Measuring a dynamic and flooding river surface by close range digital photogrammetry

Current state of the art computational fluid dynamics (CFD) and in particular, river flow modelling, require accurate estimation of the “free surface” in order to accurately predict the three dimensional flow field along a river. Such models are increasingly important for management of river basins and mitigation of river floods as the incidence of extreme rainfall events increases, causing widespread flooding, disruption and loss of life. To increase the accuracy of such flow models, it has become necessary to calibrate model outputs using field data, demanding improved measurement of the flow field along rivers.A substantial and funded research project being conducted at Loughborough University is developing image based river measurement methods using a combination of Particle Image Velocimetry (PIV) and close range photogrammetry (CRP). This paper will report on the use of digital close range photogrammetry to measure the dynamic topographic water surface exhibited by real and flooding rivers.A pair of Nikon D80 (10 Mega-pixel) digital cameras have been purchased, each equipped with a variables zoom lens (f: 18-70mm). The two cameras have been synchronized using two cables connected via a single relay operated switch, tests demonstrating that the accuracy of synchronization is better than 100th of a second. The cameras are mounted on two standard camera tripods, providing convergent and stereoscopic coverage of the river reach, which is between 10 and 20m distant. In initial tests, conventional photogrammetric control was provided using temporary targeted points, coordinated using a Reflectorless Total Station. Subsequent work is being conducted on a semi-engineered river at Farnborough, in the UK, where fixed targets have been permanently installed and coordinated to be in position necessary for the two year duration of the project.Imagery is being processed using the Leica Photogrammetry System (LPS); commercial software which provides the ability to automatically extract digital elevation models. A key issue is the targeting of the water surface and a variety of seeding particles have been tested. Natural materials are clearly preferred, to avoid polluting the natural riverine environment. Leaves, sawdust, and wood chips have all demonstrated some localised success but suffer because of a basic lack of contrast in the imagery captured. It has always been recognized that white polystyrene chips used for packing and posting delicate objects would resolve such difficulties but are clearly undesirable from an environmental perspective. Fortunately, biodegradable packaging chips are now available and tests demonstrate that LPS can successfully generate DEMs representing the dynamic and flooding water surface. Such material degrades within a few days, less if subjected to the mechanical action associated with a flowing river.This presentation will outline the development and application of the methodology, focusing on the accuracies attainable. Specific tests were developed to assess accuracy using a fixed, horizontal survey staff located just above the water surface. Accuracies of 3mm were achieved using the Nikon cameras located 11m away, more than sufficient for the uncertainties associated with the flow modelling. Key issues to be discussed include: synchronization, seeding and significantly, the use of a convergent camera configuration to increase the accuracy of data generated still further. The approach will be used over the next two winter seasons and combined with PIV to parameterise water flow and develop the 3D computerised flow models further.