Visitors’ positive emotions in National Forest Parks: Nonlinear and threshold effects of landscape attributes across seasons and urban gradients

• Published in: Ecological indicators Vol. 178; p. 113963
• Main Authors: Jia, Xing-xing, Xu, Yu-cheng, Zhao, Bing, Li, Hong-yi, Li, Ling-yu, Zhang, Jin-guang
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.09.2025; Elsevier
• Subjects: Big data; Landscape attribute; National Forest Park; Nonlinear association; Visitors’ sentiment; XGBoost-SHAP; Landscape attribute; Nonlinear association; Big data; National Forest Park; XGBoost-SHAP; Visitors’ sentiment
• ISSN: 1470-160X
• DOI: 10.1016/j.ecolind.2025.113963

[Display omitted] •Quantified visitors’ positive emotions in 511 National Forest Parks using big data.•Exurban forest parks showed higher positive emotion index than urban ones.•Identified site environment and 3D attributes as top drivers of positive emotions.•Found seasonal shifts in forest age and density effects on positive emotions.•Discovered stable and varying thresholds across urban, suburban, and exurban parks. National Forest Parks (NFPs) on the outskirts of cities provide visitors with forest therapy and ecotourism experiences. However, the nonlinear and threshold effects of key landscape attributes on visitors’ positive emotions, as well as their seasonal and urban gradient variations, remain poorly understood. This study analyzed 511 NFPs in China using 378,950 valid Sina Weibo comments to calculate a positive emotion index (PEI). Landscape attributes were assessed from pedestrian and aerial perspectives. Social media images were used to quantify three-dimensional landscape attributes (3DLA), including single-element indicators and perceived sensory dimensions. Two-dimensional landscape attributes (2DLA) were categorized into vegetation, site environment, and humanistic attributes. After comparing the performances of the multiple models, we selected the eXtreme Gradient Boosting-SHapley Additive exPlanations (XGBoost-SHAP) model for this study. The results indicated that (1) 3DLA, such as the Green View Index (GVI) and the Natural sensory dimension, along with site environment attributes within 2DLA (e.g., PM2.5 and cooling effects), significantly influence PEI; (2) most landscape attributes exhibit nonlinear and threshold effects on PEI; (3) interactions between certain attributes, such as cooling effects and GVI, reveal potential synergistic or antagonistic effects; (4) attributes such as forest age, density, and PM2.5 demonstrate seasonal variability in their nonlinear relationships with PEI; and (5) attributes such as forest density and age demonstrate heterogeneous nonlinear effects across urban, suburban, and exurban NFPs, PM2.5 and GVI exhibit relatively stable patterns across the urban gradient. This study provides valuable insights for NFP planning and management agencies aimed at enhancing visitor experiences.