Many mountain regions in Europe have experienced massive migrations towards metropolitan areas, with far-reaching implications for societies and the environment, especially croplands and grasslands. In this work, we tailored a geospatial framework to envisage the probability of land abandonment in the Spanish Pyrenees at moderate spatial resolution. We predicted the likelihood of land abandonment combining machine learning algorithms, geospatial data and historical observations of land abandonment in the period 1980 to 2019. The model attained a high predictive performance (AUC = 0.85) at a moderate resolution (30 m), providing insights into the spatial behavior of the potential for both abandonment and its main driving forces. The highest rates of abandonment were found in rural settlements and towns in bottom valleys where tourism and recreational activities have proliferated over the years. Fast and comfortable connections between the main metropolitan areas (e.g. Barcelona and Zaragoza) and the mountain regions foster touristic activity and lead to the creation of new settlements. Ecotourism and mountain sports promote land abandonment as evidenced by the high probability predicted over the Central Pyrenees (from Pallars Jussà to Alto Gállego counties). Our results provide spatially explicit probability and uncertainty outputs, providing insights into site-specific abandonment potential and the influence of its drivers. These results can substantially assist land planners in decision-making, enabling assessments at local scale.

欧洲的许多山区都经历了向大都市地区的大规模迁移，对社会和环境产生了深远的影响，尤其是农田和草原。在这项工作中，我们定制了一个地理空间框架，以设想在中等空间分辨率下西班牙比利牛斯山脉的土地废弃概率。我们结合机器学习算法、地理空间数据和 1980 年至 2019 年期间土地废弃的历史观察预测土地废弃的可能性。该模型在中等分辨率 (30 m) 下获得了高预测性能 (AUC = 0.85)，提供了见解进入废弃可能性及其主要驱动力的空间行为。在多年来旅游和娱乐活动激增的谷底农村定居点和城镇中，遗弃率最高。主要大都市区（如巴塞罗那和萨拉戈萨）与山区之间快速而舒适的连接促进了旅游活动，并导致了新定居点的建立。生态旅游和山地运动促进土地废弃，这从预测的中比利牛斯山脉的高概率（从Pallars Jussà到Alto Gállego县）。我们的结果提供了空间明确的概率和不确定性输出，提供了对特定地点废弃潜力及其驱动因素影响的洞察。这些结果可以极大地帮助土地规划者进行决策，从而在当地进行评估。