digest lots of data, identify trends, and create predictive models. However, left unchecked ML may output values that are unrealistic in the natural world, so we need to set boundaries for it with the ecological knowledge we have. This project is an effort to merge process-based modeling with machine learning to make better predictions about water quality across large spatial and temporal scales. Data from the Northern Temperate Lakes Long-Term Ecological Research (NTL-LTER) Program has been instrumental in training and testing models.
|
Environmental Data Initiative: data curation and the pursuit of FAIR, analysis-ready data
I am a member of the data curation team at EDI. When someone submits a dataset to be published, we review it and work with the submitter to make sure the data and metadata are complete and Findable, Accessible, Interoperable, and Reusable (FAIR).
The Environmental Data Initiative envisions a future of scientific discovery fueled by data commons accessible to everyone. We preserve environmental data for open and reproducible science, to promote synthesis across space and time, and to aid in the assessment of environmental change and its consequences. |
Past Research
{ THIS SECTION IS UNDER CONSTRUCTION }