Study on Grassland Dynamics and Livestock Behavior Changes Under Different Grazing Strategies

Dataset Description This dataset consists of GPS tracking data and remote sensing imagery used to assess the impact of grazing strategies on grassland vegetation and cattle behavior. The data was collected during a fenced grazing experiment conducted in Jinhua Village, Jinshan Town, Wenchang City, Hainan Province, China. GPS Data The GPS data includes the location trajectories of cattle tracked using GPS collars, recorded at regular intervals. These data provide precise spatial information on the cattle's movement within the grazing plots and are essential for understanding their foraging behavior, walking, standing, ruminating, and resting patterns. The dataset includes timestamps for each location point, along with GPS coordinates (latitude and longitude). Additionally,cattle behavior was observed and recorded in the field, with GPS-collected activity data integrated to develop a behavioral classification model. Remote Sensing Data The remote sensing data includes high-resolution imagery captured using unmanned aerial systems (UAS) to assess vegetation changes over time in the grazing plots. The images were processed to calculate the Normalized Difference Vegetation Index (NDVI), a key indicator of vegetation health. NDVI data were obtained before and after grazing activities to monitor vegetation recovery and degradation under different grazing intensities and regimes. Data Analysis The dataset was used to classify cattle behaviors using machine learning models, such as XGBoost, Random Forest, Decision Tree, Extra Trees, and CatBoost. The models were trained using GPS and behavioral data, providing insights into the effects of grazing strategies on cattle behavior. NDVI data were analyzed to determine the relationship between cattle behavior and vegetation changes. Purpose This dataset is intended to support the study of sustainable grassland management and grazing strategies. It provides valuable insights into the interaction between cattle behavior and grassland vegetation under different grazing pressures. The data can be used to refine grazing management practices and inform the development of more sustainable land management strategies.数据集说明 本数据集包含GPS追踪数据与遥感影像数据，用于评估放牧策略对草地植被及牛只行为的影响。该数据采集自中国海南省文昌市锦山镇金华村开展的围栏放牧实验。 GPS数据 GPS数据包含通过GPS项圈（GPS collar）追踪得到的牛只位置轨迹，以固定时间间隔进行记录。此类数据可提供牛只在放牧样地内移动的精确空间信息，对于解析其觅食、行走、站立、反刍及休憩行为模式至关重要。本数据集包含各位置点的时间戳，以及GPS坐标（纬度与经度）。此外，研究人员在野外开展了牛只行为观测记录，并结合GPS采集的活动数据构建行为分类模型。 遥感数据 遥感数据包含通过无人机系统（Unmanned Aerial Systems, UAS）采集的高分辨率影像，用于评估放牧样地内植被随时间的动态变化。研究人员对影像进行处理以计算归一化差分植被指数（Normalized Difference Vegetation Index, NDVI），该指数是衡量植被健康状态的核心表征指标。本数据集分别在放牧活动前后获取NDVI数据，以监测不同放牧强度与放牧模式下的植被恢复与退化进程。 数据分析 本数据集被用于基于机器学习模型开展牛只行为分类任务，所涉模型包括极端梯度提升（XGBoost）、随机森林（Random Forest）、决策树（Decision Tree）、极端随机树（Extra Trees）以及CatBoost。模型通过GPS与行为数据进行训练，以揭示放牧策略对牛只行为的影响机制。同时，研究人员对NDVI数据展开分析，以探明牛只行为与植被变化之间的关联关系。 数据集用途 本数据集旨在支撑可持续草地管理与放牧策略相关研究。其可为解析不同放牧压力下牛只行为与草地植被间的相互作用提供宝贵见解，亦可用于优化放牧管理实践，并为制定更具可持续性的土地管理策略提供参考依据。