Developments and customizations based on open source software have lower costs and higher flexibilities than the large-scale, high-cost, and high-complexity routine maintenance of commercial geographic information system (GIS) platforms that is limited by the function of the system. An increasing number of GIS applications are migrating from commercial platforms to open source GIS platform frameworks. The open source GIS platform facilitates the application of GIS and promotes the rapid development of spatial data management and processing technologies. However, developers may still face the challenge of scenarios of integration or mashup applications of tile datasets from multiple digital map sources in many GIS projects. Therefore, the data source of the base map based on the open source GIS platform must be increased to extend the space and time coverage of the base map in the study area. Cesium is an open source 3D map engine developed based on the Web Graphics Library with various characteristics, such as cross-platform, cross-browser, 2D/3D integration, and dynamic geospatial data visualization. This library offers a flexible and efficiency base map customization environment for WebGIS applications with low costs. Cesium provides access and loading support for two mainstream spatial references: WGS84 and Web Mercator coordinate systems. On the basis of these considerations, we comprehensively examined the tile data organization mechanisms and spatial reference differences of the Cesium platform and various tile data sources, such as Google maps, Tianditu, and Bing maps. Two types of tile data loading schemes, namely, static and dynamic loading, were designed for the mashup of multi-source tiles with different spatial references. (1) Experiment involving the single data source tile loading scheme illustrated that single data source tiles could be loaded successfully by the two image services, and the createTileMapServiceImageryProvider image service was suitable as a choice of single data source tile loading scheme. (2) The static and dynamic schemes showed favorable visualization effects when tiles from different data sources with the same spatial coordinate system were loaded. (3) The static tile mashup loading scheme could lead to tile image deformation when multi-source tile data with different spatial references were loaded. (4) The dynamic mashup scheme could overcome the deformation problem and shown better visualization effects than the static one when multi-source tile data with different spatial references were loaded. The proposed method that is based on Cesium and involves different spatial references and multi-source tile data in offline maps, data fusion, and integration applications can be used to extend the base map data source of GIS and remote sensing projects and has evident potential practical application values. © 2019, Science Press. All right reserved.
