Data quality assurance in pressure transducer-based automatic water level monitoring

Authors: Endale, Dinku; Fisher, Dwight; Jenkins, Michael; Schomberg, Harry

Submitted to: Proceedings of the American Society of Agricultural and Biological Engineers International (ASABE)
Publication Type: Proceedings
Publication Acceptance Date: 12/17/2009
Publication Date: 2/21/2010
Citation: Endale, D.M., Fisher, D.S., Jenkins, M., Schomberg, H.H. 2010. Data quality assurance in pressure transducer-based automatic water level monitoring. Proceedings of the American Society of Agricultural and Biological Engineers International (ASABE).

Interpretive Summary: The computer and associated technology era has allowed a proliferation of development of environmental monitoring technology. Advances in submersible pressure transducer design and manufacturing has made these devices easily accessible and affordable for use in water level, water flow and other hydrologic investigations in streams, wetlands, ponds, groundwater, etc. By connecting these devices to suitable data recording devices (data loggers) it is now possible to automatically and continuously monitor and record hydrologic data at the desired time interval from minutes to hours and days. Unfortunately unless such recording systems are closely watched to check for the proper functioning of all part of the recording system, erroneous data could be collected for long periods. Researchers at the USDA-ARS, J. Phil Campbell Sr. Natural Resource Conservation Center in Watkinsville, GA, carried out over a two year period continuous monitoring of flow from a spring with a flow measuring structure fitted with a submersible transducer connected to a data logger. Over the same period, periodic manual checking of flow rate was performed. The researchers identified periods where the automated system under predicted, correctly measured, and over predicted the flow rate. Initially, over a 162-day period in 2003, the automated system under predicted the flow rate by approximately 17% (mean 6.07 versus 7.28 gallons per minute). Over a 349-day period in 2004 flow rate were 4.66 gpm for the manual and 4.74 gpm for the data logger measurements (a 1.7% difference only). Over 235 days during the latter part of the monitoring period, the automated system over predicted the flow rate by approximately 29% (mean 9.49 versus 7.34 gpm). In a similar setup a pond outflow was underestimated by 27.4% (3,471,485 gallons) over a 30-day period in 2009. Flow rate is an important hydrologic variable in water quantity and quality investigations including estimation of daily pollutant loads. Proper data quality assurance is needed in flow rate estimation utilizing pressure transducers to avoid the propagation of errors from flow estimation to that of pollutant load. The same principle should apply in other sensors used in flow measurement. These observations should be on interest to researchers, teachers, environmental groups, regulators, engineers and local to regional water resources managers.

Technical Abstract: Submersible pressure transducers integrated with data loggers have become relatively common water-level measuring devices used in flow or well water elevation measurements. However, drift, linearity, hysteresis and other problems can lead to erroneous data. Researchers at the USDA-ARS in Watkinsville, GA, carried out over a two year period continuous monitoring of flow from a spring using a flume fitted with a pressure transducer and data logger. Over the same period, periodic manual checking of flow rate was performed. Initially, over a 162-day period in 2003 with 77 comparisons, the automated system under predicted the flow rate by approximately 17% (mean 0.383 versus 0.459 L s-1; significant based on non-parametric test). Over a 349-day period in 2004 in 66 days of comparisons flow rate was 0.294 L s-1 for the manual and 0.299 L s-1 for the data logger measurements (a 1.7% difference; non-significant). Over 235 days during the latter part of the monitoring period, the automated system over predicted the flow rate by approximately 29% (mean 0.599 versus 0.463 L s-1; significant). In a similar setup at a pond outflow, an offset of 32 mm in flow depth caused an outflow error of 27.4% (13,141 m3) aver a 30-day period. Flow rate is an important hydrologic variable in water quantity and quality investigations. Proper data quality assurance is needed in flow rate estimation utilizing pressure transducers to avoid the propagation of errors from flow estimation to that of pollutant load. The same principle should apply in other similar sensors.