Flow information is necessary in many diverse applications including water supply management, pollution control, irrigation, flood control, energy generation, and industrial use. New technologies have been developed for the establishment of stage-discharge relationships due to concerns about costs, accuracy, and safety of traditional discharge estimation methods. One emerging technology for measuring open-channel flow is Large-Scale Particle Image Velocimetry (LSPIV).

LSPIV is a system capable of measuring surface velocity by collecting and analyzing recorded images of the stream surface. LSPIV has several advantages over conventional discharge measurement techniques: LSPIV is safer, could be automated to reduce labor, and could produce "real-time" discharge measurements. Therefore, the overall goal of this study was to evaluate the accuracy and feasibility of using LSPIV to measure discharge in low-order streams. The specific goals were to develop and test a prototype under varying conditions in a laboratory flume, adapt the prototype for field conditions, test the accuracy of the prototype in the field, and assess and recommend improvements for LSPIV operation as a stream discharge measuring device.

The laboratory experiments results indicated that LSPIV accuracy was influenced by camera angle, surface disturbances and flow regime (Froude number), and particle seeding density. Camera angle was optimum around 15 degrees, with larger camera angles producing more error due to image distortion. Conditions at high Froude numbers likely produced out-of-plane displacement losses due to surface disturbances. Low Froude numbers also showed under-predictions, which were likely due to agglomeration of the tracer particles at low velocities. Finally, the laboratory results demonstrated that tracer seeding density should be maximized and that densities below three particles per interrogation window should significantly reduce LSPIV accuracy.

The LSPIV prototype was tested at two low-order streams after developing a field prototype and operating procedures. Under field conditions, the prototype acquired consistent images, performed the necessary image processing, and established rules for estimating input parameters. The accuracy of LSPIV was evaluated using a Flo-Mate 2000 current meter and a permanent weir. Overall, twenty discharge measurements were taken with each measuring device at Stroubles Creek and Crab Creek. The discharges measured ranged from 0.12 to 63 cfs, which corresponded to a large range of velocities, with both simple and complex flow patterns. Problems were encountered from surface glare reducing image quality at both sites.

The LSPIV prototype was accurate for most measuring conditions with a mean error of -1.7%, compared to the weir measurements. The LSPIV measurements tended to under-predict discharge at high stages and had greater error at moderate flows (up to 39%) compared to the weir. However, at low flow conditions LSPIV showed improved discharge accuracy over the current meter, in comparison to the weir measurements. The LSPIV discharge measurements were not statistically different from either the current meter or weir (á = 0.05). Finally, the LSPIV discharge measurements had an uncertainty of approximately ±14% (at a 95% confidence interval).

In conclusion, LSPIV accuracy can be degraded by surface disturbances, inadequate illumination, and poor seeding densities. However, LSPIV showed adequate accuracy with the potential to become competitive with conventional discharge measurement techniques and therefore, has the potential to reduce costs and increase the geographic extent of surface water monitoring networks.