Using aerial photos to produce digital surface models, otrhophotos, and land cover maps of a coastal area in Puget Sound, WA

Abstract

A new method in photogrammetry (the practice of taking spatial measurements from photographs) called Structure from Motion (SfM) produces automated measurements at much lower cost than older techniques. Research has shown that the spatial resolution and accuracy of measurements taken from SfM varies with photo properties such as camera position (terrestrial, aerial), spectral resolution (black and white, color, near infrared), and subject matter (bare earth, buildings, trees). Our first goal was to assess the spatial resolution and accuracy of SfM measurements taken from color infrared aerial photos of a 400 hectare coastal peninsula. We did so by producing and validating several types of Digital Surface Model (DSM, maps displaying elevation data) and orthophotos (photo maps that have been geometrically corrected using elevation data such that the scale is uniform). Results show that SfM measurements we derived from aerial photos are of high spatial resolution (5 points/m2) and accuracy. Orthophotos and DSMs derived from similar SfM measurements have been shown to be suitable for many common geospatial applications, but SfM measurements have not yet been used to produce comprehensive land cover maps. Our second goal was to assess the accuracy of a land cover map derived from fusing a DSM and orthophoto and analyzing similarity between adjacent areas using a technique called Object Based Image Analysis. The land cover map we designed is intended to be suitable for monitoring and regulating land cover at small spatial scales in order to evaluate anthropogenic alterations toward conservation of shoreline ecological function. In this case we have semi-automatically mapped trees, shrubs, groundcover vegetation, bare ground, impervious surfaces, and water with 86% accuracy, and much higher resolution than the best available land cover map of the same area. We believe that the low cost of SfM measurements, and the high accuracy and resolution of products derived from them make SfM well suited to aid in monitoring and regulating land use to conserve shoreline ecological function.