05.12.2023 Views

Environmental Internship Program - 2023 Booklet

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Jamie Kim ’24<br />

CIVIL AND ENVIRONMENTAL ENGINEERING<br />

Certificate: Applications of Computing<br />

WATER AND THE<br />

ENVIRONMENT<br />

102<br />

PROJECT TITLE<br />

The Interface of Hydrology<br />

and Machine Learning:<br />

Generating Better<br />

Information for Decisionmakers<br />

and Educating the<br />

Decision-makers of the<br />

Future<br />

ORGANIZATION(S)<br />

Integrated GroundWater<br />

Modeling Center (IGWMC),<br />

Department of Civil<br />

and <strong>Environmental</strong><br />

Engineering, Princeton<br />

University<br />

LOCATION(S)<br />

Princeton, New Jersey<br />

MENTOR(S)<br />

Reed Maxwell,<br />

William and Edna Macaleer<br />

Professor of Engineering<br />

and Applied Science,<br />

Professor of Civil and<br />

<strong>Environmental</strong> Engineering<br />

and the High Meadows<br />

<strong>Environmental</strong> Institute;<br />

Lisa Gallagher, Education<br />

and Outreach Specialist,<br />

IGWMC, Department of Civil<br />

and <strong>Environmental</strong><br />

Engineering<br />

Providing accurate information about our<br />

water sources is important for future decisionmakers<br />

to make wise and sustainable plans<br />

concerning water management. However,<br />

due to climate change, future environmental<br />

conditions have become further unpredictable,<br />

making it challenging to understand our water<br />

supplies. Using both physics-based and datadriven<br />

modeling in combination is essential<br />

for predicting conditions, as both modeling<br />

methods have advantages and disadvantages.<br />

To inform the integration of these methods, I<br />

analyzed the correlations between pumping<br />

data and climate condition variables and<br />

determined how groundwater pumping affects<br />

water table depth. I focused on implementing a<br />

machine learning, data-driven model that used<br />

a regression-enhanced random forest method<br />

to estimate water table depth in New Jersey. In<br />

addition, as part of the educational aspect of<br />

the internship, I helped teach at The Watershed<br />

Institute’s Watershed Academy for High School<br />

Students where I introduced the idea of pattern<br />

recognition and discussed the importance of data<br />

quantity and quality for training, all of which are<br />

fundamental to machine learning. Through this<br />

experience, I learned various practical skills in<br />

machine learning and data analysis which I hope<br />

to utilize for my senior thesis and other future<br />

work.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!