Developing a High-Frequency Water Quality Data Collection System based on in situ Spectroscopy for the Advancement of Hydrological Science and Water Resource Management

Poster 211 – Click on poster below to view presentation from author.

Click on poster to view presentation from author.

Joe Carter

University of Florida

Co-Authors: Eban Bean, Aditya Singh

Progress in hydrological science and water resource management is largely limited by the rate at which water quality data can be collected and transformed into actionable information. Water quality data is needed to understand the overall health of water bodies and the underlying physical, chemical, biological, and ecological processes so that resource managers can make well-informed decisions. In recent studies, spectroscopy-based sensing technologies have been used to generate high-frequency data for multiple water quality parameters. While several devices designed to perform in situ spectroscopy are currently available off-the-shelf, their broad use is limited by cost (in the range of $20,000) and lack of control over how the system operates and manages data. Recent technological advancements have enabled investigators to custom build spectroscopy-based in situ water quality data collection systems. A low-cost (~$2,500) benchtop prototype was developed which can detect multiple water quality parameters using a miniature spectrophotometer and light source housed in a 3D-printed enclosure and controlled using a microcomputer. In the future, this system will be adapted to perform high-frequency collection of water quality data, in situ, and used to investigate the effects of data collection frequency on hydrological model performance in urban settings and to analyze nutrient dynamics within urban stormwater infrastructure. Beyond the goals of this study, the same data collection framework can be used in future applications that require highly controlled in situ water quality data collection, such as adaptive control of stormwater systems, watershed fingerprinting, and improving pollutant detection through advanced statistical techniques.

Post comments and questions for author below.

All posts are publicly visible after review by site administrator. Students’ responses to posted questions is factored into scoring for the poster competition. Finalists announced May 25 and awards presented May 26, 2021.

4 thoughts on “Developing a High-Frequency Water Quality Data Collection System based on in situ Spectroscopy for the Advancement of Hydrological Science and Water Resource Management

  1. Hi Joe, very important and useful work you are involved in. I was wonder if you could comment on the potential effects/disturbance from ambient water characteristics, such as color and salinity. Also, do send us a few sensors down to Tampa when you are ready!

    Like

    1. Hi Mauricio, thank you for the comments/questions! Color and salinity could both affect the absorbance spectrum of the water. However, since most of the absorbance takes place in the UV range, color, being a characteristic of the absorbance in the visible range, may not necessarily decrease the performance of the system. Also, this method has been shown to work well in tidal systems with highly variable salinity (Etheridge et al., 2014). In short, the device needs to be calibrated for every unique environmental setting because a practical general model for relating absorbance spectra to solute concentrations has not yet been developed.

      Like

  2. Great study, and a very nice poster! I like the simplicity of your presentation on the topic. I’d be a little cautious about the nitrate results, where the goodness of fit might come most from the clustering of your data in two groups. Do you think more samples across a broader range will help with prediction?

    Like

  3. Hi David, thank you for the comments/questions! Yes, it is true that the higher concentration nitrate samples have high leverage in the model and are causing the r-squared value to increase. The model was also fitted using just the lower concentration samples, which resulted in a slightly lower r-squared value. Right now, I would be wary of using this device/model to make predictions outside of the geographical and temporal context that it was developed in. So, yes, having more samples from a broader range of settings could help in developing a more general model for making water quality predictions. The performance of the general model might not be as good as a site-specific model, however.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: