Landscape characterization using photographs from crowdsourced platforms: content analysis of social media photographs

Landscape characterisation using social media photographs from popular platforms has been proposed as a landscape and ecosystem services approach. However, popular crowdsourced websites provide uncharacterized data and are only representative of the general public. Photographs from crowdsourced sports platforms, whose users are more homogeneous, could help to characterise landscape more uniformly. In this study we use automated content analysis from photographs on Wikiloc, a crowdsourced sports platform, to characterize landscape in the Ebro Delta Natural Park, a protected area in Spain. Our approach applies big data procedures and spatial analysis to provide in-depth information regarding what draws visitors' attention to a landscape and to ascertain their intrasite flow. Our results show that sports users are keen on natural landscapes and pay less attention to rural and degraded landscapes, and that areas closer to paths are more photographed than more distant areas.