集成地理建模与历史文献挖掘的近300年苏皖地区土地利用格局重建

Historical land use and cover change (LUCC), which is an important part of global change research, has been one of the international hot topics. Most methods to reconstruct historical LUCC take the amount of total land area or population as the quantity control factor and use the results of land suitability evaluation or modern land use pattern as the guideline. There are some weaknesses in these methods especially in improving spatial resolution, concatenating simulation results and enhancing explanatory capability. To overcome these drawbacks, this proposed study mainly aims at: 1) applying data mining technology to extract and analyze geography elements in historical documents. Geographical dictionary will be established. Gazetteers, land features, phenomenon and important events of the historical period will be explored by applying methods of Optical Character Recognition (OCR) and ancient map vectoring. These together serve as the base for geography modeling for historical LUCC reconstruction; 2) integrating both top-down and bottom-up modeling approaches and following the flow of "allocation, simulation and coordination", a historical land use spatial reconstruction model will be built by integrating geo-spatial allocation based on suitability evaluation and geo-spatial simulation based on raster unit evolution; and 3) taking the past 300 years as the study period and Su-Wan region (including Jiangsu province, Anhui province and Shanghai municipality with current administrative boundary) as the study area, a continuous historical land-use dataset (with the resolution of 1 km×1 km in a given time period of 10-20 years) will be established. In addition to explore a new idea of analyzing land-use changes and its drive mechanism, the study is also expected to provide research references to reconstructing historical land use dataset for studies with a larger area and/or a longer period.

历史时期土地利用与覆被变化是全球变化研究的重要组成部分，是开展长时期气候与环境变化模拟的关键参数。针对以适应性评价或现代土地利用格局为控制进行空间配置的传统重建方法存在的局限性，围绕深入历史地理要素获取和拓展历史土地利用格局重建方法这两项关键问题开展研究。一方面借助现代信息挖掘技术，探索对历史文献记载中地名、地物、现象、事件等地理要素的识别、解译、校正和空间化，为丰富地理建模提供基础数据；另一方面将基于适宜性的空间配置、基于栅格组织的空间演化和基于土地利用主体的空间决策方法进行综合，构建集成“配置-演化-协调”的重建方法体系。以苏皖地区为研究区，通过数据挖掘、试验模拟和综合验证，实现近300年时空连续（空间分辨率为1km×1km，时间分辨率为10-20a）的全地类重建。本项目对深入识别热点地区地表变化过程，拓展历史土地利用格局重建方法，提升中国传统农区土地利用变化规律解析等具有积极意义。

以突破传统历史土地利用格局时空重建研究在细化空间精度、连续重建结果、提升成果解释力等方面面临的限制为对象，沿袭拓展数据源和集成重建模型的整体思路，从历史文献挖掘、重建方法体系构建和全地类历史土地利用格局重建三方面开展了深入研究。在历史文献挖掘方面，以5部明清官修《苏州府志》为典型对象，采用地名演变规律分析、古今地名匹配、基层区划模拟、地名信息关联等方法，实现了古文献记载的村落、市镇、官署、寺庙、驿站、学校、园林等地理要素的自动空间化，解决了在不依赖古今地名对照词典的情形下解析、转换与空间化县以下古代地理要素的难题。在重建方法体系构建方面，将传统民居特征分析引入重建模型，结合数量控制和空间配置准则，重建形成了100m×100m空间格网下的苏皖农村居民点用地数量和空间格局。在全地类历史土地利用格局重建方面，重点开展了历史时期城镇用地和农村居住用地的重建工作，采用历史地理学的历史城市形态复原方法，复原了明清时期苏皖地区的城镇建成区，并结合现代地理数据，研究了近600年来苏皖地区城镇建成区的扩展过程、特征与城镇体系演变。同时，在总结完善历史城镇建成区重建方法的基础上，进行了明清全国范围城镇建成区的重建工作，形成了一套较为可靠的明清时期全国城镇建成区数据集。