à£à£Ö±²¥Ðã

Xufei Yang, a South à£à£Ö±²¥Ðã State University assistant professor and Extension environmental quality engineer, is developing a novel method for assessing surface water quality through drone imagery and smells.

South à£à£Ö±²¥Ðã is home to 11,929 miles of rivers and streams and 577 lakes and reservoirs. But a revealed underlying problems in many of these water bodies.

The 2024 report assessed 6,148 stream miles. Over 70% of the miles showed water quality challenges. It also assessed 180 lakes. Over half of those lakes did not support designated water quality use per standards set forth by the U.S. Environmental Protection Agency.

Assessment practices are essential for identifying water quality issues and managing surface water quality. With a state as large as South à£à£Ö±²¥Ðã, an efficient and cost-effective system for assessing water quality is necessary for identifying problematic waters. Xufei Yang, a South à£à£Ö±²¥Ðã State University assistant professor and Extension environmental quality engineer, is developing an artificial intelligence-powered, noncontact method for the rapid assessment of surface water quality based on imagery and smells. 

"Traditionally, water quality assessment involves on-site surveys using water quality probes and lab analysis of grab samples," Yang explained. "However, the increasing number of samples due to growing concerns over water quality results in substantial expenses and labor, even though many samples exhibit good water quality, making further analysis unnecessary. Therefore, developing a noncontact method for rapid screening and identifying 'problematic' water samples that warrant further analysis would significantly facilitate surface water quality management in South à£à£Ö±²¥Ðã."

People have utilized drones paired with hyperspectral cameras to analyze water quality through images. But this tried-and-true method often faces a challenge: sediments, aquatic plants and other factors causing turbidity, or color in the water, and can lead to inaccurate analysis. Yang believes he can overcome this by complementing the hyperspectral imagery with scent data. Gas sensor arrays — Yang refers to this as an "e-nose" — will be utilized to gather smells from the water, which will be integrated into a system combining both measurements. A neural network algorithm will be utilized to train the system.

Two other SDSU faculty members — ZhengRong Gu and Sushant Mehan — will assist Yang in this project. Gu, a professor in SDSU's Department of Agricultural and Biosystems Engineering, brings a wealth of experience developing sensors and training artificial intelligence models. Mehan, assistant professor in SDSU's Department of Agricultural and Biosystems Engineering and an Extension water resource engineer, has significant experience in water quality monitoring projects.

The team will validate its system through measurements from 150 surface water samples across eastern South à£à£Ö±²¥Ðã. The project, if successful, would greatly expedite the identification of problematic water samples that need further analysis, saving time and money.

"If proven effective, this novel approach will supplement existing surface water quality assessment methods, including both onsite surveys and lab analyses, potentially reducing labor and costs associated with water quality monitoring," Yang added.

In the future, Yang believes this system could be further applied to nonsurface water samples, such as livestock wastewater and groundwater.