Match the claim to the evidence in the image

Start with a claim close to the recorded evidence: a dark patch, a bright band, a boundary or a difference between dated observations. Then identify the measurement and processing context. An interpretation such as water, smoke or a geological unit may be reasonable, but it requires more than naming a familiar-looking shape.

NASA’s image-reading guidance emphasizes context because several physical features can share a color or texture. Shadows can resemble dark surface material; bright areas can arise from clouds, snow or reflection. A comparison band, a different viewing time or independent observations can help distinguish the alternatives.

Consider two fictional images of a lake. One appears to have a smaller dark area, but the images differ in date, viewing angle and cloud cover. Before claiming a measured shoreline retreat, align the maps, define the classification method and identify which boundary is actually visible in both observations.

Suppose a documented method finds 120 valid water-classified grid cells in the first image and 100 in the second on the same equal-area grid. That is a decrease of 20 cells, or about 16.7 percent of the original count. It remains a result of that classification and coverage, with uncertainty at mixed boundary cells.

Observation, measurement and cause differ

A supported area difference does not alone explain why it occurred. Rainfall, inflow, evaporation, human management and classification effects may need investigation, depending on the setting. The image can be one part of the evidence without answering every causal question.

Write a final statement with the quantity, location, interval, method and limitation: “Under this mapping and valid-area rule, classified area decreased between these observations.” Then state what additional information would test the proposed cause. This makes the result both useful and appropriately bounded, rather than either overconfident or dismissed as merely a picture.