Kashgar : GIS and spatial modeling of cultural heritage sites at the heart of the Silk Road GiladUri 2020 For many centuries, Kashgar has been a major crossroads between East Asia, Central Asia, and the Indian subcontinent. Its location in the heart of the Silk Road has led to a diverse cultural heritage, and subsequently there exists a variety of cultural heritage sites, including Buddhist stupas, mazars (Islamic cemeteries), mosques, irrigation canals, and artifacts. Nowadays, the cultural heritage sites of Kashgar are under-explored and under threat of destruction. This thesis aims to produce new knowledge that would lead to a better understanding, and therefore better protection and preservation opportunities, for the cultural heritage sites of Kashgar. This is achieved by providing new information about the spatial distribution of the existing cultural heritage sites of Kashgar, establishing the relationships between the locations of the cultural heritage sites and their surrounding landscapes, and mapping and modeling the likelihood of the distribution of cultural heritages in the region through integrating Geographical Information Systems (GIS) and spatial modeling techniques. The thesis also aims to evaluate the modeling and predictive power of Weights of Evidence (WoE) - a GIS based predictive modeling method - for cultural resource management (CRM). Initially, a database of the cultural heritage sites of Kashgar was constructed by obtaining data from Chinese sources and fieldwork. The data were used to perform a site-landscape analysis that quantified the relationship between the occurrence or absence of sites and the landscape features. This information was then used to create probability models, mapping the relative likelihood of site occurrence. The information derived from the analysis and modeling can be used by local Chinese development authorities for CRM. The predictive modeling involves identifying landscape features which have significant spatial associations with the observed cultural heritage sites by WoE, and then producing probability maps of potential cultural heritage sites based on the identified landscape features by both WoE and Logistic Regression (LR). The results produced by both methods were assessed and compared. The site-landscape analysis using WoE shows how strongly different landscape features or factors are associated with the occurrences of cultural heritage sites in Kashgar. It is found that while in the oasis regions cultural heritage sites are closely related to urban and rural centers and their agricultural fields and canals, in the non-oasis regions they are correlated with grazing areas and natural rivers. The predictive models result in maps showing the probabilities of areas being of cultural significance. Such areas are in the heart of the oasis regions within the reach of canals and in the valleys of the Tashkurgan area. Model validation shows improvements in site prediction abilities of the models compared to random prediction, and suggests the combined use of WoE and LR methods to achieve better modeling outcomes. It concludes that LR generates more accurate results than WoE. However, WoE is powerful in establishing the relationships between landscape features and site occurrences and identifying the landscape features with predictive powers. Predictive modeling increases our capability of defining site-sensitive areas, a necessary step in improving prospects for site preservation and other future CRM issues. The findings from this research are also valuable for those who wish to learn about the geographic distribution of culturally significant sites in Kashgar, either as an independent landscape or as a part of the Silk Road.