传统地理学/GIS对人群时空行为的研究集中在中宏观尺度，对微观时空行为的研究相对不足。针对这一问题，提出了微观人群数字孪生的概念及基础逻辑框架，从人群行为实验方法、基础行为/场景行为建模、典型应用及案例分析等方面总结了微观人群数字孪生的技术基础及其研究现状，并从地理学的视角，对基于系统论的人群实验思路和可信人群实验方法、以及利己-利他、个体-群体、确定-随机统一的人群行为建模方法、微观行为地理学和微观人地关系理论体系构建等未来研究进行探讨，希望推动地理学、GIS和虚拟地理环境对微观人群时空行为理论、方法和技术的研究。

Traditional geography/GIS researches on the spatiotemporal behavior of people focus on the meso and macro scale, while the research on the micro spatiotemporal behavior is relatively insufficient. Aiming at this problem, first of all, a new concept of micro-crowd digital twins is proposed. The idea is to map the real crowd and their behaviors into the virtual space, calculate and simulate the crowd behavior, mine the evolution mechanism, discover the operation rules, predict the trends, locate the abnormalities in advance, propose optimization strategies, and improve the efficiency and safety of real crowd and real society. The micro-crowd digital twins can be divided into two stages, the special digital twins and the general digital twins, according to whether it is synchronized in time between the real crowd and the virtual one. Then, the technical basis and research status of micro-crowd digital twins are summarized from the aspects of crowd behavior experimental approaches, basic behavior/scenario behavior modeling, and typical applications. (1) According to the types of places and events, the crowd behavior experiment approaches can be divided into "real place-real event", "real place-simulated event", "simulated place-simulated event" and "virtual place-virtual event"; (2) The status of crowd behavior modeling is summarized which can be divided into basic micro crowd behavior models and scenario behavior models. The basic micro crowd behavior models are to model individual behaviors in a simple and abstracted space, with general interactions between individuals and obstacles. The scenario behavior models are to model crowd behaviors in complex and detailed space, with comprehensive interactions between crowds, environments, and geo-events. The advantages and disadvantages of typical basic micro crowd behavior models such as Social Force Model, Velocity Obstacle model (ORCA model) and Artificial Intelligence model are analyzed. (3) A typical case of micro-crowd digital twins in a school evacuation is analyzed. The platform of digital twins of crowds is developed to simulate and analyze the disordered behaviors and orderly behaviors, and the spatial layout of the school and the crowd evacuation strategies are optimized. By this way, the special micro-crowd digital twin is realized. Finally, the future research of crowd behavior experiments, crowd modeling, and theoretical system of micro human-land relationship are looked forward to from the perspective of geography. (1) New approaches such as credible micro-crowd behavior experiment should be developed. In the crowd behavior experiments, the behavior errors caused by subjective and objective factors in crowd experiments need to be calibrated qualitatively and quantitatively, and the deviation correction algorithm should be developed to obtain reliable experimental results; (2) new micro-crowd behavior modeling ideas should be proposed. Behavior models should focus on not only emergency disordered behaviors, but also those orderly behaviors under the constraints of social rules and social orders in common daily conditions; at the same time, it is also necessary to expand the rules of individual "self-interest" used in traditional behavior modeling, and the crowd interests should be introduced to build behavioral models considering both "self-interest" and "altruism" rules; (3) the deterministic and stochastic characteristics of crowd behaviors should be taken into account. In view of the complementary characteristics of AI behavior models and traditional mathematical behavior models as well as the advantages of AI model in stochastic modeling, they can be combined to achieve this goal; (4) on the basis of the digital twinning of micro-crowd, the innovative theoretical system of "micro human-land relationship" can be developed in the future, which will break through the traditional concept of "macro human-land relationship" modeling, and will make the theoretical system and modeling system of the study of human-land relationship in geography more complete, then the research of geography, GIS and Virtual Geographic Environments (VGE) on modelling of micro-crowd behaviors, micro-spaces and the applications can be promoted.