Quantifying the amount of water flowing in storm sewers is vital for calibrating hydrologic models and estimating urban pollutant loads. Measurements of such flows are often made by placing flow measurement devices that work on acoustic principles within sewer systems (see figure at left). However, the uncertainty associated with these measurements and sensors as a function of various flow conditions has never been quantified.
Recently, PhD students Marcus Aguilar and Walter McDonald, and associate professor Randel Dymond, conducted a study to benchmark this uncertainty for the first time through a series of laboratory experiments. The findings show that the uncertainty associated with the measurements can vary substantially with flow condition and sensor type. The details of the study, along with a discussion on the implications of uncertainty for stormwater science and management, can be found in the Journal of Hydrology. The abstract from this paper is given below.
Summary: The uncertainty associated with discharge measurement in storm sewer systems is of fundamental importance for hydrologic/hydraulic model calibration and pollutant load estimation, although it is difficult to determine as field benchmarks are generally impractical. This study benchmarked discharge uncertainty in several commonly used sensors by laboratory flume testing with and without a woody debris model. The sensors were then installed in a field location where laboratory benchmarked uncertainty was applied to field measurements. Combined depth and velocity uncertainty from the laboratory ranged from ±0.207–0.710 in., and ±0.176–0.631 fps respectively, and when propagated and applied to discharge estimation in the field, resulted in field discharge uncertainties of between 13% and 256% of the observation. Average daily volume calculation based on these observations had uncertainties of between 58% and 99% of the estimated value, and the uncertainty bounds of storm flow volume and peak flow for nine storm events constituted between 31–84%, and 13–48% of the estimated value respectively. Subsequently, the implications of these observational uncertainties for stormwater best-management practice evaluation, hydrologic modeling, and Total Maximum Daily Load development are considered.