Potential and constraints of Unmanned Aerial Vehicle technology for the characterization of Mediterranean riparian forest

Abstract

Unmanned Aerial Vehicle (UAV) technology provides potential for very high spatial resolution (<25 cm) mapping of relatively large areas at a user-defined re-survey frequency. In a riparian context, UAV technology provides a mechanism for riparian managers to (a) quantify riparian terrain and vegetation units and (b) identify standing dead wood and canopy mortality. In this study a paraglider UAV was used to survey 174 ha at 6.8–21.8 cm ground resolution. Pixel-based and object-oriented classification approaches were used at the scale of a single image and a channel mosaic. Significant potential was demonstrated: vegetation units were classified with an accuracy of kappa = 0.79 and standing dead wood units were identified with an average accuracy with respect to omission and commission errors of 80% and 65%, respectively. Work across multiple images identified that major constraints currently result from factors such as illumination conditions and sensor movement during flight, which create variations in spatial resolution and radiometry. It is expected that with further methodological refinement and more complex methods of automated radiometric correction UAV technology can provide the flexibility to rapidly produce very high resolution map products to aid riparian management.