1.3 Thinking Geographically: Maps, GIS, and Fieldwork

Chapter 1: Thinking Geographically - How Place, Scale, Maps, and Regions Help Us Read the World | Section 1.3: Maps, GIS, and Fieldwork
Background image: John Snow's 1854 map of cholera deaths in Soho, London.

Chapter 1 · Thinking Geographically · Section 3 of 4

How place, scale, maps, and regions help us read the world

Maps, GIS, and Fieldwork

Maps do not simply display the world. They gather evidence, reveal relationships, and make arguments. Their power depends on the questions we ask and the evidence we refuse to overlook.

John Snow's cholera map, 1854 · Public domain

A map is a machine for directing attention.

It can shrink a continent onto a page, gather a million observations into a pattern, and let us compare places that no person could see at once. Yet the same act of compression leaves things out. A road map may omit income. A climate map may omit borders. A map of reported crime may show incidents but not policing practices, population, street lighting, housing conditions, or the crimes never reported.

This is not a defect that better ink or a faster computer can eliminate. A map must select. The serious question is whether its selections help us think honestly about the problem.

What has been measured?

What has been left out?

How were the categories and boundaries chosen?

What evidence would we need on the ground?

A Map Can Turn Location into Evidence

In 1854, cholera struck the Soho district of London. At the time, many people still blamed poisonous air. Physician John Snow suspected that contaminated water was carrying the disease. He gathered addresses of cholera deaths, investigated where households obtained water, and marked deaths on a street map. The bars clustered around the public pump on Broad Street.

The map became famous because it made the geography of the outbreak visible. Yet the map did not solve the case alone. Snow combined spatial pattern with interviews, observations, comparisons among water sources, and an argument about transmission. The pattern guided the investigation; the investigation gave the pattern meaning (Clarke, McLafferty, and Tempalski 1996).

John Snow's 1854 street map of Soho, London, with black bars marking cholera deaths clustered near the Broad Street pump.
A Map of DeathEach black bar represents deaths at an address. Together they form a cluster near the Broad Street pump. The map did not prove the mechanism by itself, but it helped Snow turn scattered tragedies into a testable geographic argument.Map: John Snow, 1854 · Public domain

This is : using location and spatial relationships to extract meaning from evidence. A cluster may suggest exposure. A corridor may suggest movement. A sharp boundary may suggest a law, landform, market, or classification rule. None automatically proves a cause. Each tells us where to look next.

The map is not the answer. It is a way of arranging the evidence so that a better question becomes possible.

GIS: One Place, Many Layers

A , usually called GIS, links information to location. It can store roads, rivers, property parcels, census data, elevation, temperatures, land cover, hospitals, bus routes, interviews, photographs, and thousands of other kinds of .

GIS is more than software for drawing maps. It organizes evidence into , each tied to the same locations, and lets us compare them. Where do flood zones overlap homes? Which clinics are farthest from public transportation? Do hot surfaces coincide with sparse tree cover? Which farms combine steep slopes, erodible soil, and heavy rain?

1. People

Where are residents most sensitive to extreme heat?

Larger circle = more sensitive residents

2. Environment

Which blocks are hottest, and where is tree canopy sparse?

Orange = hotter; green dots = canopy

3. Access

Who can reach transit, parks, and public cooling sites?

Blue line = transit; cross = cooling site

4. Land and buildings

What surfaces, parcels, and structures occupy each block?

Brown = buildings; dashed lines = parcels

The geography stays fixed; the question changes. Because every layer uses the same coordinates, GIS can test where conditions overlap. A heat-relief decision might compare the people, environment, and access layers rather than treating any one of them as the whole story.

GIS commonly represents features in two broad forms. use points, lines, and polygons: a well, a road, a watershed. use grids of cells or pixels: a satellite image, an elevation surface, or a temperature map. The distinction sounds technical, but it shapes what can be measured. A bus stop is conveniently a point. Summer heat does not stop at a property line; it is better represented as a continuous surface.

One layer can reveal a pattern invisible in another. In an ordinary aerial image, a roof is merely bright. In a thermal image, it may appear dramatically cooler than nearby pavement and darker roofs. The relationship is not decorative. It can guide choices about roofing, shade, energy, and health.

Side-by-side aerial and thermal images of the same Salt Lake City area, with a large reflective roof appearing cooler in the thermal view.
One Place, Two LayersThe visible-light image records roofs, streets, and buildings. The thermal layer reveals differences in surface temperature that the eye cannot see.Image: NASA · Public domain

Now try the logic of layers yourself. The model below is fictional, so there is no hidden trivia to remember. Its purpose is to show how a decision changes as different kinds of evidence enter the map.

Interactive Geography Lab 1 of 2

Layer the City

A city has funding for one immediate heat-relief project. Turn on evidence layers, select the district where work should begin, and then test what your map can and cannot justify.

Teaching model - the same fictional city mapped four ways

Choose an evidence layer to display its map.

Surface heat

Darker means hotter. Values run from 1 to 10.

Northgate7
Riverside5
Midtown8
Eastworks10
Hillview6
Lakeside4

Tree canopy

Darker means more shade and vegetated cover.

Northgate6
Riverside7
Midtown3
Eastworks2
Hillview8
Lakeside9

Heat sensitivity

Darker means more modeled health and housing risk.

Northgate4
Riverside9
Midtown6
Eastworks7
Hillview5
Lakeside3

Cooling access

Darker means better transit and nearby public cooling sites.

Northgate8
Riverside3
Midtown9
Eastworks4
Hillview5
Lakeside7

1. Add evidence

Turn on at least three layers. Each value runs from 1 to 10. Higher heat and sensitivity mean greater concern; lower canopy and cooling access mean greater concern.

2. Choose a district

No district selected. The map is still only a collection of layers.

Remote Sensing: Evidence from a Distance

allows geographers to observe Earth without touching every part of it. Sensors aboard satellites, aircraft, and drones measure reflected or emitted energy. Some record visible light. Others record infrared, radar, or additional parts of the electromagnetic spectrum. The result may look like a photograph, but it is also a set of measurements.

Repeated observations make change visible. These two MODIS satellite views show the same part of northwestern Rondônia, a state in western Brazil near the Bolivian border. The Rio Jaciparaná runs through the center; the town of Buritis lies to the east; BR-364 crosses the northwest corner. Dark green is intact forest. Tan and pale green mark cleared land, including farms and cattle pasture. Roads enter the forest like spines, and narrow clearings spread outward from them like ribs.

Northwestern Rondônia: the same landscape, ten years apart

Satellite image of forest clearing in Rondônia, Brazil, in 2000, showing roads and branching clearings surrounded by dark green forest.
Rondônia, 2000Most of the center and west remain dark-green forest. Near Buritis and along the road toward Nova Mamoré, narrow tan clearings branch from roads in the early stages of the Amazon's characteristic "fishbone" pattern. The scale bar represents 40 kilometers.MODIS image: NASA Earth Observatory · Public domain
Satellite image of the same Rondônia landscape in 2010, showing substantially expanded forest clearing.
Rondônia, 2010The branching clearings have widened and joined. Around Buritis, much of the former forest has become a continuous patchwork of fields, pasture, settlements, and forest remnants; new fishbones also press into the forest west of the Rio Jaciparaná.MODIS image: NASA Earth Observatory · Public domain

Remote sensing can track urban growth, wildfire damage, crop stress, glacier retreat, sea ice, flooding, and coastal change across areas too large or dangerous for continuous field observation. It also has limits. Clouds can block some sensors. A pixel may contain several land covers. The time of day, season, , and processing method affect what appears. An image shows what the sensor measured, not everything that matters.

Two false-color satellite images comparing Louisiana coastal wetlands in 1976 and 1999, with more open water visible in the later image.
A Coastline ChangesSatellite observations compare Louisiana wetlands in 1976 and 1999. Slow changes that are difficult to grasp from one visit become visible across time.Image: Jesse Allen, NASA Earth Observatory · Public domain

Fieldwork: The Map Meets the Ground

For all the reach of satellites and GIS, geography cannot live entirely on a screen. A database may place a grocery store within half a mile of a neighborhood, yet a freeway, steep hill, missing sidewalk, or dangerous crossing may make it difficult to reach. A satellite may show new housing but not who was displaced. A land-use code may say "park" while residents describe the space as unsafe or unwelcoming.

brings geographers into contact with the places represented in their data. It can include observing land use, interviewing residents, collecting water or soil samples, counting traffic, recording GPS points, measuring temperature, sketching a streetscape, or walking the route a map claims is accessible.

The hardware changes; the geographic logic persists

A researcher standing in shallow surf while using field equipment to collect a water sample.
The Map Meets the GroundA USGS researcher collects a water sample at Gulf Shores, Alabama, after the Deepwater Horizon spill. Fieldwork creates and checks data that may later appear on a map.Photo: U.S. Geological Survey · CC0 / public domain
Hands using a stylus on an early handheld computer equipped with GPS and GIS mapping software.
Before the SmartphoneAn early handheld GPS and GIS computer. A phone can now perform many similar tasks, but location, observation, and evidence still have to be recorded together.Photo: Nathan Wood / USGS · Public domain

Fieldwork is sometimes called , especially when observations are used to check classifications made from imagery. The phrase is useful, but the ground does not always offer one tidy truth. Residents may disagree about a boundary, a place name, a risk, or what a neighborhood needs. Those disagreements are themselves geographic evidence.

The screen gives distance. The field gives texture. Good geography moves between them.

Maps Are Made from Choices

A polished map can feel more certain than the evidence beneath it. Lines are crisp. Colors are decisive. Labels sit exactly where they were placed. That visual authority is useful, but dangerous. Maps are arguments built from data, categories, boundaries, scale, symbols, and institutional decisions.

Sometimes a map does more than describe a landscape. It helps create one. In the 1930s, federal housing appraisal maps graded urban neighborhoods for lending risk. Categories that treated race and ethnicity as hazards became part of a larger system of exclusion from credit and investment. The colors did not merely report inequality; institutions used such classifications to help organize it.

Map of Minneapolis and Saint Paul showing 1934 Home Owners' Loan Corporation neighborhood grades on a modern street grid: green A, blue B, yellow C, and red D. Minneapolis Saint Paul
A: bestB: still desirableC: decliningD: hazardousBusiness or industrial
The Twin Cities Were Sorted TooThis reconstruction places the 1934 HOLC grades for Minneapolis and Saint Paul on a modern street grid. Green and blue marked neighborhoods considered safer investments; yellow and red marked areas labeled declining or hazardous. Students can compare the pattern with familiar features such as downtown Minneapolis, the Mississippi River, and the freeway corridors later cut through the cities.Map: Ethan Osten, using Metropolitan Council digitized HOLC data · Streets.mn, CC BY-NC-SA

Even without deliberate discrimination, ordinary technical choices alter the story:

  • Data: Who was counted, when, and by whom? Which people or activities are missing?
  • Scale: Does a national average conceal local extremes? Do large districts blur neighborhood differences?
  • : Where are the break points between "low," "medium," and "high"? Would another method produce another pattern?
  • : Which sizes, shapes, distances, or directions are distorted when a round Earth is flattened?
  • Design: Do colors, symbols, and labels clarify the evidence or exaggerate it?

Interactive Geography Lab 2 of 2

Same Data, Different Map

These six fictional districts contain the same records in every view. Change the measure and the classification method, then watch the apparent geography of need move across the city.

Build the map

1. Choose the measure

Counts show total workload. Rates divide by population so districts of different sizes can be compared more fairly.

2. Choose the classification

Equal intervals divide the numerical range evenly. Quantiles place roughly the same number of districts in each class.

Check the reasoning

For comparing the relative incidence of heat illness among districts with very different populations, which measure is the sounder starting point?

Choose an answer after experimenting with the map.

Fictional teaching data

Reported heat illnesses: total cases

Equal-interval classes divide the current numerical range into three equal parts.

Low Middle High

Selected district

Northgate

Population
2,000
Cases
18
Rate
9.0 per 1,000

With raw counts, Northgate appears less urgent than larger districts even though its population-adjusted rate is among the highest.

The records have not changed. The choice to map totals rather than rates changes which districts command attention.

The method and the data both matter

Two maps of the San Francisco Bay Area comparing choropleth and dasymetric representations of population.
The Same Data, a Different MapA spreads values across administrative units. A places population more selectively within them. Neither changes the people; the method changes the picture.Image: U.S. Geological Survey · Public domain

Disconnected roads

If digitized lines stop just short of one another, routing software may conclude that a real intersection does not exist.

Overlapping parcels

A thin sliver counted twice can inflate acreage, taxes, population, or environmental exposure.

Unassigned gaps

A narrow empty strip between boundaries can leave addresses outside every district, service area, or voting precinct.

When Digital Geometry BreaksThe old technical terms are undershoot, overshoot, sliver, and gap. What matters is the consequence: a tiny drawing mistake can become a false route, a double count, or a place that disappears from the database.
World map in the Winkel Tripel projection covered by red circles and ellipses showing changing map distortion.
Every Flat World Is DistortedTissot's indicatrices begin as equal circles on the globe. Their changing shapes and sizes show how the projection distorts area and form across the map.Map: Eric Gaba (Sting) · CC BY-SA 4.0

AI Can Accelerate the Work, Not Remove the Judgment

AI systems can classify land cover, identify buildings and roads, compare enormous archives of satellite imagery, detect change, and help convert old paper maps into searchable spatial data. In February 2026, the U.S. Geological Survey published its first agency-wide AI strategy, emphasizing faster and broader scientific analysis while also stressing responsible and trustworthy use. NASA Earthdata likewise reported in May 2025 on a foundation model trained to recognize features in MODIS satellite imagery (USGS 2026; NASA Earthdata 2025).

The gain is scale and speed. A task that once demanded years of manual inspection can sometimes be attempted across a continent. The risk is scale and speed too. An automated classifier can repeat the same mistake millions of times. Training data may be uneven. A model that recognizes roofs in one country may fail where materials, building shapes, or image quality differ. Communities with poor data can become poorly mapped, then poorly served. When errors systematically burden particular places or groups, they become .

A faster mistake is still a mistake.

Geographers therefore ask the same questions of an AI-generated map that they ask of any other: Who built the data? What was the model trained to recognize? Where was it tested? What does it miss? How certain are the classifications? Has anyone checked the result against the ground?

The Toolkit Is a Cycle, Not a Toolbox

The tools of geography are most powerful when they correct one another. A map reveals a cluster. GIS compares it with other layers. Remote sensing shows change through time. Fieldwork checks what the pixels and categories mean. Interviews reveal experience. Historical records show how the pattern developed. Statistics test whether a relationship is stronger than chance. Then the map is revised.

This is not a straight march from ignorance to certainty. It is a cycle of observation, inference, testing, and revision.

A person can own expensive software and still ask a shallow question. Another can walk through a neighborhood with curiosity and care and notice something profound. The best geography joins both kinds of vision: the wide view from above and the close view from the ground.

The Main Ideas

Use these points as a compact study guide to the section.

  • Maps direct attention by selecting and organizing evidence; they are arguments rather than neutral mirrors.
  • GIS links information to location and analyzes relationships among spatial layers.
  • Vector data represent points, lines, and polygons. Raster data represent grids or pixels.
  • Remote sensing offers broad, repeated observation, while fieldwork checks what pixels, categories, and databases mean on the ground.
  • Counts, rates, classification methods, scale, projection, and design can make the same evidence appear to tell different stories.
  • Maps can participate in systems of power, as historic housing classifications demonstrate.
  • AI can accelerate spatial analysis, but it can also accelerate error and bias. Human judgment and ground verification remain essential.

Thinking Geographically

  1. Return to John Snow's map. What does the cluster suggest, and what evidence beyond the map would you want before claiming a cause?
  2. Choose a public decision in your community. Name four GIS layers that would help make the decision and one important condition the layers might miss.
  3. Why might surface temperature measured by a satellite and air temperature experienced by a person produce different maps of heat risk?
  4. Find a map you encountered this week. Who made it, what did they choose to show, and what would change if one missing layer were added?
  5. When should an AI-generated spatial classification be trusted enough to guide policy? What kind of ground check would you require?

Key Terms and Definitions

algorithmic bias
A systematic pattern in an automated system that produces unfair or distorted outcomes for particular people, places, or categories.
choropleth map
A thematic map that shades predefined areas according to a statistical value.
classification
The process of grouping observations or values into categories that can be compared or mapped.
dasymetric map
A thematic map that uses additional information to place a statistical distribution more selectively within administrative areas.
fieldwork
The direct study of places through observation, measurement, interviews, surveys, sampling, mapping, photography, or related methods.
Geographic Information System (GIS)
A system for storing, managing, analyzing, and displaying information connected to geographic locations.
ground truthing
Checking remotely sensed or mapped information against direct observations and measurements on the ground.
layer
A separate set of spatial information that can be displayed and analyzed with other location-based data.
map projection
A mathematical method for representing Earth's curved surface on a flat map, necessarily distorting some spatial relationships.
raster data
Spatial data stored as a grid of cells or pixels, often used for imagery and continuously varying values.
remote sensing
Detecting and monitoring Earth from a distance by measuring reflected or emitted energy with instruments on satellites, aircraft, drones, or other platforms.
spatial analysis
The examination of locations, distributions, distances, connections, and spatial relationships in order to discover or test patterns.
spatial data
Information that includes a geographic position, boundary, route, area, or other locational reference.
spatial resolution
The amount of geographic detail represented by a dataset or image, often expressed by the ground area covered by each pixel.
vector data
Spatial data represented as points, lines, and polygons.

Works Cited & Further Reading

Clarke, Keith C., Sara L. McLafferty, and Barbara J. Tempalski. 1996. "On Epidemiology and Geographic Information Systems: A Review and Discussion of Future Directions." Emerging Infectious Diseases 2 (2): 85-92. https://wwwnc.cdc.gov/eid/article/2/2/96-0202_article.

Fouberg, Erin H., and William G. Moseley. 2017. Understanding World Regional Geography. 2nd ed. Hoboken, NJ: Wiley.

Monmonier, Mark. 2018. How to Lie with Maps. 3rd ed. Chicago: University of Chicago Press.

Plain, Charlie. 2025. "Applying AI to MODIS Data Analysis." NASA Earthdata, May 9, 2025. https://www.earthdata.nasa.gov/news/feature-articles/applying-ai-modis-data-analysis.

National Aeronautics and Space Administration Goddard Space Flight Center. 2011. "Amazon Deforestation in Rondônia, Brazil, 2000-2010." Scientific Visualization Studio. https://svs.gsfc.nasa.gov/10872.

National Research Council. 2006. Learning to Think Spatially: GIS as a Support System in the K-12 Curriculum. Washington, DC: National Academies Press. https://doi.org/10.17226/11019.

Nelson, Robert K., LaDale Winling, Richard Marciano, Nathan Connolly, et al. n.d. “Mapping Inequality: Redlining in New Deal America.” American Panorama, Digital Scholarship Lab, University of Richmond. Accessed August 17, 2026. https://dsl.richmond.edu/panorama/redlining/.

Osten, Ethan. 2016. "Twin Cities Redlining (HOLC) Map, 1934." Streets.mn. https://streets.mn/2016/10/24/map-monday-twin-cities-redlining-holc-map-1934/.

U.S. Environmental Protection Agency. 2026. "Measuring Heat Islands." Last updated March 10, 2026. https://www.epa.gov/heatislands/measuring-heat-islands.

Gordon, Janice M., Alison P. Appling, Alfredo Aretxabaleta, et al. 2026. Artificial Intelligence Strategy for the U.S. Geological Survey. U.S. Geological Survey Circular 1562. Published February 18, 2026. https://doi.org/10.3133/cir1562.

U.S. Geological Survey. n.d. "Geographic Information Systems (GIS) Educational Resources." https://www.usgs.gov/centers/wetland-and-aquatic-research-center/science/usgs-geographic-information-systems-gis.

U.S. Geological Survey. n.d. "What Is Remote Sensing and What Is It Used For?" https://www.usgs.gov/faqs/what-remote-sensing-and-what-it-used.

World Regional Geography in Action: Understanding a Connected World · David Rand-McKay · Chapter 1: Thinking Geographically · How Place, Scale, Maps, and Regions Help Us Read the World · Section 1.3: Maps, GIS, and Fieldwork · Updated August 2026

Comments

Popular posts from this blog

1.1 Thinking Geographically: What is Geography? How Place, Scale, Maps, and Regions Help Us Read the World

2.1 Global Connections: Globalization and the World System

1.2 Thinking Geographically: Reading the Landscape