An ancient olive tree in the garden. Mapping the deep history of land use from a single image

Landscapes with a deep history of land use are the legacy of millennial interactions between ecological, social and cultural elements, which can be investigated to obtain new knowledge about our present-day ecosystems. This paper presents the application of a supervised contextual post-classification technique to extract, from a single orthoimage, geospatial objects (classes) representing different temporalities of the same land use in a historical landscape. With a rural area of Sicily as case study and its century-old olive trees as geospatial ‘control points’, we analyse the degree of category similarity between historica
lly contingent classes of the same land use. We map and interpret from present space their dynamics of change and persistence over time, cross-validating our results with evidence from local plant microfossils (phytoliths) analysis. We demonstrate how Earth observation products and contextual geospatial analysis are multidimensional sources of information enriching our understanding of past-present landscapes and their biocultural heritage.

拥有悠久土地利用历史的景观，是生态、社会与文化要素历经数千年相互作用所留存的遗产，对其开展研究可助力我们获取有关当代生态系统的全新认知。本文提出将监督式上下文后分类技术（supervised contextual post-classification technique）应用于单幅正射影像（orthoimage），以提取代表历史景观中同一土地利用类型不同时间维度的地理空间对象（类别）。本研究以西西里岛某乡村区域为研究案例，以当地百年橄榄树作为地理空间“控制点”，分析同一土地利用类型下随历史条件变化的类别间的相似度。我们从当下的空间视角，对该区域随时间推移的变化与存续动态进行制图与解读，并借助本地植物微体化石（植硅体，phytoliths）分析结果对研究结论开展交叉验证。本研究阐明了地球观测产品与上下文地理空间分析作为多维度信息源，如何深化我们对古今景观及其生物文化遗产的认知。