Hydrologists are used to jumping through hoops to access data. But it doesn’t have to be that way all the time! In a single semester, the stellar Tecnológico de Monterrey (ITESM) WaterFront Project Team developed software that will easily display monthly summaries of streamflow at multiple USGS stream gages. As a bonus, we can quickly view flow duration curves for the same gages.

Thanks to the ITESM team’s expertise and hard work, this software will be used to generate Extension hydrology reports for the Platte River in Nebraska. Platte River streamflows are critical for agricultural production and for important wildlife habitat in Nebraska.

The UNL GaugeCam Team, along with Doug Hallum at the West Central Research, Education and Extension Center, presented this challenge to the capstone student group in the Departamento de Computación and Escuela de Ingeniería y Ciencias. ITESM students, led by Professor Elizabeth López Ramos, tackled this project in two phases.

PHASE 1: Gather list of client requirements and develop proposal.

• Requirement gathering and analysis based on meetings with client.
• Research and brainstorm to devise innovative and practical solutions.
• Create a concise proposal outlining solutions, timeline, budget, and benefits for the client.
• Present refined proposal to the client, highlighting alignment with requirements and receiving feedback.

PHASE 2: Build out the software based on the accepted proposal.

• Construct platform, including features and functionality described in the proposal.
• Test software functionality, performance, and security to meet client requirements and industry standards.
• Present the software to the client for feedback and iterate to meet expectations and requirements.
• Deliver a functional Windows installer and documentation to the client.

Just like real world situations, we (UNL) came into this project with many ideas for the team. In other words, we gave them the very real challenge of (1) setting realistic expectations for the client, and (2) helping the client know what they actually want for a final product. And what we got was a streamlined, professional Windows application and good documentation.

Check out the animation below to see some of the WaterFront features.

• Select a range of months to summarize
• Display monthly statistics by hovering mouse
• Display additional USGS gage sites
• Show flow-duration curves
• Download graphics and data

We are truly grateful to the ITESM WaterFront Team for their dedication to this project.

• Joel Fernando Santillán Santana
• Jorge Luis Salgado Ledezma
• Milton Eduardo Barroso Ramírez
• Miriam Paulina Palomera Curiel
• José Ricardo Vanegas Castillo

Special thanks also to Professor Elizabeth López Ramos and Professor Gildardo Sánchez Ante for a wonderful experience working with your class. We hope we can continue working together!

There are six key components to be concerned about when setting up the GaugeCam octagon calibration target in the field.

1. The facet lengths of the octagon must be the exact same length.
2. The background in the stream must be mounted orthogonal to the water surface.
3. The camera should be mounted as directly in front of the background as possible.
4. Field measurements must be made from the upper left vertex to the water level.
5. The size of the octagon (in pixels) in the image must be large enough to be detected by the GRIME2 search algorithm.
6. The thickness (in pixels) of the black border around the octagon must be large enough to be detected.

As you might guess, based on the bold text above, we are focusing on the last two items in this blog post, which are both related to the size of the octagon in images to achieve successful calibration for water level measurement.

To provide guidelines for the octagon dimensions (in pixels) required for successful calibration, we used imagery from the Kearney Outdoor Learning Area (KOLA). The image used was about 1MB when stored as a .jpg, as captured with a Reconyx trail camera.

In this simple test we resampled the images to reduce resolution. As the resolution was reduced, the size of the octagon was reduced. In other words, there were fewer pixels displayed across the width and facet lengths of the octagon. Similarly, the black border around the octagon became more pixelated.

The animation above shows when the octagon search algorithm began to break down as the resolution of the image is decreased. The decrease is represented by the scale annotation in the image. The scale value shown is the size of the image in percentage of the original image size (2304 x 1295 pixels, including the black strips at top and bottom of the image). Major failures of the octagon find and calibration started to occur at less than 60% of the original image resolution. The horizontal width of the blue area in the octagon target is 137 pixels in the 60% image.

Below is a table showing the desired width of the black border around the octagon. The suggested width for robust detection is 15 pixels, although the 60% image had a width of only about 11 pixels. The width of the black border can be measured in your images using the measurement tool in GRIME2.

The major takeaway here is that, whatever the size of the images captured, the current GRIME2 octagon search and calibration algorithm should be robust for images where the blue part of the octagon is at least 137 pixels wide and the width of the black border is at very minimum 11 pixels wide but preferably at least 15 pixels wide. These values are based on a very simple test with an ideal image. Greater sizes (in pixels) of the octagon target are generally going to be preferable, and the overall performance of the system is still highly dependent on image scene quality (fewer shadows, glare, etc.) and proper installation of the target background in the stream.

We had a great online and in-person audience for Ken Chapman’s dissertation defense on Thursday May 8, 2023. Ken gave an excellent overview of his dissertation, as shown below. He will graduate in December 2023.

Congratulations Dr. Chapman!

A pdf of Ken’s presentation is available here.

Ken’s three major dissertation projects have resulted in a robust, free open-source water level measurement software and scientific publications.

More information and the GRIME2 software can be found here.

Publications are as follows:
Chapman, K. W., Gilmore, T. E., Chapman, C. D., Birgand, F., Mittlestet, A. R., Harner, M. J., et al. (2022). Technical Note: Open-Source Software for Water-Level Measurement in Images With a Calibration Target. Water Resources Research, 58(8), e2022WR033203. https://doi.org/10.1029/2022WR033203

Chapman, K., Gilmore, T., Mehrubeoglu, M., Chapman C., Mittelstet, A., Stranzl, J.E. (2023). Is there sufficient information in images to fill large data gaps of stage and discharge measurements with machine learning? PLOS Water [revised, in review]

Chapman, K. W., Gilmore, T. E., Harner, M. J., Stranzl, J.E., Chapman, C. D., Birgand, F., Mittlestet, A. R., et al. (2023). Technical Note: Improved Calibration Target for Open-Source, Image-Based Water-Level Measurement. In prep for Water Resources Research.

Research:

We monitored water quality in the Nebraska Sandhills, combining water sampling with time-lapse imagery from multiple cameras, including two co-located Platte Basin Timelapse cameras. The USGS 104b program supported this project. The animation below illustrates changing water color with changing dissolved organic carbon (DOC)* measured in water samples.

At the Kearney Outdoor Learning Area (KOLA), we installed two different GaugeCam targets as we develop less intrusive methods for image-based water level measurement. This installation, which also includes a traditional water level sensor (HOBO transducer), is the focus of PhD candidate Ken Chapman’s final dissertation chapter. This work is supported in part by the UNL Collaboration Initiative.

In Bazile Creek, we monitored water levels using a camera and GRIME2 software. This work is partially supported by USDA (PR-HPA LTAR Network).

Software Development:

GRIME2 (water level) software was published in a peer-reviewed article.

GRIME2 has now been updated to accommodate both a bowtie target (see right-hand side of KOLA image, above) and an octagon calibration target (left-hand side of KOLA image).

Partnerships:

GaugeCam Team Presentations:

T.E. Gilmore, Stranzl, J., Chapman, K., Harner, M. GRIME-AI Open-Source Ecosystem for Time-lapse Imagery. USGS CDI Script-a-thon, October 11, 2022 (virtual, 50+ attendees)

Stranzl, J., T.E. Gilmore, M. Harner, and K. Chapman. 2022. GRIME-AI software for incorporating information from ground-based camera imagery with other sensor data (talk). Joint Aquatic Sciences Meeting, Grand Rapids, Michigan. May 2022.

Harner, M., T. Gilmore, K. Chapman, J. Bajelan, A. Klein, and C. Wagner. An introduction to the Kearney Outdoor Learning Area (KOLA) for experiential learning in ecohydrological research. 2022 Platte River Basin Conference. Kearney, NE. October 2022.

The GaugeCam (GRIME Lab) team, including PI Troy Gilmore at University of Nebraska-Lincoln and Mary Harner at the University of Nebraska at Kearney, have been using imagery in eco-hydrologic studies and science communication for about a decade. The University of Nebraska is also home to a large time-lapse image archive from high-resolution (DSLR) cameras deployed across the Platte River Basin. Since 2010, the Platte Basin Timelapse project has acquired and archived over 3 million high quality images of water features across the watershed. These images are captured hourly during daylight hours and contain large amounts of untapped scientific information. Our team is devoted to (1) extracting ecological and hydrological information from imagery, and (2) building software that makes these tasks easy for other scientists.

Over the last 9 months, the GaugeCam GRIME Lab has benefited greatly from a fast-developing collaboration with scientists at Tecnológico de Monterrey (ITESM). Our collaboration involves both teaching and research.

Teaching

Dr. Gildardo Sánchez Ante, Professor in the Department of Computation in the School of Engineering and Sciences at the Guadalajara Campus has incorporated image-based hydrology projects in two courses. The GaugeCam team has been joining these classes via Zoom. We have had the opportunity to introduce the students to the dataset and hydrology concepts. We have also heard updates from the student project teams and are looking forward to final project presentations this week. The students are doing a fantastic job of extracting information from imagery and building machine learning models that successfully predict streamflow in the North Platte River in Nebraska!

The two courses where Platte Basin Timelapse-derived data are being used are:

• Advanced Artificial Intelligence for Data Science
• Business Solution Development Capstone project

Research

We see many exciting opportunities for image-based water monitoring, including in Mexico. Based on our background with image-based hydrology projects and our collaboration with Dr. Sánchez Ante, we are also working closely with Dr. Pabel Antonio Cervantes Avilés to set up and pilot camera-based monitoring at a site on the Atoyac River (see video below for background on this river). Dr. Cervantes Avilés is in the Water Science and Technology Group, in the School of Engineering and Sciences at the Puebla Campus.

We are excited to have partners at ITESM with much-needed expertise in artificial intelligence and water quality. We look forward to new insights into water quality and camera-based monitoring approaches in 2023 and beyond.

Since 2010, the GaugeCam team has been working on open-source software for ground-based time-lapse imagery (recently referred to by a colleague as “camera trap hydrology”).

The project started in François Birgand’s lab at North Carolina State University. GaugeCam was my undergraduate research project and Ken Chapman wrote the open-source software to measure water level in highly conditioned time-lapse images. We published a paper in 2013 (Gilmore et al. 2013, Journal of Hydrology) showing potential precision of +/- 3mm, or about the height of a meniscus, in carefully controlled laboratory conditions. Application of GaugeCam showed increased uncertainty in a precisely installed and well-maintained field site in a tidal marsh in eastern NC (Birgand et al. 2022, PLOS Water). Uncertainty is still reasonable compared to other common water level monitoring methods. The current iteration of this mature, open-source software for measuring water level is the GaugeCam Remote Image Manager 2 (GRIME2), as described in a technical note in Water Resources Research (Chapman et al. 2022). More details, including major recent improvements to the ease of installation, are available at https://gaugecam.org/grime2-details/.

More recently, we have been working with Mary Harner (University of Nebraska at Kearney, see https://witnessingwatersheds.com) and her colleagues with the Platte Basin Timelapse (PBT) project (https://plattebasintimelapse.com). PBT has over 3 million high-resolution (DSLR) hourly daytime images from 60+ cameras across the Platte River basin. These images are available to University of Nebraska researchers for research and teaching purposes. Mary and her colleague Emma Brinley-Buckley had previously published research based on PBT imagery, for example, extracting the extent of wetland inundation from imagery. Similarly, Ken Chapman (now a PhD candidate at UNL) used PBT imagery to fill simulated data gaps in USGS stream gauge data (withdrawn preprint in HESS, now revised and in review in PLOS Water).

Thus, the inspiration for GRIME-AI, a software suite that will allow much broader application of time-lapse (camera trap) imagery in hydrological and ecological studies. In short, GRIME-AI is intended to help us move across the spectrum of potential image types that can be acquired from fixed ground-based cameras. The figure below shows this vision, starting with highly conditioned GaugeCam water level imagery on the left, to more flexible (but more difficult to handle) unconditioned imagery on the right. UNL PhD student John Stranzl is the lead programmer on GRIME-AI. More information is available at https://gaugecam.org/grime-ai-details/.

Long-term, our goal is for GRIME-AI is an open-source ecosystem that has the simplicity for new (very low barrier to entry for non-programmers) combined with the opportunity for experienced programmers to contribute to the capabilities of the software. A key guiding concept for GRIME-AI is that it should reduce mundane and repetitive tasks to the bare minimum, while still being powerful and flexible. Our software development philosophy is below.

Early features enabled in GRIME-AI are as follows:

• Automatic download of NEON data
• Automatic download of PhenoCam images for NEON sites
• Image triage (removal of dark and/or blurry imagery), with csv report
• Visualization (overlays) of common edge detection algorithms (Canny, Sobel, ORB, etc.)
• Custom region of interest (ROI) placement on reference image
• Calculation of scalar image features for each ROI and whole image, including:
  • Color cluster analysis for each ROI and whole image
  • Greenness calculations for each ROI and whole image
  • Intensity and entropy for each ROI and whole image
• Generation of csv with all selected scalar image features extracted for folder(s) (recursive) of images

The obvious next steps are:

• Data fusion (combining other sensor data with scalar image features extracted in GRIME-AI)
• Anchoring images/ROIs to account for camera movement
• Capabilities to build classifiers from fused datasets
• Automated image segmentation

In our last post, we promised to tell you about some ongoing work. Our current target backgrounds use pattern matching to find and precisely locate calibration points. Given our experiences with GaugeCam testers, we’re looking to simplify the installation and calibration process without sacrificing too much accuracy. Currently in testing is our “stop sign” calibration target. This feature is not yet live in the GRIME2 software, but we are installing these backgrounds in two locations for testing. At both locations we have HOBO water level loggers for comparison. At one location we will have an adjacent bowtie target installed. Our first location was installed yesterday, with deep gratitude to the landowner who is interested in this work. A photo and time-lapse of the installation are below!

Hydrologic modelers like to say “all models are wrong, but some are useful.”

And more recently (via @bisnotforbella), “All models are wrong but some I am emotionally attached to.”

Those of us working on GaugeCam have an emotional attachment to bowties. Because bowties are the shapes that have helped us model the real world by relating pixels in an image to real locations in the image scene.

This post is the first in a series that will explain the history of this emotional attachment, and why it may change in the future.

First, a step back to 2009. Here are some of the first calibration target designs we tried. Circles, bowtie-ish circles, and secchi disk shapes. Tossed in with some horizontal lines to test our water line finding algorithms.

Our first testing used 5.5-inch circular patterns. We did this work in the lab and in the field. Below is the Pullen Park site where we eventually installed two columns of circular fiducials.

Along with the calibration shapes, we had to test line find algorithms to detect the water line. We had some hits and some misses, both in the lab and field.

In the images above, you can quickly see why we were looking for calibration patterns other than classic staff gauges. Our low resolution images at the time didn’t help, but it was very clear that the intuitive idea of just putting staff gauges in the image scene for calibration was not our best path forward.

By 2011 we had developed and tested bowtie targets using both benchmark patterns (synthetic water lines) and real water levels in the lab at North Carolina State University. We presented these results at the American Society of Agricultural and Biological Engineers Annual Conference in Louisville, KY.

Honestly, these are great memories. This was my undergraduate research and some of the early research in the Birgand Lab at NCSU. We also had great volunteers on the project, most notably Ken Chapman, who brought the machine vision and software programming skills necessary to create GaugeCam.

Bowties are still central to our work. But be sure to check back to find out how we’re moving GaugeCam forward!