Cartography is the science and practice of making maps, but that definition undersells what the field has become. Modern cartography sits at the intersection of satellite imagery, artificial intelligence, cognitive psychology, and political debate. The hand-drawn charts of earlier centuries have given way to automated pipelines that extract road networks from orbital photographs and three-dimensional surface models of other planets. Yet the core tension of the discipline remains unchanged: every map is a set of choices about what to show, what to hide, and how to frame a piece of the world for the person looking at it.
From Satellites to Street-Level Detail
The raw material of modern cartography is overwhelmingly digital. Satellites, aircraft, drones, and ground-based sensors generate enormous volumes of imagery, and much of the labor that once fell to human drafters now falls to algorithms. Deep learning models trained on satellite photographs can identify and trace road networks with high accuracy. One study using a modified neural-network architecture on satellite images from Massachusetts reported road-extraction accuracy above 95 percent.1e-Prime – Advances in Electrical Engineering, Electronics and Energy. Advanced road extraction using CNN-based U-Net model and satellite imagery A broader review of the field notes a shift from small-scale local studies to global-scale extraction projects, and from simple pixel-level detection to practical mapping applications.2ISPRS Journal of Photogrammetry and Remote Sensing. Deep learning-based road extraction from remote sensing imagery: Progress, problems, and perspectives
This matters because road maps are one of the most frequently updated cartographic products on the planet. Every new housing development, highway interchange, and rural track needs to appear in navigation databases within weeks, not years. Automating the extraction process means maps can keep pace with the physical world in a way that was impossible when a technician had to trace every road by hand.
Automation extends to other map features as well. Line-simplification algorithms handle the “generalization” problem: when you zoom out on a digital map, coastlines and rivers need to be smoothed so they remain legible at the smaller scale. Researchers have developed hybrid methods that segment a line into sections, analyze each section’s shape characteristics, and then simplify accordingly, preserving the visual character of a coast or border even at low zoom levels.3Pattern Recognition Letters. Hybrid line simplification for cartographic generalization
Mapping the Ocean Floor and Other Planets
Some of the most dramatic cartographic work happens in places no human eye can see directly. A landmark effort combined ship-based depth soundings with marine gravity data from the Geosat and ERS-1 satellites to produce a digital map of the global ocean floor at resolutions between one and twelve kilometers. That project revealed previously unknown features, including a seamount chain stretching roughly 1,600 kilometers across the South Pacific.4Science. Global Sea Floor Topography from Satellite Altimetry and Ship Depth Soundings Even today, much of the deep ocean remains mapped at lower resolution than the surface of Mars.
Planetary cartography has its own dedicated toolchains. A photogrammetric system called Planetary3D, developed at the Hong Kong Polytechnic University, builds three-dimensional surface models of Mars, the Moon, and other bodies from orbital camera images. It processes data from instruments aboard missions like NASA’s Mars Reconnaissance Orbiter and China’s Chang’E-2 lunar probe, producing high-quality digital elevation models through an automated pipeline.5ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Planetary3D: a photogrammetric tool for 3D topographic mapping of planetary bodies The resulting maps are not just scientific curiosities. Future landing-site selection, rover navigation, and even resource prospecting depend on accurate topographic data, making planetary cartography a practical prerequisite for space exploration.
How Maps Shape the Way You Think
Cartography does not just represent spatial reality; it actively structures how people understand and remember it. Research on spatial cognition suggests that people who spend time navigating a city tend to build internal reference frames oriented to the north, likely because that is how most maps are drawn. In one study, participants were most accurate at pointing toward unseen landmarks when they happened to be facing north, and errors grew as they rotated away from that orientation. The researchers concluded that even though participants spent far more time physically navigating the city than studying maps, their mental spatial framework appeared to derive from a north-up map orientation.6PubMed. Is the map in our head oriented north?
A follow-up study in a real-world environment found a related pattern: when people were asked to point toward magnetic north, their estimates clustered only loosely in the correct hemisphere, and their guesses frequently aligned parallel or perpendicular to the nearest road rather than to true north.7PLoS ONE. The Map in Our Head Is Not Oriented North: Evidence from a Real-World Environment In other words, the map in your head is powerfully shaped by maps you have actually looked at, but it is also distorted by whatever features happen to surround you at the moment.
There is also growing concern that digital navigation tools undermine spatial learning altogether. Research has shown that traditional turn-by-turn GPS navigation promotes passive wayfinding and leads to poor spatial learning of the surrounding environment.8PubMed Central. Rethinking GPS navigation: creating cognitive maps through auditory clues When an app tells you every turn, you stop building a mental map entirely. You arrive at your destination without knowing how you got there. That same study explored whether audio-based spatial cues could restore some of the cognitive engagement that turn-by-turn directions strip away, suggesting that the design of navigation interfaces is itself a cartographic problem with real consequences for how well people understand the places they live in.
Design Choices That Help or Mislead
The visual language of a map is never neutral. Color schemes, classification methods, symbol sizes, and shading techniques all influence what the reader notices and what they miss. A study comparing different ways to color a population-density map found that the choice of classification method and color scheme significantly affected how accurately readers interpreted the data. Among the approaches tested, a diverging color scheme outperformed others in effectiveness, based on eye-tracking measures of how quickly and correctly readers extracted information.9The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. EFFECTIVENESS OF CLASSIFICATION METHOD AND COLOR SYMBOL SCHEME ON CHOROPLETH MAP OF POPULATION DENSITY IN SPECIAL REGION OF YOGYAKARTA
Terrain representation is another area where subtle design decisions carry outsized weight. Shaded relief, the technique that makes mountains look three-dimensional on a flat page, was historically done by hand, with skilled cartographers painting shadows as if sunlight were striking the landscape from the northwest. Transferring that craft to software has been a long process. One approach combines different shading methods depending on slope: steep areas get aspect-based shading, lowlands get diffuse reflection, and flat areas get a bright neutral tone. It also simulates aerial perspective, the haziness that makes distant mountains look lighter. Tests showed that with minimal time investment, experienced cartographers using the software could significantly improve the quality of computer-generated relief compared to purely automatic methods.10Cartographica. An Interactive Approach to Analytical Relief Shading
The point is not that one color scheme or shading method is always right. It is that these choices carry information whether the cartographer intends them to or not. A poorly chosen classification method can make a densely populated area look sparse, or make a gentle hill look like a cliff. Professional cartography involves managing those distortions deliberately rather than letting defaults do the thinking.
The Politics of Place Names and Borders
Maps have never been politically innocent. The act of drawing a border, naming a river, or centering a projection on a particular continent embeds assumptions about who matters, who owns what, and whose perspective counts. Scholars in critical cartography have argued that alternative cartographic representations can be effective tools for challenging conventional geopolitical narratives, and that any single map is just one way of illustrating emerging political realities.11Geography Compass. Imagining a Critical Geopolitical Cartography
Place naming is one of the most loaded cartographic acts. European colonialism systematically renamed non-European landscapes as a way of extending imperial control over both physical environments and the people living in them.12Geography Compass. Historical geographies of place naming: Colonial practices and beyond The study of place naming, or toponymy, has undergone what scholars describe as a “critical turn,” moving beyond questions of etymology to examine the political power that naming carries.13Progress in Human Geography. Geographies of toponymic inscription: new directions in critical place-name studies When a colonizing power erases an Indigenous name from a map and replaces it, the map becomes a tool of dispossession as much as a tool of navigation.
Indigenous communities have long practiced their own forms of mapping. For over five centuries, cartographic map-making served as a technology of empire-building and the dispossession of Indigenous lands, yet Indigenous peoples themselves maintained mapping practices to share ancestral knowledge and challenge colonial rule. Scholars working in this area argue that genuinely decolonial cartography requires recentering Indigenous geographical knowledge, respecting Indigenous protocols, and ensuring active Indigenous participation in the mapping process.14Cartographica. Decolonizing the Map: Recentering Indigenous Mappings This is not a fringe academic debate. Decisions about what names appear on Google Maps, which borders are drawn as solid versus dashed lines, and whose territorial claims get visual weight are live controversies with real diplomatic consequences.
Privacy on the Map
When health researchers map disease cases, environmental exposures, or social outcomes, they face a tension between analytical value and individual privacy. Plotting exact home addresses on a map can reveal who is sick, and in sparsely populated areas even approximate locations can be identifying. The field has developed a set of techniques called geomasking to address this.
One approach, known as the “donut method,” displaces each geocoded address in a random direction by at least a minimum distance but less than a maximum distance, creating a ring-shaped zone of uncertainty around each point. Compared with simple random perturbation and geographic aggregation, the donut method performed at least about 43 percent better on privacy-protection measures while reducing cluster-detection performance by less than about 5 percent.15PubMed Central. Mapping Health Data: Improved Privacy Protection With Donut Method Geomasking In practice, this means epidemiologists can publish maps showing disease clusters without giving away individual locations.
A newer technique called “street masking” takes a different approach. Instead of requiring researchers to obtain supplemental population data to calibrate the displacement, it automatically retrieves road-network data from OpenStreetMap and uses that as the basis for relocating points. Street masking achieves comparable privacy protection to population-based donut geomasking at similar displacement distances, without demanding extra data from the user.16PubMed Central. Street masking: a network-based geographic mask for easily protecting geoprivacy The practical implication is that smaller research teams without access to detailed census data can still publish responsible maps. Despite these tools being available, geomasking remains highly underutilized in published research, leaving individual privacy at risk in many mapping studies.
Mapping for People Who Cannot See
Most maps assume a sighted user, but a significant strand of modern cartographic research focuses on tactile and non-visual maps. One project, a web application called Semantic Interior Mapology (SIM), allows anyone to trace a building’s floor plan and automatically convert it into a tactile map at different scales. The system was developed with input from blind participants and tested with users who explored the resulting maps by touch. Participants were asked to perform tasks like pointing toward locations across different map scales, finding paths, and determining turn directions during imagined walks. They completed most tasks successfully, suggesting that tactile maps generated this way could support pre-journey spatial learning for blind and visually impaired individuals.17PubMed Central. Experimental Evaluation of Multi-scale Tactile Maps Created with SIM, a Web App for Indoor Map Authoring
The challenge with tactile cartography has always been production cost and customization. A sighted person can open any map on a phone; a blind person typically needs a purpose-built physical artifact or a specialized digital interface. Tools like SIM lower the barrier to creating these maps, but the gap between mainstream digital cartography and accessible cartography remains wide. As indoor navigation becomes more important in complex environments like hospitals and transit hubs, the demand for non-visual mapping is likely to grow.
Indoor Navigation and Augmented Reality
Outdoor cartography benefits from GPS, but satellite signals degrade or vanish inside buildings. Indoor navigation has become its own research area, drawing on a range of technologies. A comprehensive review of the field notes that beyond traditional inertial navigation and satellite positioning, sensors for environmental perception, including high-definition indoor maps, LiDAR, cameras, 5G signals, and Internet-of-Things beacons, are all becoming important inputs for indoor positioning.18Satellite Navigation. Indoor navigation: state of the art and future trends
Augmented reality adds another layer. Instead of reading a map on a screen and translating it to the space around you, AR overlays directional cues onto your real-time view through a phone camera or headset. A recent comparative study evaluated two ways to build the 3D environment that an AR navigation system requires: traditional manual modeling versus automated cloud-based spatial mapping using the Immersal SDK. The manual approach is time-consuming and can introduce spatial inconsistencies with the actual building, while the automated approach captures the real geometry directly but introduces its own constraints around lighting and sensor coverage.19PubMed Central. Comparative Analysis of Automated Cloud-Based Mapping and Manual 3D Modeling for AR-Based Indoor Navigation Neither method is mature enough to displace conventional wayfinding signs, but the trajectory is clear: cartographic products are migrating from two-dimensional screens into the three-dimensional space around the user.
Crowdsourced Maps and Their Limits
One of the most significant shifts in cartography over the past two decades has been the rise of volunteered geographic information, or VGI. Platforms like OpenStreetMap let ordinary people contribute spatial data, and the results have been substantial. VGI complements conventional mapping by supporting disaster response, urban planning, environmental monitoring, and humanitarian operations, and it provides a flexible, low-cost source of spatial data particularly in underserved areas where official datasets are scarce.20International Journal of Environmental Sciences, Development & Sustainability. Integrating Volunteered Geographic Information into National Spatial Data Infrastructures: Opportunities, Data Quality Challenges, and Governance Implications in Nigeria
The weaknesses are real, though. Positional inaccuracies, inconsistent labeling, digital inequality (wealthier urban areas get mapped more thoroughly than rural ones), volunteer burnout, and institutional fragmentation all limit what VGI can deliver. A road traced by an enthusiastic but untrained volunteer may be in roughly the right place but offset by dozens of meters, or labeled with a name that does not match official records. For critical infrastructure mapping or emergency routing, those errors matter.
The governance question is also unresolved. If a national mapping agency wants to integrate crowdsourced data into its official spatial data infrastructure, it needs quality-assurance protocols that do not currently exist in standardized form. Some countries are experimenting with hybrid approaches, blending authoritative data with volunteer contributions while flagging the provenance and confidence level of each feature. The result is a kind of cartographic ecosystem where professional and amateur contributions coexist, each filling gaps the other cannot.