Landscape Schema

The AI4SH database uses two schemas for landscape data: landscape_utility holds reference catalogues and landscape holds the observations. Landscape observations provide the environmental context for soil measurements — land use, land cover, terrain attributes and similar properties that describe the broader setting of a sampling site.

Process files

File	Schema	Purpose
landscape/landscape_utility_v10_sql.json	landscape_utility	Creates landscape reference catalogue tables
landscape/landscape_observation_v10_sql.json	landscape	Creates landscape observation tables

Schema: landscape_utility

The landscape_utility schema holds reference tables for landscape classification. These are analogous to the observation_utility tables but specific to landscape-level attributes. Typical content includes land use classes, land cover classifications, and terrain descriptors.

The reference tables in landscape_utility must be populated before any landscape observation records can be inserted, because landscape observations reference these catalogs through foreign keys.

Schema: landscape

The landscape schema stores actual landscape observations linked to sampling locations. A landscape observation typically characterises the environment at or around a sampling point with attributes such as:

• Land use type
• Land cover class
• Terrain position
• Vegetation cover

These observations help contextualise soil property measurements — the same soil property can behave differently under different land use or cover conditions.

Relationship to observation schema

Landscape observations are linked to the same sampling infrastructure as soil observations (sampling logs, campaigns, datasets). A sampling location may have both soil observations in the observation schema and landscape observations in the landscape schema, connected via the shared sampling log.

This separation keeps AI4SH-specific landscape classification independent of the more generic observation infrastructure, making it easier to extend or replace the landscape classification system without affecting core soil data tables.