OSW

SIGNATURE WORK
CONFERENCE & EXHIBITION 2022

Water Management Assessment using Satellite Imagery and Deep Learning

Name

Aya Lahlou

Major

Data Science

Class

2022

About

Aya Lahlou is a senior student from Morocco studying Data Science. Aya has been researching applied machine learning throughout her undergraduate career before discovering her passion for applied machine learning in the water energy nexus. She has been pursuing research at the Duke Energy Analytics Lab in parallel with her signature work research. Aya Lahlou will pursue a Doctoral degree in environmental engineering at Columbia university researching applications of machine learning for climate modeling.

Signature Work Project Overview

Improving water and sanitation quality can contribute to five of the 17 UN Sustainable Development Goals (SDGs). However, in many countries, the lack of necessary data or monitoring methods to measure water and sanitation quality constitutes a major hurdle in tracking progress toward these goals. This work demonstrated the use of publicly available satellite imagery data for the prediction of water and sanitation quality estimates. We trained convolutional neural networks (CNN) on satellite images from Landsat 9, Landsat 8, and Sentinel 2 with corresponding ground truth labels from the Afrobarometer Round 6 survey to assess a region’s water and sanitation quality, access, and reliability. Our models reached an accuracy of 0.80, which is a significant improvement from the existing analysis of water and sanitation quality assessment. Additionally, we investigated the predictive ability of CNN and satellite imagery on other water and sanitation markers such as county-level water use, sanitation quality, and the geographical domain adaptability of our methods. This project offers a new perspective in the domain of water and sanitation quality assessment with real-time, faster, and cheaper estimates in all regions of the globe.

Signature Work Presentation Video