Near my childhood home was a small river. It wasn’t much more than a creek at the best of times, and in dry summers it would sometimes almost dry up completely. But snowmelt revived it each Spring, and the remains of tropical storms in late Summer and early Fall often transformed it into a raging torrent if only briefly before the flood waters receded and the river returned to its lazy ways.
Other than to those of us who used it as a playground, the river seemed of little consequence. But it did matter enough that a mile or so downstream was some sort of instrumentation, obviously meant to monitor the river. It was — and still is — visible from the road, a tall corrugated pipe standing next to the river, topped with a box bearing the logo of the US Geological Survey. On occasion, someone would visit and open the box to do mysterious things, which suggested the river was interesting beyond our fishing and adventuring needs.
Although I learned quite early that this device was a streamgage, and that it was part of a large network of monitoring instruments the USGS used to monitor the nation’s waterways, it wasn’t until quite recently — OK, this week — that I learned how streamgages work, or how extensive the network is. A lot of effort goes into installing and maintaining this far-flung network, and it’s worth looking at how these instruments work and their impact on everyday life.
Inventing Hydrography
First, to address the elephant in the room, “gage” is a rarely used but accepted alternative spelling of “gauge.” In general, gage tends to be used in technical contexts, which certainly seems to be the case here, as opposed to a non-technical context such as “A gauge of public opinion.” Moreover, the USGS itself uses that spelling, for interesting historical reasons that they’ve apparently had to address often enough that they wrote an FAQ on the subject. So I’ll stick with the USGS terminology in this article, even if I really don’t like it that much.
With that out of the way, the USGS has a long history of monitoring the nation’s rivers. The first streamgaging station was established in 1889 along the Rio Grande River at a railroad station in Embudo, New Mexico. Measurements were entirely manual in those days, performed by crews trained on-site in the nascent field of hydrography. Many of the tools and methods that would be used through the rest of the 19th century to measure the flow of rivers throughout the West and later the rest of the nation were invented at Embudo.
Then as now, river monitoring boils down to one critical measurement: discharge rate, or the volume of water passing a certain point in a fixed amount of time. In the US, discharge rate is measured in cubic feet per second, or cfs. The range over which discharge rate is measured can be huge, from streams that trickle a few dozen cubic feet of water every second to the over one million cfs discharge routinely measured at the mouth of the mighty Mississippi each Spring.
Measurements over such a wide dynamic range would seem to be an engineering challenge, but hydrographers have simplified the problem by cheating a little. While volumetric flow in a closed container like a pipe is relatively easy — flowmeters using paddlewheels or turbines are commonly used for such a task — direct measurement of flow rates in natural watercourses is much harder, especially in navigable rivers where such measuring instruments would pose a hazard to navigation. Instead, the USGS calculates the discharge rate indirectly using stream height, often referred to as flood stage.
Beside Still Waters
The height of a river at any given point is much easier to measure, with the bonus that the tools used for this task lend themselves to continuous measurements. Stream height is the primary data point of each streamgage in the USGS network, which uses several different techniques based on the specific requirements of each site.
The most common is based on a stilling well. Stilling wells are vertical shafts dug into the bank adjacent to a river. The well is generally large enough for a technician to enter, and is typically lined with either concrete or steel conduit, such as the streamgage described earlier. The bottom of the shaft, which is also lined with an impervious material such as concrete, lies below the bottom of the river bed, while the height of the well is determined by the highest expected flood stage for the river. The lumen of the well is connected to the river via a pair of pipes, which terminate in the water above the surface of the riverbed. Water fills the well via these input pipes, with the level inside the well matching the level of the water in the river.
As the name implies, the stilling well performs the important job of damping any turbulence in the river, allowing for a stable column of water whose height can be easily measured. Most stilling wells measure the height of the water column with a float connected to a shaft encoder by a counterweighted stainless steel tape. Other stilling wells are measured using ultrasonic transducers, radar, or even lidar scanners located in the instrument shelter on the top of the well, which translate time-of-flight to the height of the water column.
While stilling well gages are cheap and effective, they are not without their problems. Chief among these is dealing with silt and debris. Even though intakes are placed above the bottom of the river, silt enters the stilling well and settles into the sump. This necessitates frequent maintenance, usually by flushing the sump and the intake lines using water from a flushing tank located within the stilling well. In rivers with a particularly high silt load, there may be a silt trap between the intakes and the stilling well. Essentially a concrete box with a series of vertical baffles, the silt trap allows silt to settle out of the river water before it enters the stilling well, and must be cleaned out periodically.
Bubbles, Bubbles
Making up for some of the deficiencies of the stilling well is the bubble gage, which measures river stage using gas pressure. A bubble gage typically consists of a small air pump or gas cylinders inside the instrument shelter, plumbed to a pipe that comes out below the surface of the river. As with stilling wells, the tube is fixed at a known point relative to a datum, which is the reference height for that station. The end of the pipe in the water has an orifice of known size, while the supply side has regulators and valves to control the flow of gas. River stage can be measured by sensing the gas pressure in the system, which will increase as the water column above the orifice gets higher.
Bubble gages have a distinct advantage over stilling wells in rivers with a high silt load, since the positive pressure through the orifice tends to keep silt out of the works. However, bubble gages tend to need a steady supply of electricity to power their air pump continuously, or for gages using bottled gas, frequent site visits for replenishment. Also, the pipe run to the orifice needs to be kept fairly short, meaning that bubble gage instrument shelters are often located on pilings within the river course or on bridge abutments, which can make maintenance tricky and pose a hazard to navigation.
While bubble gages and stilling wells are the two main types of gaging stations for fixed installations, the USGS also maintains a selection of temporary gaging instruments for tactical use, often for response to natural disasters. These Rapid Deployment Gages (RDGs) are compact units designed to affix to the rail of a bridge or some other structure across the river. Most RDGs use radar to sense the water level, but some use sonar.
Go With the Flow
No matter what method is used to determine the stage of a river, calculating the discharge rate is the next step. To do that, hydrographers have to head to the field and make flow measurements. By measuring the flow rates at intervals across the river, preferably as close as possible to the gaging station, the total flow through the channel at that point can be estimated, and a calibration curve relating flow rate to stage can be developed. The discharge rate can then be estimated from just the stage reading.
Flow readings are taken using a variety of tools, depending on the size of the river and the speed of the current. Current meters with bucket wheels can be lowered into a river on a pole; the flow rotates the bucket wheel and closes electrical contacts that can be counted on an electromagnetic totalizer. More recently, Acoustic Doppler Current Profilers (ADCPs) have come into use. These use ultrasound to measure the velocity of particulates in the water by their Doppler shift.
Crews can survey the entire width of a small stream by wading, from boats, or by making measurements from a convenient bridge. In some remote locations where the river is especially swift, the USGS may erect a cableway across the river, so that measurements can be taken at intervals from a cable car.
From Paper to Satellites
In the earliest days of streamgaging, recording data was strictly a pen-on-paper process. Station log books were updated by hydrographers for every observation, with results transmitted by mail or telegraph. Later, stations were equipped with paper chart recorders using a long-duration clockwork mechanism. The pen on the chart recorder was mechanically linked to the float in a stilling well, deflecting it as the river stage changed and leaving a record on the chart. Electrical chart recorders came next, with the position of the pen changing based on the voltage through a potentiometer linked to the float.
Chart recorders, while reliable, have the twin disadvantages of needing a site visit to retrieve the data and requiring a tedious manual transcription of the chart data to tabular form. To solve the latter problem, analog-digital recorders (ADRs) were introduced in the 1960s. These recorded stage data on paper tape as four binary-coded decimal (BCD) digits. The time of each stage reading was inferred from its position on the tape, given a known starting time and reading interval. Tapes still had to be retrieved from each station, but at least reading the data back at the office could be automated with a paper tape reader.
In the 1980s and 1990s, gaging stations were upgraded to electronic data loggers, with small solar panels and batteries where grid power wasn’t available. Data was stored locally in the logger between maintenance visits by a hydrographer, who would download the data. Alternately, gaging stations located close to public rights of way sometimes had leased telephone lines for transmitting data at intervals via modem. Later, gaging stations started sprouting cross-polarized Yagi antennas, aimed at one of the Geostationary Operational Environmental Satellites (GOES). Initially, gaging stations used one of the GOES low data rate telemetry channels with a 100 to 300 bps connection. This gave hydrologists near-real-time access to gaging data for the first time. Since 2013, all stations have been upgraded to a high data rate channel that allows up to 1,200 bps telemetry.
Currently, gage data is collected every 15 minutes normally, although the interval can be increased to every 5 minutes at times of peak flow. Data is buffered locally before a GOES uplink, which is about every hour or so, or as often as every 15 minutes in peak flow or emergencies. The uplink frequencies and intervals are very well documented on the USGS site, so you can easily pick them up with an SDR, and you can see if the creek is rising from the comfort of your own shack.
Thank you for this, I’ve often wondered exactly how the data was captured. As a whitewater kayaker I checked the CFS on the river before heading out and there are wonderful prediction sites that will tell you how high the water is going to go in the near future. Great if you like watching huge flow over nearby waterfalls after a big rain. (the river peaks more or less 24 hours after the rain event)
My dad, who was a rafter, used river height. So it was hard for us to explain to each-other what a given river was doing :-)
Interesting. We had several of these gages in the town where I grew up, and while I frequently saw people opening the gages, I’d always wondered why the biggest gage had a small cableway and gondola erected next to it. Makes a lot more sense than my working theory that there was more instrumentation across the river, where there wasn’t easy road access.
Intersting read, especially some of the water height measurement methods were new to me and I’ve been in the business for about 13-years :)
As the guy that is responsible for monitoring and maintaining a BUNCH of these sites – you’re exactly right about the pros/cons of each type of well and the recovery of data.
Happy to answer any other questions you may not have found the information for – the only thing I’ll add off the top is that bubblers require regular calibration based on manual measurements, because they’ll frequently drift over time. We visit our highest priority sites 2x a week to ensure they’re calibrated correctly, and we’re measuring down to 0.01′ accuracy/resolution (about 3.4mm).
Why not use metric system? It’s better suited for scientific measurements and calibrated equipment.
Because USGS standard is decimal feet – so whatever they use is what we (environmental consulting firm) have to use. Otherwise yeah, that would be way more appropriate.
Decimal feet is the standard fitted US civil engineering, that’s not going to change because it’s tied to literally every road, bridge and building in the country. It’s a simple consequence of how standardisation worked out. Notably USGS doesn’t use it when they don’t have to.
The article mentions a short pipe length and a known size orifice. This implies measurements while air flow is present. One could also give a short blow into the pipe just before measuring, eliminating the flow mechanics of the pipe and the orifice (as long as it is pointed downwards). What are the problems with this method?
For the drift I assume, the combination of ruggedness and precision doesn’t allow for “common” pressure sensors like in cheap blood pressure displays which can go without calibration for longer periods of time. What kind of pressure sensors are used here, and are the deviations more of a back-and-forth motion, or keep the values wandering away with age?
It’s mostly to do with real-world things like ice, bugs, mud, etc – stuff that gets into the pipe (also, not really a pipe, more of a 1/4″ ID vinyl tube in a lot of cases). We constantly purge the lines, both manually and automatically, to try to keep them clear. We keep our lines within ~20′ or so, so there’s more flexibility in a lot of cases than the article makes it sound.
That’s also the reason for the drift in terms of bubbler lines – stuff gets in them, they get jostled around by us/weather/bugs/water itself, and the sensors are so incredibly sensitive that it will change the reading the bubbler is getting. It’s not like the sensor itself is drifting, so much as it’s deployed in some really dynamic environments and there isn’t a good way to eliminate that dynamicity, even with a stilling well.
I’m not sure what the exact part# is for the pressure sensors, but I’ve got over 100 of the Solinst LevelLoggers deployed on my sites, mostly in groundwater wells, but also in some surfacewater sites. While we don’t necessarily pay sticker price, they can run between $700-$1,700 a pop, and run unattended/unpowered for years. The manufacturer says they last at least 10 years, but I’ve got some that have been out for 17 and are still going strong – that’s got to be impressive!
Thanks for the details, I wouldn’t have guessed that the pipes/tubes are cause of the drift (I know an installation on a small hydropower plant, the owner used a compressor, a differential pressure sensor from an old washing machine and the smallest standard copper tube available to trigger the rinsing of the …. Rechenanlage — my dictionary trolls me and translates that to “computer”, no other options; I mean the grill fishing out the leaves and bottles. Open the valve for some seconds, then wait some seconds, then start the cleaning if there is a measurable difference before and after the Rechen. He never had any problems with that, but compared to a station in the wild giving absolute values this is just a quick and dirty hack).
For the pressure sensors used, I’m not close enough to the topic to know part numbers or brand names and their reputation :( but I guessed there might be a special measurement principle with interesting advantages and disadvantages, partly because I falsely attributed the drift to the sensor instead of the tube.
If insects are an issue then why have you not considered hanging some flypaper around sensors? A single roll of flypaper can be had on Aliexpress for like $0.45 USD. I find it dubious that richest country on earth would not be able to afford simple piece of sticky paper to catch flies, which is commonly used everywhere, even in Russia, China or Uganda.
it would be under water or inside the tube
it or the insects stuck on it would block the tube
one would have to change it daily
it would (potentially) release chemicals into the environment
it would put a constant load on the local insects population
Very well done – I used to work on building some of these for researchers. I kept wanting to chime in with “but there’s… THIS!” except every time I thought that, it would be in the next paragraph.
But there’s this! Another measurement method is capacitive: http://www.geoscientific.com/dataloggers/AquaRod_Freeze_Tolerant_Water_Level_Recorder.html The precision was nice when we didn’t need a lot of measurement range – i.e., not a great choice for flood monitoring down where we were but so sensitive we could just about measure wind speed with it.
And if your ultrasonic sounder goes bananas, clean the spiderwebs out. So many spiderwebs. :-)
GOES is not the only game in town for relaying data. Many, many gages transmit their data over VHF radio using the ALERT or ALERT2 protocols. By now, everyone may have made the switch to ALERT2 since it gives higher capacity and better weak signal performance.
And… I’m old and I forget stuff now. ALERT/ALERT2 is used for RAIN gauges, not stream gages.
ALERT can certainly transmit water level data, as well as wind speed and direction if need be. And Alert 2 only improves on its ability to do so. The ELPRO units I service as a hydrographer historically can accept quadrature rotary encoders and 4-20mA signals. The old ALERT units have a limitation that the data is transmitted as a 8 bit sensor address then a 8bit data packet which is modulated and transmitted at 300baud. This means the range of the sensor needs to be split into 2047 units then decoded at the other end with a computer. Typically this results in 4 significant figures of data so xx.xxm or x.xxxm. The system is usually set to transmit on a change of 10 or 20 mm for flood monitoring purposes. The ALERT protocol means that a single site may have 3 or so data address as each variable it transmits needs a unique address. So battery might be 1000, rainfall 1001 and stage 1002 ect.
Alert 2 is a significant improvement as the equipment is much newer, transmits at a faster baud rate, reads sdi12 sensors, transmits in SI units and has a single data address for each site. It can send a single message that has all the variables the site can monitor and can transmit arbitrarily long messages depending on how many variables your sdi12 sensor produces.
Wonder if it is possible to intercept and decode the signal from these with a ham rig so I can tell if there’s water in the creek to swim in before I hike down there. There’s a web portal for the system, but it is terrible and unreliable
I find the web portal good and quite reliable, and improving as well.
I visit this site almost daily in the winter time.
https://waterdata.usgs.gov/monitoring-location/11173200/#dataTypeId=continuous-00065-0&period=P7D
Thank you, very interesting topic and well-written article.
This article brings back some good memories back when I worked as a design engineer for a company that designed and manufactured products used in stream gauging and water monitoring.
I was the lead design engineer and work on both a GOES DCS transmitter design and Bubbler with an integrated pressure transducer.
Data telemetry through GOES is very popular at the time due to little to no reoccurring data service costs. The big caveat was you need to be a government entity or sponsored by one to be assigned an GOES ID.
At the time 100bps using BPSK modulation still was widely used and being migrated to 300bps which uses QPSK modulation. Eventually all the 100bps sites were migrated and 300 bps and 1200bps channels were split into two 300bps channels with the V2 spec. Interestingly the main difference with V2 was just a tighter filtering of the side lobes.
The 1200bps assignments were available but were difficult to get and primary were assigned to NOAA affiliated sites. It sounds like this has loosened up.
The transmitter that we designed was a SDR modulator made by Intersil. I ended up adding support for other satellite protocols (EUMETSAT, INSAT) and were working towards others before the company was bought and subsequently shut down.
At the time I was big into exploring the backcountry and one weekend I was bored, so I wrote a firmware for the goes transmitter to act as a GPS tracker so my wife could keep track of me when I was off in the wilderness. Basically, the firmware would transmit my obfuscated location every 500ft of movement to the GOES WEST satellite on the random test channel. It allowed wife to get a trail of breadcrumbs of where I was and if I was still moving.