agent-auto-sci-geospatial
地理空间科学与遥感子 skill。用于 GIS、GeoPandas、栅格/矢量处理、CRS 审计、空间连接、缓冲区、网络可达性、OSM、遥感指数、LCZ、热暴露、绿地暴露、空间公平、空间机器学习、地图合规和论文地图设计。触发于 geospatial、GIS、remote sensing、GeoPandas、rasterio、spatial accessibility、green exposure、
SKILL.mdskills/agent-auto-sci-geospatial/SKILL.md
GitHubAgent Auto Sci Geospatial
Use this subskill for spatial data, maps, and exposure/accessibility workflows.
Fast Workflow
- Inventory all spatial layers and CRS.
- Separate vector, raster, network, remote-sensing, and tabular sources.
- Reproject before distance/area/network calculations.
- Define exposure, accessibility, availability, quality, and use separately.
- Run spatial analysis with reproducible code and sensitivity checks.
- Design maps for evidence, not decoration.
- For China locator maps, check standard map compliance before final figures.
Read references/geospatial_remote_sensing_workflows.md.
For full paper projects, this skill owns spatial data discovery, download planning, CRS/geometry audit, OSM/network accessibility, green/heat exposure indicators, LCZ/remote-sensing preprocessing, spatial equity metrics, map compliance, and handoff to analysis and figure-writing agents. It should produce explicit data paths, assumptions, and sensitivity checks.
For deeper K-Dense-style encapsulation:
references/k_dense_geospatial_mapping.md: how GeoPandas, geomaster, raster/vector, remote sensing, and spatial workflow skills are adapted.references/sport_geography_spatial_playbook.md: sport park/facility accessibility, green/heat exposure, equity, LCZ, OSM, and China map-compliance playbook.
Related Helper Skills
geomastergeopandasnetworkxagent-auto-sci-data-vizsport-geography-sci-writingsport-geography-review-bibliometric
Must Not Do
- Do not calculate distance or area in unprojected geographic CRS.
- Do not equate proximity with actual use.
- Do not treat LCZ, NDVI, or park area as interchangeable exposure metrics.
- Do not use unofficial China locator maps for publication figures.