High Resolution, Low-Cost 3D Riverscape Mapping Using Field Photography

We present a new approach to mapping both the above- and below-water three-dimensional riverscape environment using nothing but off-the-shelf (and geometrically unconstrained) cameras and a new approach called “Structure from Motion” (SfM). Multiple images taken from different vantage points can be converted to 3D topographic point clouds using automated computer vision algorithms. This new approach stands in contrast to traditional photogrammetric techniques which require significant image geometric control, expensive and difficult-to-use software, and a high degree of user training. The almost-free cost of the SfM approach also stands in stark contrast to the high expense of aerial lidar and terrestrial laser scanning, the current standards for three-dimensional environmental mapping. The SfM process uses user-friendly and free online software, as well as ground or aerial images taken of a river environment, and converts these to 3D point clouds with resolutions between aerial lidar and terrestrial laser scanning depending on the camera height.

Such an approach immediately opens up a wide range of uses for riverine mapping and monitoring. In addition to the three-dimensional information, this approach allows the direct rectification of imagery taken at low altitude. Such imagery can be simply processed to extract useful habitat metrics such as water depth and quality, bottom albedo and particle size, and potentially water motion information (velocity, Froude number). Non-water information such as woody debris emplacement and riparian vegetation structure is also amenable to the SfM process.  Instrument platforms for using the SfM approach range from simple hand-held cameras to aerial imagery from planes. In this presentation, we demonstrate the utility of the helikite, a hand-held, low-cost aerial (yet tethered) combination of a balloon and kite. Helikites can be adjusted to many different heights with off-the-shelf cameras suspended beneath to image rivers and streams at resolutions from sub-centimeter up to several meters. By walking a river segment with a helikite-camera setup, a series of high-resolution images are collected and for which desktop image processing using the SfM process and spectral techniques produce maps of riverscape habitats with unprecedented scope and resolution. The ease and very low cost of data collection make this process much easier to support for repeat data collection in comparison with other field or aerial mapping methods.