Accurate measurement of the amount and timing of surface runoff at multiple
scales is needed to understand fundamental hydrological processes. At the
plot scale (i.e., length scales on the order of 1–10 m) current methods
for direct measurement of runoff either store the water in a collection
vessel, which is not conducive to long-term monitoring studies, or utilize
expensive installations such as large-scale tipping buckets or flume/weir
systems. We developed an alternative low-cost, robust and reliable
instrument to measure runoff that we call the “Upwelling Bernoulli Tube”
(UBeTube). The UBeTube instrument is a pipe with a slot machined in its side
that is installed vertically at the base of a runoff collection system. The
flow rate through the slot is inferred by measuring the water height within
the pipe. The geometry of the slot can be modified to suit the range of flow
rates expected for a given site; we demonstrate a slot geometry that is
capable of measuring flow rates across more than 3 orders of magnitude
(up to 300 L min
Surface runoff, or overland flow, is a fundamental process of interest in hydrology. Surface runoff generation can occur at multiple scales, ranging from small pools of excess water that propagate downhill to stream networks that drain large catchments (Horton, 1939; Betson, 1964; Hewlett and Hibbert, 1967; Dunne and Black, 1970; Goodrich et al., 1994; Van de Giesen et al., 2000; Stomph et al., 2001, 2002a; McDonnell, 2003; Descroix et al., 2007; Blume et al., 2008; McGuire and McDonnell, 2010; Ali et al., 2013; Jones et al., 2013; Radatz et al., 2013; Steenhuis et al., 2013; Blair et al., 2014; Stewart et al., 2014). Runoff is a primary cause of erosion and can drive nutrient losses from watersheds (Aksoy and Kavvas, 2005; Butler et al., 2008; Nearing et al., 2011). Accurate measurement of runoff quantity is therefore vital to understand the mechanisms and effects of overland flow.
A number of instruments have been used to quantify runoff. At the plot scale
(encompassing lengths on the order of 1 to 10 m), the most basic measurement
method involves diverting flow to a barrel or similar structure (Hudson,
1993; Meals and Braun, 2006; Dosskey et al., 2007). Water quantity,
chemistry and sediment measurements can then be taken on the collected
water. This setup is typically inexpensive and easy to install, but requires
that the barrels be periodically emptied if long-term monitoring is desired.
Alternative systems have been designed to mitigate these problems, including
dividing flow into multiple containers (Pinson et al., 2004), or using
electronic water sensors (Srinivasan et al., 2000) or tipping buckets
(Hashim et al., 1995; Yu et al., 1997; Zhao et al., 2001; Nehls et al.,
2010). Flow dividers still necessitate the capture and storage of the runoff
water, while the electronic sensor system only detects the presence or
absence of surface flow. Tipping bucket systems are self-emptying and can be
used for long-term deployments, but may have significant error at both low
and high flow rates. For instance, the Belfort-type tipping bucket rain
gauge was shown to have a per-minute accuracy of only 12 mm h
V-notched weirs and flumes have also been used to measure runoff at the plot
scale (Hashim et al., 1995; Radatz et al., 2013), as well as for
measuring surface runoff in larger catchments (Hudson, 1993). However,
these installations are often expensive, with a per-plot cost that can
exceed USD 5000 (Pinson et al., 2004). Further, maintaining the
required up-stream condition of the bed being well below the notch of the
weir requires frequent maintenance in natural streams. Finally, Stomph
et al. (2002b) designed a flowmeter to measure small discharge rates (2 to
60 L min
Seeing the need for a low-cost, reliable and accurate method for measuring
runoff in the field, we developed a new instrument called the “Upwelling
Bernoulli Tube”, or “UBeTube” for short. Similar in function to a v-notch
weir, the instrument is self-emptying, features no moving parts, and can be
configured to minimize sensitivity to sedimentation. Our tested design
possessed the ability to accurately measure flows as low as
0.05 L min
The UBeTube design employed here consisted of a vertical 10 cm (4 inch) diameter pipe with a slot machined into one side (Fig. 1). Schedule 40 aluminum pipe (alloy 6063-T52, though others could be used with equal success) was employed, due to its relatively low cost, strength, rigidity, resistance to corrosion, and machinability. Schedule 40 or higher PVC may also be used, although in our experience the lack of rigidity can make it difficult to accurately machine the slot, and thermal stability is of concern with plastics. The UBeTube pipe can then be attached to a runoff collection system through use of water-tight neoprene rubber gaskets or similar connection method.
We attached the runoff collection system to the bottom of the UBeTube
instrument for several reasons:
the pressure head needed to drive flow into the pipe is reduced compared to having water enter through the top; splashing due to incoming water, which causes pressure fluctuations, is minimized; the runoff system piping can be buried below grade, which protects it, buffers
temperature swings, and secures the system. Example installations are shown in Fig. 1a, b and c.
It should be noted that having the inflow arrive through the bottom of the
pipe could create complicated backwater conditions within the runoff
delivery pipe, which can alter the shape and timing of the runoff
hydrograph. Thus, in certain situations, it may be preferable to have the
inflow enter the UBeTube from the top.
The UBeTube instrument's machined slot can be any shape and dimension, providing the ability to accurately measure a wide range of discharge rates. Our example system used a slot formed by two superimposed trapezoids: the lower trapezoid had dimensions of 0.2 cm bottom width, 1 cm top width and 10 cm height, while the upper trapezoid had dimensions of 6 cm top width and 6 cm height (Fig. 1c). This allowed the system to be operated with less than 30 cm of pressure head.
By measuring the water height within the pipe, the volumetric flow rate of
water through the trapezoidal slot can be calculated using Bernoulli's
equation. Assuming steady-state conditions, the volumetric flow rate (
Water height can be measured through a number of methods; we used a vented pressure transducer system (Decagon Devices CTD) for its combination of low noise, reliability and economy. For our installations, we placed the water level sensor within a pipe located concentrically inside of the main tube (Fig. 1e). This second pipe had a diameter of 4.2 cm (1/4 in. Schedule 40 PVC), and was perforated with 0.6 cm diameter holes beginning 1 cm below the bottom of the height of the slot. This allowed the inner pipe to act as a stilling well, with the goal of helping to reduce momentum effects on the water level at high flows and to prevent non-suspended sediment from interfering with the sensor.
The rating curve (flow rate,
Derivative of the rating curve (d
The effect of slot width on instrument sensitivity can also be seen by
plotting the derivative of the rating curve (d
To calculate inflow into the UBeTube instrument (rather than outflow), it is
necessary to quantify the water storage within the instrument itself. We
recommend the storage equation provided by Stomph et al. (2002b):
Equation (4) allows for the near-instantaneous calculation of runoff and can allow for the study of runoff timing and shape of the hydrograph. It also is useful for times when the water level is not precisely at the bottom of the notch, due to evaporation or capillary effects (as discussed in Sect. 4).
The instrument was validated using a simple test, where various steady-state
flows were added to the system. Five different flow rates were measured
across a range of
Results of the laboratory validation experiment. The uncorrected measurements are represented by the gray-filled circles, while the measurements corrected using Eq. (5) are represented by the open diamonds. Each point represents the mean flow rate measured over a 5 min period.
Based on the mean value for each 5 min repetition, the measurement error
ranged from 1 to 25 % (Fig. 4). Error increased as a function of flow
rate as momentum effects began to dominate and the instrument response
became more sensitive to water height (as demonstrated by the d
A calibration factor can be included in the calculation of flow rate to
account for roughness in the slot surface and deviation from steady-state
flow conditions. While a number of different correction factor techniques
may be suitable, we found that for this particular design a simple
first-order correction factor of
Examples of field applications of the UBeTube instrument
in a long-term study measuring highway stormwater runoff produced by highway
surfaces within western Oregon.
As part of a long-term study focused on quantifying the efficacy of roadside
vegetated filter strips at infiltrating stormwater generated by the
impervious areas, six UBeTube instruments were installed at runoff plots
around the western part of the state of Oregon. Two of the runoff plots were
constructed to be 3
As can be seen, not all precipitation events caused a corresponding runoff
response. For instance, at the Alsea site (Fig. 5b) the first rainfall
event on 5 March 2014 did not produce any measurable runoff, likely due to
dry antecedent conditions. However, subsequent rainfall events of
approximately the same magnitude produced runoff rates that approached or
exceeded the rainfall rate (the latter occurrence due to run-on being
delivered from the adjacent highway surface). Moreover, comparing the runoff
rates from two examples demonstrates the dynamic range of the UBeTube
system, as it proved itself capable of adequately measuring low flows at the
Otis site (
Although the UBeTube instrument has proven its capability in measuring flow
for plot- and field-scale experiments, we here list several considerations
that practitioners should keep in mind when installing and/or using the
instrument, to ensure the quality of collected data:
When using the instrument in cold weather, extra attention is needed when the
ambient temperature drops close to or below freezing, as ice can form inside of the
tube. At the same time, some water level sensors (including the Decagon Devices CTD
model used in our example system) can become damaged if the water around/within them
freezes. While installing the sensor below grade (as shown in Fig. 1e) should provide
some protection from freezing, we nonetheless recommend field inspection of the
installation before and/or after snow events to ensure the quality of data and to
verify proper operation of the water level sensor (Fig. 6a). Under ideal conditions, the water level inside the tube would be maintained
at the bottom of the thin slot so that whenever there is an inflow event, the water
will flow out of the tube instantaneously. However, this can be difficult to achieve
in field installations due to water films forming in the slot due to capillary rise
(Fig. 6b) or inevitable water loss from evaporation (Fig. 6c). Therefore, the tube
should either be refilled to the outlet level prior to expected flow events, or a
calibration should be developed to account for flows that occur before the water level
reaches the bottom of the slot. Equation (4), which accounts for storage within the
instrument, can also be used in such instances. As water flows into the tube, it will carry sediments and small debris that
can pass through the filter mesh installed at the front edge of the collection
channel. These sediments and debris can be accumulated in the thin slot and block the
flow path, thus, jeopardizing the reliability of data (Fig. 6c). Regular cleaning may be necessary. The presence of suspended sediment within the stilling well could increase
the fluid density and, thus, cause measurement error. In high sediment environments
it may therefore be necessary to account for this effect and/or use alternative methods
for measuring water level, such as capacitance probes. When the UBeTube is installed at places where stormwater runoff does not
quickly drain (e.g., at the bottom of a roadside vegetated swale), it is exposed to
the risk of being flooded (Fig. 6d), which will prevent the collection of reliable
data until the surrounding area drains. Momentum effects can cause pressure fluctuations in both the water level and
in the pressure measurements. These effects become more prominent as the flow rates
increase, as seen in Fig. 4. Therefore, it is important to have a carefully designed
stilling well that can help alleviate some of the higher-frequency fluctuations. The use
of alternative (non-pressure) water level sensors, such as capacitance probes, may also
reduce error from momentum effects. Moreover, as previously mentioned, such sensors might
also be preferable in high sediment systems.
Hydrological studies require accurate measurement of water balance
components such as runoff. We presented a new instrument, called the
UBeTube, which can monitor runoff flows at the plot scale. The design is
small, sturdy (no moving parts), and can measure both low (
Most important, the instrument is low-cost, as a single instrument can be manufactured and installed for less than USD 150 (not including water measurement sensor and data-logging costs). The instrument has thus far been used in two field-based studies, providing multiple years of near-continuous runoff data. We presented sample plot-scale data showing that the instrument is capable of providing reliable, near-continuous measurements of surface runoff. Overall, the combination of reliability, accuracy and affordability makes the UBeTube a practical choice for measuring runoff.
This work was performed under two grants: National Science Foundation award 0943682, and Oregon Department of Transportation award 13-025. The authors would like to acknowledge the contributions of Majdi Abou Najm of the American University of Beirut, Jon Hesseltine and Zane Rogers of Oregon State University, and recognize Frank Selker of Selkermetrics, LLC, for sparking the original idea that lead to the development of this instrument. Finally, we would like to thank Kevin Christopher of Christopher CNC for his assistance and patience in prototyping the instrument. Edited by: A. Benedetto