利用逻辑回归的南极常年考察站选址定量研究

南极常年考察站是人类在南极地区科考、探索以及观光等活动的据点，其选址对南极自然环境产生一定影响并制约考察站本身功能。本文突破传统考察站选址方法，基于地理信息系统，采用逻辑回归模型模拟已建站情况，使用定量化方法建立指标体系，评价常年站建站适宜性程度。实验过程摒除了人工指标适宜性分类以及专家打分等过程，降低了人为因素带来的不确定性。实验证明，其对已建站情况的拟合精度大于全指标模型，可以更好地模拟常年站的选址情况。

Quantitative Research for Site-selection of Antarctic Year-round Research Stations Based on Logistic Regression

Antarctic year-round research stations are the foothold for Antarctic expedition, exploration, and sightseeing activities, whose selection affects the natural environment of Antarctica as well as the function and operational efficiency of the stations. Different from traditional site selection methods, this study utilizes the logistic regression, a quantitative method, to build the criteria system and identify their weights and model the construction environment of Antarctica with the aids from geographical information systems. Due to the elimination of artificial data classification and expert scoring process, the logistic model generated in this study largely reduced the uncertainties brought by human decisions. It turned out that the logistic model was more suitable to the construction environment than the model with all criteria, and more precise and detailed, showing the feasibility of the results.