Welcome to EnviroDIY, a community for do-it-yourself environmental science and monitoring. EnviroDIY is part of WikiWatershed, a web toolkit designed to help citizens, conservation practitioners, municipal decision-makers, researchers, educators, and students advance knowledge and stewardship of fresh water. New to EnviroDIY? Start here
Tips & Pointers

Evaluating Maxbotix Ultrasonic Sensors for Monitoring Stream Base Flow

Evaluation by the Save our Water, Watershed Hydrologic Analysis Team, Marion Waggoner and Dave Yake.

Background

Broad Run drains a 10 km2 watershed straddling southeastern Pennsylvania and northwest Delaware. Commercial groundwater withdrawal from the aquifer underlying the Broad Run watershed has been approved, but questions about the effect of this withdrawal on Broad Run flow and water quality remain. To address these questions, our organization (Save Our Water) started a stream monitoring program on Broad Run in late 2014.

Our primary objective was to establish a baseline of stream discharge throughout the year that could be used to evaluate the effect of commercial pumping when it begins.

We used an acoustic doppler velocimeter (ADV) to measure the velocity of water flow and calculate discharge. We then developed a stream rating curve to relate water stage (depth, an easy-to-continuously-measure parameter) to water discharge (volume per time, a more difficult to measure parameter but more relevant to our objective).

Our rating curve allows us to continuously predict water discharge from water stage. Multiple rating curves were developed at several locations with particular emphasis on the low flow periods because these periods are when the effect of commercial pumping from the aquifer is expected to be most apparent.

Real-time continuous data from New Garden Township’s EnviroDIY monitoring station allows us to convert stream depth measurements to discharge using our rating curve, but this site is upstream from the commercial well. We needed additional stream flow data from further downstream but did not have money for a submerged pressure (depth) sensor.

We decided to evaluate the Maxbotix ultrasonic sensors for possible monitoring of baseflow stream levels.

Equipment List and Setup

We selected the MB7389 sensor for distance resolution and weather resistance. We initially tested just one sensor but added another when unexpected issues surfaced.

The MB7389 sensors were tested by pointing them downward at a stationary dry target. In the stream, sensors were mounted on standpipes and, later, on a boom extending over open water. We limited the standpipe diameter to 4 inches because a larger standpipe would have been too disruptive to flow in a small stream and would catch lots of debris.

Reviewing the Results

The Positives

Delivers stable data: The two sensors delivered very stable distance data on test stands indoors — stationary plates as targets — at a constant ambient temperature. Generally, over many hours and at a distance near 1 meter, the values varied only about +/- 1 to 2 millimeters.

Functionally impeccable: Both sensors continued to function after extensive time outdoors in operation at temperatures from below 0 degrees Celsius to as high as 39 degrees Celsius.

Extremely rugged: One sensor was washed away during a huge flood on August 7, 2020. A couple of days later we found the sensor and the mounting washed up on the bank in a pile of debris. In spite of the prolonged immersion, the sensor was completely functional and still very accurate.

Remarkable performance: After mounting the sensors over open air above the water, one sensor has operated for nine months, taking ten data points (for averaging) at each 5 minute interval, without returning a single “bad” data value. This is remarkable performance compared to the earlier erratic problems with “bad” data when mounted in standpipes.

Low cost: The sensors cost about $125 each.

Very good depth change performance: During a runoff event where the temperature changes are minor, the Maxbotix sensor delivers very good depth change performance as illustrated in the chart below. At peak flow there is about a 21 millimeter average difference between the Maxbotix and Onset Depth sensors out of a 370 millimeter rise in depth (or about 5.7%). The Onset depth sensor was located on the side of a large stake and could have detected some depth depression at higher peak flow compared to the Maxbotix which was located 2 feet downstream where there was less surface turbulence. Therefore, the 5.7% difference at peak flow is likely a real detected difference in stream depth. Overall, when the Maxbotix is used to measure stream surface height changes using a boom configuration performance is excellent. However, using the Maxbotix as an absolute measure of distance is complicated by a strong temperature variance bias.

During a runoff event where the temperature changes are minor, the Maxbotix sensor delivers very good depth change performance.

The Negatives

“Bad” distance data: When mounted in standpipes, the Maxbotix MB7389 sensors sometimes delivered “bad” distance data — generating values of 300 millimeters when the range was much larger than 300. On rare occasions, the distances might be intermediate between 300 and the real value. Note that these MB7389 sensors are not very accurate below 500 millimeters (50 centimeters) but range well out to about 5 meters. The lower cutoff is approximately 300 millimeters.

We noted that the “bad” data seemed to occur at times when the temperature was dropping and believe that the issue was condensation on the walls of the standpipes. Therefore, the situation should have improved when we insulated the exterior of the standpipes with bubble wrap (would not waterlog) which was over-wrapped with foil. Unfortunately, this helped only marginally. The “bad” data might occur for a single interval or might occur for many consecutive intervals for long time periods (up to a few hours).

We tried adding delays between startup and taking data (to allow the sensor to stabilize) and start to average/integrate distances with built-in software. No improvement was seen.

We had to go to a boom mount of the sensor in open air to eliminate bad data.

Whether mounted over a standpipe or boom-mounted in open air in a stable test stand at a fixed distance to the target, the sensors show large changes in reported distances (as much as +/- 1 centimeter or more) with normal ambient temperature cycling. In fact, we observed this in three conditions:

  • Using the sensor’s internal temperature monitor/ correction for temperature.
  • Using the Maxbotix external temperature sensor for temperature compensation.
  • Attempting to correlate the Maxbotix temperature compensated data with either the Mayfly board temperature or a DHT temperature and % RH sensor. Those results will be discussed along with data in the next section.

Test Data versus Temperature

The issue with temperature seen with the MB7389 sensors is well illustrated in the three figures below. The results are from a stable test stand outdoor mounting over a stationary plate target when the weather was warm/sunny days and cooler nights. The MB7389 was equipped with the Maxbotix external temperature sensor (HR-Maxtemp sensor) which was mounted in the shade under the logger box. The logger box was covered with foil to minimize solar heating of the logger box interior.

Note the +/- spread of the distance data above at any given temperature and % RH. (Click on an image to enlarge.)

A “Multiple Regression Analysis” of the distance vs. Mayfly board temperature or DHT temperature, and percent relative humidity delivered a root squared error the same as using only board temperature: 37%. As would be expected from theory, the relative humidity has very low significance.

Conclusions

  • If used in a boom mounting to eliminate issues associated with condensation inside standpipes, the MB7389 sensors reliably deliver distance data over a wide temperature range and other weather conditions. Furthermore, the sensor delivered excellent stream depth change performance because the temperature change during the rain event and runoff was minor.
  • In spite of using the Maxbotix external temperature sensor for better temperature compensation, the distance data (fixed test stand) vary significantly with changes in ambient temperature — as much as +/- 1 centimeter or more. It is questionable whether the Maxbotix temperature compensation software can be considered worthwhile since the same results were seen for both ultrasonic sensors.
  • Work to correlate the distance changes with other measures of temperature (Mayfly board temperature or an independent temperature as from the DHT digital temperature and humidity sensor) could account for at most 40% of the changes.
  • We recommend careful consideration of these results before deploying these sensors if the objective is to have an absolute measure of depth/distance. However, the sensor is excellent for measuring changes in depth over a time frame where the temperature is not changing significantly.
4 1 vote
Article Rating
Subscribe
Notify of
4 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
neilh
5 months ago

Thankyou so much for doing such detailed work. I’d been thinking of a boom situation for some measurements. However the atmospheric temperature will vary.

James_NZ
5 months ago

Great work guys. We are also interested in these sensors and have also been considering a boom-type setup.
I think it might be worth getting one of these to see if it’s fit for our purpose.

Rory G
3 months ago

Hey Dave,
As part of my technical institute’s graduation requirements, I’m currently designing a project to test the accuracy of MaxBotix’s sensors (as well as a $75 vented pressure transducer) to determine their applicability for streamflow monitoring. I’m also a hydrometric tech so do put a lot of thought into collecting accurate data.

With that said, I do have a couple of thoughts to consider in regards to how your test was designed.

Here they are in point form for brevity’s sake:

  • The U20-001-04’s (13 ft range) accuracy is 3-6 mm, so that needs to be considered. It also needs to be compensated with barrometric pressure (not mentioned in the report) which adds additional inaccuracies.
  • When measuring water in a stilling well with a pressure transducer, and the surface of the water with an ultrasonic sensor, you can’t consider the two data sets as reporting the same values. In high-velocity situations, you can get positive or negative pressure, thus lowering or raising the water level, in your stilling well depending on how it is designed.
  • And most importantly, in your on-ground test, the image shows that the sensor is on a metal boom. This is problematic as long metal objects deform/bend with temperature. You metnioned that you also did this test with the sensor at the top of a stilling well, but I didn’t see those images.

Anyhow, I don’t mean to be a downer. I basically just listed these points as I’ll be critiquing your work in a literature review as part of my project.

Really though I think it’s great that you’ve gone to all this work to test the sensor. I learned a few ideas from reading this that I’ll be incorporating into my tests.

Give me a shout at gallaugher.consulting@gmail.com if you are up to chatting about your tests and helping me to design mine.

4
0
Would love your thoughts, please comment.x