Social media and crowdsourcing

Using the Knowledge Hub for Landscape Practices for analyzing social media photographs

The rapidly growing number of landscape photographs published in social networks such as Flickr, Panoramio, Instagram, Twitter or Facebook has a large potential for the elicitation of the values of landscape features and the ecosystem services they provide.

A recent study within the HERCULES project, explored the challenge of this methodological approach, suitable for eliciting the importance of cultural ecosystem services and the landscape features underpinning their provision across different sites in Europe. This tutorial introduces to this use case and the method that was developed in this study and teaches how to perform such a study using the Knowledge Hub for Landscape Practices.

Read the paper: Elisa Oteros-Rozasa, Berta Martín-López, Nora Fagerholm, Claudia Bieling, Tobias Plieninger (2017): Using social media photos to explore the relation between cultural ecosystem services and landscape features across five European sites. Ecological Indicators. DOI: 10.1016/j.ecolind.2017.02.009

Set your goal

Choose a subject that relates to the potential that the analysis of landscape photographs provides.

The use case example aimed to develop a methodological approach for eliciting landscape features and cultural ecosystem services (CES), as well as their relationships (challenge 1). To do so, we used crowdsourced geotagged photographs posted on Flickr and Panoramio (challenge 2), and applied this method across five contrasting study landscapes in Europe (challenge 3). Our specific goals were: (1) to identify the most frequent CES that are represented by landscape photographs and the associated landscape features in five different European landscapes; (2) to specifically explore how landscape heterogeneity is associated with CES diversity; and (3) to identify bundles of CES mediated by landscape features that emerge from diverse landscape perceptions. Finally, we also aim to assess the methodological approach by (4) analysing differences between two social media (Flickr and Panoramio) and by (5) comparing the land cover shown in the analysed photographs with the CORINE land cover across the study landscapes.

Choose a study landscape

Identify a study landscape to locate your research. You can choose a larger landscape scale or use municipality borders to delimit the study area and the related images.

Our user case example was conducted in five sites across Europe:

  • Peipsiääre and Alatskivi (Estonia),
  • Gera (Lesvos, Greece),
  • Colmenar Viejo (Spain),
  • Obersimmental (Switzerland), and
  • Börje (Uppland, Sweden)
These study landscapes do not cover the full diversity of European cultural landscapes, but they span over major environmental and land use history gradients throughout Europe, representing diverse geographical and ecological contexts, landscape dynamics and land uses.

Gather data

By now, the Knowledge Hub for Landscape Practices provides data of the social media platform Panoramia because the platform has been sufficiently used for sharing geotagged photograph of landscapes. But of course data of other platforms like Instagram, Facebook, Twitter and Flickr can be used as well if the use applies to copyrights and data protection.

The use case example downloaded the data from the platforms Panoramio and Flickr. A total of 6298 photographs geotagged in the study landscapes were found in Flickr, and 3595 in Panoramio. After the first selection of photographs according to the defined criteria, we had 782 photographs from Flickr and 1807 from Panoramio. The final sample analysed was composed of 465 photographs from Flickr and 939 photographs from Panoramio.

Analyse the content

Review the pictures. It is useful to develop a review protocol including a set of indicators to be identified in each photograph

The use case example identified specific landscape features and cultural ecosystem services for the assessment.

Landscape features
Trees Water body Mountain
Shrubs River Infrastructures
Agriculture Lake/pond Urban
Grasslands Sea Snow/Ice
Lawn Coastal/beach/dune Livestock
Wood pasture Rocks Wild

Cultural ecosystem services
Cultural Heritage: Recreation: Social: Spiritual:
Presence/absence of sites relevant to local history and culture: memorials, commemorations, historical sites in the photograph Presence/absence of sites used for recreational activities and that are indicated (e.g. hiking trails with signs, recreational facilities, benches, subsistence gardens, hunting facilities) Presence/absence of sites serving as meeting points with friends Presence/absence of sites that foster a sense of authentic human attachment: folklore, traditions, art

In order to check the adequacy of the method, and in order to account for the importance of local perspective and knowledge of the study area, you might get the protocol tested by some experts with local knowledge. It is also useful to assess the reliability of the methodological approach by analyzing the differences between the representation of land cover in photographs and the actual landscape by comparing it for example with CORINE based categories of land cover and uses. It is probably that people tend to overportray certain landscape features like water bodies or rocks while other features like arable land or urban features might ted to appear less frequent in relation to their land-cover.

In our use case example, the results show that the method applied provided an interesting portray of the five study landscapes, their similarities and their differences in relation to the features people depict from them. The identification of bundles of features of European landscapes and cultural ecosystem services clarified the relation of recreation with mountain areas (terrestrial recreation) and with water bodies (aquatic recreation). Cultural heritage, social and spiritual values were particularly attached to landscapes with wood-pastures and grasslands, as well as urban features and infrastructures, i.e. to more anthropogenic landscapes. Particularly wood-pastures and shrubs showed were more frequently portrayed in all study sites in comparison with their actual land cover.

Share your results

Once you have the results, you can upload them to Knowledge Hub to either communicate them or to further visualize them. You can also use it to actively ask for feedback on the adequacy of the results. For this purpose, you could create a Lab, or make use of already existing labs, like Share your landscape practice.

Finally, our use case example also reflects on pending critical questions to be answered in future research: what landscapes, landscape features and ecosystem services still remain socially invisible in social media, to whom and why? Therefore, conclusions on the importance and spatial distribution of cultural ecosystem services based on social media need to be considered carefully due to several methodological constraints that our study has pointed out, namely uncertainty about the social and ecological representativeness of publicly shared photographs. This call for caution applies widely under the increasingly attractiveness of crowdsourced social media for geographical research, which on one hand provide free and easily retrievable big data, but on the other hand are created and used by particular groups of social actors in specific geographic areas.