Landscape aesthetics: Spatial modelling and mapping using social media images and machine learning. (October 2020)

• Record Type: Journal Article
• Authors: Gosal, A.S.; Ziv, G.
• Is Part Of: Ecological indicators. Volume 117(2020)
• Publication Date: 2020-10
• Subjects: Bradley-Terry model -- Flickr photos -- LUCAS photos -- Google Vision -- Probability models -- Image content analysis; Environmental monitoring -- Periodicals; Environmental management -- Periodicals; Environmental impact analysis -- Periodicals; Environmental risk assessment -- Periodicals; Sustainable development -- Periodicals; 333.71405
• DOI: 10.1016/j.ecolind.2020.106638
• Languages: English
• ISSNs: 1470-160X

Abstract

Highlights: Social media images have more aesthetic value than random ones from the same area. Probability models can be used to rank-order images for aesthetic value. Incomplete paired comparison datasets can be used to rank order images for aesthetic value. Model results were upscaled for 22, 615 images to map aesthetic value for the National Park. Abstract: Cultural ecosystem services such as aesthetic value are highly context-specific and often present difficulties in their assessment. Here we present a case study in the northern English Protected Area of the Yorkshire Dales National Park. Utilising publicly available images, paired-comparison surveys, probability modelling, machine-learning based text annotations, natural language processing and regression analysis, we developed a spatial model to predict and map landscape aesthetics across the whole site. The predictive model found eighteen significant variables, including the positive role of rural areas, mountainous landforms and vegetation for aesthetic value. Finally, we demonstrate the potential of our approach to varying size datasets and partial paired-comparison matrices, finding a very good agreement with only 20% of paired comparisons. This study demonstrates the use of freely available data and mostly open source tools to ascertain landscape aesthetic value in a large Protected Area.