Open-Source Underwater

Instrument design aids Northwestern professor’s exploration of underground world


Divers use lights to illuminate Hoyo Negro, an underwater cave on Mexico’s Yucatán Peninsula. Photo by AP Photo/Roberto Chavez Arce via Science

The cave system of Rio Secreto in Mexico lies underground in impenetrable darkness, locked away from all natural light. The creatures within have adapted to this environment. Blind fish once had eyes but evolved to lose them. Helmet lights give at most a pale glow as guides while give tours. But there is another area, off of the beaten path, touched by a light even more ephemeral. In one chamber, a multitude of small, green LED lights can be seen blinking from time to time. Each light is connected to a drip counting sensor. The sensors are placed beneath a stalactite. Each time that a drop of water hits the drum of the sensor, the data for the drip time is logged and the LED light flashes as an indicator that the data logger has sensed the drip and that the replenishment of critical groundwater supplies is being monitored.

These data loggers are created by geologist and cave diver Patricia Beddows, an associate professor of instruction in Earth and planetary sciences at Northwestern University. The sensors measure the “recharge” of the cave — how much water is making its way into the underground tunnels and subsequently, the ground water aquifers from the surface above. These sensors are part of Beddows’ research into the water balance in the Yucatan Peninsula’s cave system, along with ones that measure the water depth.

Beddows has data loggers in dozens of other unlikely places. There are data loggers on the tops of rooftops of dive shops and local houses measuring rainfall, humidity and barometric pressure. Sensors measure water flowing through mangrove swamps where the muck reaches waist-high and log other parameters in dead zones where desiccation cracks reveal water’s long absence. Flow and temperature meters are mounted on the walls of underwater caves and at coastal outflow points to the ocean. Beddows estimates that she has at least 160 data loggers out in the world. What’s more impressive? All of the data loggers that Beddows has deployed have been created from a workshop in her house, which is actually why she is able to deploy so many. Eliminating the high costs associated with buying commercial sensors, she is able to create a variety of customized sensors that are designed exactly for a specific application. The commercial sensors have price tags ranging from $10,000 to $40,000.  Beddows’ sensors range from around $20 to $200 in material costs. The data loggers she has created for The Cave Pearl Project are all open source. This year, Beddows and Edward Mallon published a paper in Sensors titled “Cave Pearl Logger: A Flexible Arduino-Based Logging Platform for Long-Term Monitoring in Harsh Environments.”  So far, the paper has had more than 14,000 views and more than 6,000 downloads. If someone else wants to create their own data logger, there are simple step-by-step instructions published online that they can follow. Which is exactly what Beddows’ students do in class to create their own data loggers.

The Yucatan

Let’s take a step back to talk about the Yucatan Peninsula, the area of Mexico in which Beddows does most of her research. The Yucatan Peninsula is a carbonate platform. The carbonate was created by microorganisms called diatoms that float in shallow marine water. These diatoms pull carbonate out from the water around them and eventually sink to the bottom, piling up into a muddy sediment that cements itself into rock. When global sea levels fell about 10 million years ago, the land of the Yucatan Peninsula emerged from the ocean. The salt water of the ocean is beneath the land. As it rains, that water starts to move down through the rock as well. Eventually there is a layer of stratified water inside the rock, where the freshwater sits on top of the heavier saltwater.

Pearl device

Edward Mallon documents the condition of a water level sensor after retrieval from Jaguar Cenote

When the two waters mix, the brackish mixture cuts into the carbonate to carve out a cave. The more caves that exist, the more the water is able to flow and mix, expanding the voids. Over time the peninsular became riddled with these cave systems that now contain the only freshwater in the area.  There is no surface water there, like rivers or lakes. The freshwater flows underground through the cave systems and out to the ocean. The vegetation also plays an important role in the water cycle, as some of the water is taken up by the trees and evaporated out into the atmosphere.

“We are still struggling with absolutely fundamental questions of how much water is there,” Beddows says.  “One of the things we are finding is that we’re missing about 30 percent of the water that needs to be in the system.”  Where it belongs in the system is a mystery.

Beddows says that a lot of publications infer that the majority of the rainfall in the area of the Yucatan Peninsula is undergoing evapotranspiration, returning it to the atmosphere.

Beddows has traced this idea back to a paper from 1976 where a geologist came up with an estimation that was so high he shaved the number down. The model created in this paper has been repeated by scientists since then. Beddows aims to create a more accurate model with the data she is gathering.

“I’m looking at flows every step along the way. I want to know what’s happening at the water table. I want to know how high the water table gets, how low it gets, why it’s being pushed around,” Beddows says. “I track the water flows through the caves. I look at mixing and then all of the way out to the coastal discharges. And so I am building from the bottom up, what can be called a critical zone observatory. I am looking at all the different components of these fluxes through the system from the atmosphere through to the coastal discharge and all the steps in between.”

Off the Shelf

Beddows started her research using commercial sensors, many of which were very expensive. During her PhD research, when she needed to measure flow in caves, she had to use sensors developed for use in the ocean. They were large and heavy and dangerous to install because they had to be anchored into either the ceiling or floor of the cave using lead weights, all while underwater. The first flow meters she used were also unable to get data when the water started to flow in the opposite direction of the sensor’s setting.  Some of the sites in which she deployed flow meters were out in the jungle and required a lot of effort to transport.

Even when an instrument was able to collect data through a flow reversal, she still had issues. One that was installed in a location called The Pit required a multi-day expedition where Steve Bogarts, the diver who was installing the sensor, had to dive down so far that he needed decompressing underwater before he could return to the surface safely. When Bogarts repeated the dive a month later to retrieve the instrument, the team found that the sensor hadn’t logged any data points. The battery had disconnected at some point in the sensor’s transport, but there was no way to tell. “Steve literally risked his life and we didn’t get any data,” Beddows says.

Another problem with commercial instruments is that they are prone to flooding. “It became financially hazardous to use them because they were designed in such a way that they weren’t withstanding the constant opening and closing I needed to do and the flooding incurred a lot of expenses,” Beddows says.

Building Sensors

Beddows then started to build drip sensors with custom PCBs, or circuit board electronics, so she could afford to create a wider network monitoring cave drips at 21 different sites. She was able to obtain good data, but still had a higher rate of failure from the sensors she built. “Batteries ran out. Corrosion was a constant problem. It was next to impossible to keep them sealed properly. And it turned out that the instruments were of great interest to cave rats and other rodents in the cave that would chew on cables,” Beddows says.

As her research progressed, she continued buying commercial instruments and experiencing failure with them at high cost, limiting the number she could buy, and therefore the amount of data sampling that she could do. Beddows compares the sampling with sensors to a physician taking blood samples. The time of day and location that a blood sample is taken from can cause variation in the quality and content of it. In the same way, the location and time in which you take samples from groundwater in cave systems also matters. If you only take one sample, there is a lot of data you will miss out on. “Even if I don’t get as many decimal places in the quality of each data point, I can actually start asking bigger and better scientific questions and get better answers by just having more sensors throughout the system,” Beddows says.

Beddows and Mallon

Beddows and Mallon set up an “alpha” prototype

While Beddows started making some of her own instruments, with the help of husband, Edward Mallon, she was still largely using commercial instruments. But she was using and also teaching students how to use Arduinos, a kind of microcontroller, to build instruments. One probe she owned which cost $10,000 died while two other people were using it. “That was extremely painful to literally in a matter of minutes, lose $10,000 worth of instrumentation,” Beddows says.  She explained this frustration to her husband.  “And Ed at that point had been involved in some of the Arduino work that I was doing with my students. He was like ‘I’m sure we can do this better!’ And so by the end of the first day he mocked the first prototype that would ultimately become the Cave Pearl,” Beddows says. From that point, The Cave Pearl Project was born. It was made from a plastic cup and its batteries were kept in by Legos. It ran for only six hours.

“He and I are thoroughly convinced that none of this would exist if it weren’t for the combination of the two,” Beddows says of Mallon. They both continued to invest in the project, making it open-source.  Mallon is inspired by people such as Joshua Pierce, who believes in making 3D printed open-source hardware a norm in science.

Cave Pearls

The Cave Pearls were named for a calcite formation in caves. A cave pearl is made when a grain of sand becomes coated with calcite from a drip of water above. The particle in calcite moves around in a body of water and builds up more layers of calcite over time, eventually resulting in a pea-sized white or brown cave pearl that rests in a nest.

Along the way, The Cave Pearl Project has been fully transparent, with blog posts walking anyone interested in how to build and create their own Cave Pearl data logger. The hardware for the Cave Pearls was chosen with a DIY ethos in mind. Many of the parts were inspired by how materials are handled during cave diving and other cave diving equipment. The designers also considered how those ideas translated to what could be bought at Home Depot.

 We’ve got about 15 people who have categorically consumed the webpage and are just building,” she says. A man named Brian Davis has Facebook pictures of his Cave Pearls in his living room, with his fireplace behind. There’s a man in Norway who is using a Cave Pearl for measuring light levels in an estuary. Strangely enough, there even is a person trying to sell a version of the Cave Pearls to interested parties.

One of the biggest challenges has involved the selection of materials used for the data loggers. Beddows and Mallon had to go through multiple types of epoxies before they found the one that worked best for underwater deployment. They also had issues with what to use for anchoring the data loggers to different surface and failures due to biofouling. Even with their current setup, units taken out of the water have a large amount of biological growth on them that needs to be cleaned off with acid before the loggers can be redeployed.

Another issue is the calibration of different units with sensors such as temperature strings. Calibration ends up being done in the kitchen in boiling water and freezing water baths. A large number of sensors are dead on arrival and need to be thrown out. Testing all the sensors and components is a large time sink. “We’re immunized from a lot of this when we buy commercial equipment,” says Beddows. Each of the components is tested and thrown away beforehand if it doesn’t work. “But with something like building a data logger, you are the process,” says Beddows. A huge amount of time is spent testing each component before any building can happen. Although the units are much cheaper than commercial data sensors, a lot of time needs to be spent creating them.

But the data sensors allow for easy and cheap detection and correction of malfunctioning parts, rather than having to replace an entire expensive unit. The user is able to test to find a bad component and replace it at a low cost.

Creating for the Future

Each unit undergoes multiple rounds of experimentation and testing in the field to fine tune its settings. In April, two different alpha units were tested. One was a vertical flow meter for use on small coastal discharges in the shallow bay to the north of the Yucatan Peninsula, a sampling situation for which there is no commercial instrument. The aim was to get the total flow volume from the discharge hole. The components were truly DIY, consisting of a large yellow chemical funnel that was connected to an impellor, which looks like a fan, along with a large tarp that belonged underneath it, anchored down by rocks and bricks. A component with a LCD screen to read out data was connected via a cord.

Unfortunately, the flow from the tiny hole, the circumference of a softball, was too strong to be contained by the tarp. Along with that, small pieces of rock got stuck in the impeller, paralyzing movement. Three scuba divers tried to hold the prototype in place and it wasn’t enough.

“I had three cinderblocks holding down the tarp in front of me and I lifted up,” says Beddows. “While I’m doing this, the pressure of the water rising up lifted the tarp up and the tree blocks I had on the tarp came tumbling towards my face. I’m swimming backwards as these three cinderblock cement blocks are coming towards my face and I look over and Natalie’s having the sample problem.”

Beddows and Mallon quickly developed ideas to improve it. One is to wedge a metal rod between each wall of the discharge hole and then mount the data logger on that rod. They also are looking at different kitchen objects to use as a way to stop pieces of rock from getting into the impeller, such as tea diffusers and wok lifts. They said it usually takes about three prototypes to get one that works the way they want.

The other data logger they setup was one able to measure the volume of multiple drips within a cave from a large column rather than a single stalactite. A frame of PVC pipes was setup, held in place with bricks that were held to the frame with zip ties. The frame held an canvas shopping bag that served as a funnel for the water, where it would be directed through an impellor and then onto a drip sensor below. Two of these data loggers were setup and left behind.

Beddows with student

Beddows helping a student make data loggers in class

Problem-Based  Solutions

In one of her classes, Beddows is teaching the next generation of scientists to create their own data loggers. They work through a method that she uses called problem-based learning, where they come up with questions they want to answer, and what kind of data they want to collect and then design their datalogger to do that job. “It’s a bit painful when you’re actually living through problem-based learning.  Instead of being given all of the components and being given really strict instructions, you have to say what is this doing? Why is it doing it? Is this what I expected?” says Beddows.

Her students have a wide variety of projects. One wants to measure temperature and humidity in beehives. Another student wants to create a sensor that could detect light in an underwater cave. One student was working on a sensor that would detect glacial movement, letting researchers know when glaciers are forming or melting.

Madeleine Lucas didn’t know what the class was when she registered for it. “I realized it was coding, like building electronics and I have never done any of this before because I’m just a sophomore. So it was kind of shocking but like once I started getting into it, I realized it was something that I could definitely do and it’s something I want to keep on doing,” she says.

One student, a theater major, took the class for a second time. She wanted to create a sensor to look at internal tree temperature, soil temperature, and outside temperature to complement class, the ethno-biology of maple syrup. She said that working with electronics helped her with the wiring of lightboards in the theater.

Beddows says that her goal for the class is to empower students to think more analytically. “Even if the people in this course never build an instrument again, even if they never touch any primary electronic components again, part of the goal is that they will be empowered to be better data users.”

“I’m really excited to see where this goes and the interplay between the research needs and research investment coupled with the teaching and then developing the materials for people to learn how to build these in short time. Including people who have no prior electronic or coding experience,” says Beddows.


Add new comment

This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.