Annual global grided livestock mapping from 1961 to 2021
Zhenrong Du, Le Yu, Yue Zhao, Xinyue Li, Xiaoxuan Liu, Xiyu Li, Pengyu Hao, Zhongxin Chen, Zhe Guo, Liangzhi You, Xiaorui Ma, Hongyu Wang
Abstract
Abstract. Understanding global livestock dynamics is essential for global food security, public health, socio-economic and sustainable development. This study developed an automated global livestock mapping framework that integrated Food and Agriculture Organization Corporate Statistical Database (FAOSTAT) and the Random Forest regression model. By implementing the mapping scheme on Google Earth Engine (GEE), we develop the first annual gridded livestock of the world (AGLW), covering the period from 1961 to 2021 at a spatial resolution of 5 km. The annual maps of AGLW were then evaluated from three perspectives: model level, finer-scale statistic level, and pixel level, with correlation coefficients (r) of 0.65–0.86, 0.78–0.97, and 0.78–0.88, respectively. The AGLW maps reveal the spatio-temporal dynamics of global livestocks over the past six decades, highlighting both global expansion and localized fluctuations, such as the notable increase in pig stock in China and the decline in horse stock in Poland. By offering a reliable and continuous dataset, AGLW overcomes the limitations of existing livestock mapping products in terms of spatio-temporal continuity and resolution. This dataset serves as a crucial resource for enhancing our understanding of global livestock dynamics, informing policy decisions, guiding sustainable agricultural practices, and promoting resilience in both ecological and human systems. We also release per-pixel, per-year uncertainty layers, and we recommend consulting these layers, especially for early decades and data-sparse regions. The full archive of AGLW is available at https://doi.org/10.5281/zenodo.17128483 (Du et al., 2025).