D.3 Post-Measurement Activities and Discharge Measurement Computation
The purpose of this section is to identify some of the common factors that lead to inaccuracies when observing widths, depths, and velocities. Inaccuracies can occur during the measurement of any of these parameters through errors introduced by technique or the type of equipment used. Errors can be categorized as human, systematic, or random.
Measuring overall stream width and defining individual verticals, using a measuring tape or tagged nylon chain, can be done with precision and neither contributes to overall error in a discharge measurement. Human error can be a factor when measuring permanently marked cross sections on bridge rails or other structures if unconventional spacing has been employed. Wind can oscillate the tape and lead to movement of the tape anchorage points and/or mistakes in observing values.
Record keeping errors can occur during boat measurements when the boat is positioned by electronic or survey equipment operated on shore with the rest of the data recorded in the boat. To avoid errors, good communication and post-measurement record comparison are essential.
Depth observations made by rod are subject to random errors such as the sinking of the rod into a soft streambed, failure to identify obstructions in the cross section between soundings, and incorrect reading of the graduated rod.
The opportunity for error is far greater when cable suspension systems are employed. In addition to the errors listed for rod measurements, certain techniques used in the measurement process can result in errors in the recording of depth. The most obvious and most common error is the failure to add the distance from the bottom of the weight to the center of the meter when the latter point of reference has been used to "zero" the reel or handline at water surface. Other errors peculiar to the handline suspension occur as a result of mistakes in the method of subdividing the distance between cable markers (streamers) and applying to obtain exact depths.
Other sources of significant error in a discharge measurement are those that relate to the measurement of velocity. Among the more readily apparent are those associated with the calibration of current meters, the direction of flow, the duration of the observation time, and the number of observation verticals as well as the number of observation points in each vertical.
The Province of British Columbia has maintained a standard, similar to the Water Survey of Canada, that all current meters have individual calibrations. These are obtained by towing the meters through a tank of still water at velocities of between 4.5 to 300 cm/s, and from this individual calibration curves are developed. Although it is generally accepted that this procedure is equivalent to the stream gauging situation where the meter is held in flowing water, there is some question about the validity of this assumption when using the meter (particularly cup-type meters) under turbulent flow conditions.
Discharge measurement cross sections are usually chosen so that the flow is perpendicular to the cross section. Even though they are carefully selected, it is not always possible to avoid oblique flows at some of the verticals. At these verticals the velocities must be corrected by applying an appropriate cosine coefficient. Random errors may be introduced when observing the angle of flow if it is assumed that the angle observed at or near the surface remains the same throughout the entire depth.
Other sources of error can be introduced when using rod suspensions and in particular when used in deep fast-flowing water. Although the current meter can be misaligned both vertically and horizontally with the direction of flows, the most significant error will result from vertical misalignment. The current meter will under register if tilted above or below the horizontal, and the magnitude of the error will depend upon both the velocity of the water and the angle of departure.
Pulsations in velocity are evident in all streams even though flow conditions are essentially steady. Because pulsations are random in nature, the effects of pulsation will be eliminated when velocities are observed for a sufficient length of time. In actual practice during a discharge measurement, velocities are observed for relatively short periods of time. The expectation is that a sufficient number of observations will be made so that pulsation effects will tend to cancel each other during the course of a measurement. Studies have shown that at low velocities, pulsation effects are usually greatest. Studies have also shown that the optimum observation duration is between 40 and 60 seconds and that accuracy decreases significantly if a duration of less than 30 seconds is used; for durations longer than 60 seconds, the increase in accuracy is generally negligible.
There are two ways in which the accuracy of a discharge measurement can be significantly affected by the number and distribution of observation verticals. First, the observation verticals are used to define the channel cross-sectional area. Appreciable errors will be introduced if the number of observations made to define the cross section are not sufficient. This particular problem can be overcome by obtaining additional depth observations.
Secondly, the velocity observations in the verticals are used to define the mean velocity in the cross section; therefore, the verticals should be spaced so that the velocities observed are more representative of those in the preceding half panel and the following half panel.
The spacing of observation verticals can be accomplished on the basis of either the equal flow method, the equal width method, or a combination. With the equal flow method, the width segment can change frequently, and using the mid-section method for computations the horizontal velocity profile tends to be distorted. That is, if the width segments change frequently, the observed velocities will not occur at the midpoint of the panels.
A good compromise is to use the equal width method and to change the spacing of verticals only a few times during a measurement to accommodate any significant changes in flow distribution.
Studies on measurement accuracy have shown that accuracy tends to be low when fewer than 16 verticals are used but the improvement becomes negligible when more than 35 verticals are used. All else being equal, the use of 20 - 25 verticals is considered optimum.
The mean velocity in a vertical is normally obtained by measuring at one or two points in that vertical. Comparing these observations with those obtained by some detailed method (a mean of observations at every tenth of the depth, plus half the value observed at the surface and half the value at the bottom), indicated that random errors do occur when determining the mean velocity in any given vertical. Furthermore, the one-point method is usually not as accurate as the two-point method. Nevertheless, surface and bottom effects become significant as the stream depths decrease, and when depths are less than 0.8 m, the one-point method should be used.
Errors in the measurement of width, depth, and velocity as well as the lack of care in choosing the number of verticals and observations in a vertical, all combine to reduce the overall accuracy of a discharge measurement. To a large extent, human errors can be avoided by careful attention to detail and by adhering to established and proven techniques and routines. Systematic errors can be reduced significantly by proper maintenance and calibration of instruments and equipment, and by adequate training. However, random errors will always occur. A significant reduction in these errors can be achieved if the field technician obtaining the measurement can recognize the potential problem areas and can take the appropriate precautionary measures to avoid or minimize them. One possible indication of measurement accuracy can be obtained by conducting several consecutive or simultaneous measurements, and by using different sets of equipment and different techniques.