Use geographic representations and geospatial technologies to investigate, analyze and communicate a geographic problem

Learning goal: Use geographic representations and geospatial technologies across the three stages of geographic problem solving — investigate, analyze and communicate — and explain how each stage depends on choices about data, layers and display.

Geographic problem solving. A geographic problem is a question whose answer depends on where: where a disease is concentrated, where a new clinic would serve the most people, where flooding is likely, where a store should open. Problem solving follows a cycle. Investigating means stating the question precisely and gathering the spatial data that bear on it. Analyzing means processing and comparing those data to find patterns, relationships and explanations. Communicating means presenting the result so that someone else can understand it, check it and act on it. Geospatial technologies — geographic information systems, global positioning systems, remote sensing and web mapping — make each stage faster and more precise, but they do not make the reasoning automatic.

A classic case. In 1854 a severe cholera outbreak struck the Soho district of London. The physician John Snow collected the addresses of people who had died and found that the deaths were concentrated near the public water pump on Broad Street; he interviewed households and investigated exceptions — for example, workers at a nearby brewery who drank beer rather than pump water and largely escaped. On that evidence he persuaded the local authorities to remove the pump's handle in early September 1854, by which time the outbreak was already declining. Only afterwards did Snow draw his famous map, marking each death and each pump on a street plan; he presented it later that year and published it in 1855, to argue that cholera spread through contaminated water rather than through bad air, the leading explanation at the time. Snow's work shows all three stages with only paper and ink: he investigated by collecting locations, analyzed by comparing the pattern of deaths with the location of the pumps and testing exceptions, and communicated afterwards with a map that others could read and check. It also shows a limit: a map of clustering is evidence of association; Snow needed the exceptions and other evidence to build a causal argument.

Investigating with geospatial technology. Today an investigator might collect field locations with a GPS receiver, download satellite imagery, and import public datasets such as census counts, road networks and elevation. Each dataset must have its coordinate system correctly defined, and the GIS must transform the layers into one common coordinate system for display and analysis; a layer whose coordinate system is missing or wrong will not line up. Each has a date and a resolution. Mixing a recent road layer with population counts from a decade earlier may produce a misleading answer. Good investigation includes a data log: source, date, resolution and known gaps for every layer.

Analyzing with a GIS. A geographic information system stores each dataset as a layer and lets an analyst ask spatial questions. Common operations include:
- overlay, stacking layers to find where conditions coincide (for example, land that is both within a floodplain and zoned for housing);
- buffer, drawing a zone of fixed distance around features (all homes within 1 kilometer of a highway);
- network analysis, measuring travel time along roads rather than straight-line distance;
- change detection, comparing images from two dates.
Each operation embeds assumptions. A 1-kilometer straight-line buffer around a clinic ignores rivers without bridges; a network analysis corrects for that but depends on an accurate, current road layer.

Communicating results. The same analysis can be communicated honestly or misleadingly. The map needs a title that states the question, a legend, a scale, a north indicator where direction matters, a date and a source. Colors should follow the logic of the data: a sequential ramp for amounts, contrasting hues for categories. The communicator should state limitations — which data were missing, what the model assumed — so that a decision-maker knows how far to trust it. A persuasive map that hides its assumptions is not good problem solving even if its conclusion happens to be right.

A worked model. Suppose, hypothetically, a county asks where to place one new ambulance station so that the most residents are within 10 minutes' driving time. Investigate: gather the road network with speed limits, residential address points or small-area population counts, and the locations of existing stations. Analyze: for each candidate site, run a network analysis to draw the 10-minute drive-time zone; overlay it on population; subtract people already within 10 minutes of an existing station; rank candidates by newly covered residents. If Site A newly covers 4,200 people and Site B 3,100 (hypothetical figures), Site A ranks first on this criterion. Communicate: map both drive-time zones, report the counts, and state the limits — speeds assume normal traffic, population counts have a date, and the criterion ignores which residents most need service. The technology produced the numbers; the geographer decided the criterion and must defend it.

The misconception. Students sometimes assume that a GIS output is objective because a computer produced it. Every output reflects human decisions about which layers to include, how to classify values, which distance measure to use and what to display. Being able to explain those decisions is what makes the analysis geographic problem solving rather than button pressing.

Worked example 1 — buffer versus network (hypothetical): A school is 800 m straight-line from a neighborhood, but a river with the nearest bridge 2 km upstream lies between them. A straight-line buffer of 1 km would count the neighborhood as "close". A network analysis along roads gives a walking route of roughly 2 km to the bridge plus the distance back downstream, far more than 1 km. Conclusion: for access questions, network distance is the appropriate measure.

Worked example 2 — overlay: Layer 1 marks the 100-year floodplain (land with a 1 percent chance of flooding in any given year according to the flood model; such floods can happen in back-to-back years); Layer 2 marks parcels zoned residential; Layer 3 marks vacant parcels. The overlay of all three identifies vacant residential land inside the floodplain — where new housing would face flood risk. The analyst reports the result together with the floodplain map's date, because floodplain maps are revised.

Worked example 3 — change detection: Two satellite images of the same valley, taken in the same month five years apart, are classified into forest and non-forest. Suppose, hypothetically, 1,200 pixels change from forest to non-forest. At 30 m resolution each pixel covers 30 × 30 = 900 square meters, so the change is 1,200 × 900 = 1,080,000 square meters, or 1.08 square kilometers. The limitation stated: cloud cover or seasonal leaf loss could be misclassified as clearing.

Problem-solving log (record task): For a local question you choose (for example, access to parks, grocery stores or libraries), complete a log with three columns — Investigate, Analyze, Communicate. List each dataset with source, date and resolution; each analysis step with its assumption; and each map element you will include.

Independent performance: Produce a map (paper or free web GIS) and a 300-word brief answering your question. Success criteria: the question is spatial and specific; at least two layers are combined with a named operation (overlay, buffer, network or change detection); the map has title, legend, scale, date and source; you state one assumption and one limitation and explain how each could change the answer; and you distinguish what the map shows (a pattern or association) from what you infer (a cause or recommendation).