User-generated data to predict visitors in environmental areas

Abstract

The economic valuation of recreational ecosystem services is challenging due to difficulties in obtaining geo-tagged information of users. The objective of this study is to validate crowdsourced and user-generated content in order to predict visitation patterns to 16 national parks in Spain. The results may serve to encourage its utilization in the study of recreational demand in other countries, particularly developing countries, where on-site visitor information may be limited or expensive to gather. The present article employs a negative binomial regression model to evaluate the validity of two sources of data: Flickr and mobile phones. The accuracy of predictions exhibited variation across the 16 parks, indicating that site-specific characteristics, such as the seasonality of visitation patterns, may be of significance. The utilization of mobile phone data for modelling visitors yielded enhanced predictive capacity, as shown by the goodness of fit of the estimated models.