Adaptive Composite Map Projections in D3

Selecting a map projection for a visualization is a challenging problem, as different projections are better suited for different purposes, and in particular for different scales. We implemented as an extension to D3 a recent paper on the topic, which produces a composite projection between zoom levels and latitudes. The implementation ensures that a reasonable projection is chosen for any given reference frame, and that the transition between these projections is seamless. We extended the method in the paper in two key ways: (1) We interpolate paths and handle edge cases more robustly than the paper’s proof-of-concept implementation, and (2) We implemented smooth animations between map locations which provide global context for the move. A demo can be found online at http://jitouch.com/map. INTRODUCTION A significant challenge in selecting an appropriate map projection, and parametrization for that projection, for visualizing geographic data. As no projection is good in every aspect, competing criteria include: • Equal area: A projection should preserve the areas of a polynomial. For instance, the Mercator projection presents inaccurate areas near the poles. In contrast, projections such as the Hammer, Lambert Azimuthal, or the Albers Conic projection preserve area. • Conformal: A projection should preserve local angles. Mercator is an instance of such a projection; in contrast to any of the equal-area projections, which must necessarily distort angles somewhere. • Equidistant: Distances are preserved from a standard reference point. Although none of the projections in this paper have this property, an example of the projection is the azimuthal equidistant projection of the UN logo. In addition, a projection should take into account to following criteria: • Surface: The choice of projecting onto, e.g., a cylinder versus a cone is often relevant. For example, showing the entire globe on a cone or other uncommon projection may serve to confuse the user more then enlighten them. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Figure 1. A screenshot of our demo. The top screen shows the view that the user of the map projection would see; the bottom screen shows the projection as it would look from a global perspective. The particular view being shown is an interpolated projection between Albers Conic and Mercator, centered on San Francisco. • Aspect: Often it is beneficial to view the map from an oblique angle; for example, for viewing areas near the poles. • Integration with existing services: In particular, many of the maps on the Internet use the Mercator projection, including importantly Google Maps and OpenStreetMap. It’s beneficial to be able to interface with these services. A recent paper by Bernhard Jenny [1] implemented an adaptive projection, which attempts to balance the criteria above depending on the user’s absolute latitude and the scale at which they view the map. More details can be found in the Previous Work section. We implement this paper as a proposed new projection in D3 (see Figure 1), and extend it to fit the production-ready standard and level of flexibility expected of D3. In particular, we address two key challenges: