NEW
Font size
WorksheetsGeoAI Quiz
Total questions: 10
Worksheet time: 5mins
What best describes Geospatial Reasoning?
Styling basemaps
Using AI + geodata to orchestrate steps that answer spatial questions
Geocoding addresses only
Drawing buffers by hand
You need areas most similar to a known lightindustrial block. Which approach fits best?
Pixel histograms only
Embedding-based similarity search
Manual site visits only
Area-weighted centroids
Pick the correct pairing:
AlphaEarth = open-source LLM; GeoGemma = virtual satellite
AlphaEarth = geospatial foundation model; GeoGemma = open-source GeoLLM for imagery
AlphaEarth = routing API; GeoGemma = DEM
AlphaEarth = SAR sensor; GeoGemma = LiDAR
Grounding with Google Maps primarily helps to
Reduce token cost only
Anchor model outputs to verified places/POIs and facts
Export PDFs
Improve 3D visualization
In a quick sitesuitability index, the most defensible formula is
Sum of raw factors
Weighted sum of normalized factors
Random weights each run
Max of any single factor
Your changedetection flags new rooftops, but images are from different seasons. Best mitigation?
Ignore seasonality
Use multitemporal stacks/embeddings and compare likeforlike windows
Lower the threshold until changes disappear
Clip a smaller AOI
For accurate urban area/distance measurement inside one state, which CRS is most appropriate?
WGS84 (EPSG:4326)
Web Mercator (EPSG:3857)
A suitable UTM zone for the region
Any local datum at random
Which practice most improves auditability of GeoAI decisions?
Hiding weights to avoid bias accusations
Using only global datasets
Maintaining a data dictionary with source/date/license/CRS + versioning
Deleting intermediate layers
A model trained on a coastal metropolis performs poorly in an inland secondary city. This is mainly
Projection error
Domain shift
Class imbalance unrelated to space
Hardware limitation
Your shortlist ignores underground utilities and heritage buffers because those datasets were missing. The primary risk is
Simpsons paradox
Sampleselection/coverage bias leading to unsafe recommendations
Overfitting due to too many layers
CRS mismatch
