Vector vs raster data models in GIS

In Geographic Information Systems (GIS), spatial data is stored, processed, and visualized using two primary formats: vector and raster. For land surveyors, GIS specialists, and revenue officials, choosing between these two data models or understanding how they interact is fundamental to precise mapping. This is especially true in modern land administration systems, such as the ongoing Andhra Pradesh land resurvey, which leverages both high-resolution raster imagery and precision vector parcel mapping.

1. The Vector data model: Precision and boundaries

The vector data model represents the real world using discrete geometric shapes. These shapes are built from coordinate pairs (X, Y, and sometimes Z for elevation) and are structured into three basic geometry types:

• Points: Represented by a single coordinate pair. Used for discrete locations like ground control points (GCPs), CORS GPS boundary stations, survey stones, or single electrical poles.
• Lines (Polylines): Formed by connecting sequence of vertices with straight segments. Used for linear features with length but no area, such as baseline measurements, roads, irrigation canals, and pipeline alignments.
• Polygons: Closed loops of coordinates enclosing an area. This is the cornerstone of cadastral mapping, used to represent individual agricultural land parcels, survey numbers, village boundaries, forest compartments, and water bodies.

Key advantages of Vector mapping in Land Surveying:

• Precise coordinate storage: Points are represented by exact coordinates, meaning there is no loss of precision regardless of zoom levels.
• Rich attribute linking: Each vector feature has a unique Identifier (ID) linked directly to an attribute database. For instance, a land parcel polygon can be linked directly to its Survey Number, Subdivision, Owner Name, Adangal/Pahani registry details, and land classification.
• Topology support: Vector models can enforce topological rules, ensuring that shared boundaries between land parcels contain no overlaps, gaps, or slivers. This is vital for legal land records to prevent boundary disputes.
• Compact file sizes: Vector geometries only store vertices and attributes, making them lightweight and easy to share as Shapefiles (SHP), GeoJSON, or GeoPackage (GPKG) layers.

---

2. The Raster data model: Continuous surfaces and imagery

The raster data model represents geography as a continuous grid of cells (pixels) arranged in rows and columns. Each cell holds a specific numerical value representing a characteristic of the land cover at that location. Aerial photographs, satellite images, and digital elevation models (DEMs) are all rasters.

Key parameters of Raster data:

• Spatial Resolution (Pixel Size): The real-world size represented by a single pixel. In Andhra Pradesh, the land resurvey uses ultra-high-resolution drone imagery with a ground sampling distance (GSD) of 5cm per pixel, whereas older satellite imagery might have a resolution of 30 meters (Landsat).
• Data bands: Multi-spectral rasters store multiple bands representing different wavelengths of light (Red, Green, Blue, Near-Infrared), which helps in identifying vegetation, crops, or soil conditions.
• Continuous variation: Ideal for continuous phenomena like land elevation (Digital Terrain Models), soil salinity, or vegetation indices (NDVI).

Limitations in Cadastral Work:

While raster data is perfect for background context, it cannot natively represent legal ownership boundaries because pixels are blocky. Zooming in too close on a raster results in "pixelation," where lines become blurry and undefined. Therefore, a raster image itself cannot have a legal database of landowners attached to individual visual lines; those lines must be traced and digitized into vector formats first.

---

3. Side-by-side comparison for Land Surveyors

Below is a technical comparison of how these two models behave in standard GIS and surveying applications:

Technical Aspect	Vector Model (Shapes & Vertices)	Raster Model (Grids & Pixels)
Data Structure	Points, Lines, Polygons (Vertices)	Row/Column matrix of grid cells (Pixels)
Ideal Use Cases	Cadastral boundaries, parcel subdivisions, roads	Drone orthophotos, satellite basemaps, elevation
Attributes	Highly flexible (Rich database tables per feature)	Single value per pixel/band (no native tables)
Precision / Accuracy	Limited only by the accuracy of the GPS/TS measurements	Limited by spatial resolution (GSD / pixel size)
File Size	Generally small (kilobytes to megabytes)	Large to massive (megabytes to gigabytes)
Editing Features	Vertices can be easily moved, snapped, or split	Requires reclassification or map algebra calculations
Andhra Pradesh Survey	Digitized FMB, CORS coordinates, LPM maps	Drone Orthophoto background, scanned village maps

---

4. Hybrid workflows: Combining Vector and Raster in AP Land Resurvey

In modern survey operations, these two models are almost never used in isolation. Instead, they are combined in a hybrid GIS environment to achieve maximum efficiency and accuracy:

Step 1: Digitizing Historical Records (Raster to Vector)

Older village maps and paper Field Measurement Books (FMBs) are first scanned. This scanning process converts the physical paper into a raster image. To make these maps useful, surveyors perform georeferencing (assigning real-world coordinates to control points on the raster map) and then digitize (trace) the boundary lines to convert them into high-precision vector polygons.

Step 2: Ground-Truthing via Drone Orthophotos (Vector over Raster)

Drones fly over the villages to capture high-resolution aerial photos. These photos are stitched together to form a georeferenced, seamless orthophoto (raster basemap). Surveyors then overlay the georeferenced FMB vector boundaries directly on top of this drone raster basemap. This allows field teams to instantly compare the legal paper boundaries with actual physical features on the ground, such as bunds, fences, or structures visible in the drone photo.

Step 3: Correcting Boundaries with CORS GPS (Vector overlay)

Any discrepancies found during ground-truthing are resolved by taking physical measurements with RTK/CORS GPS receivers. These field measurements yield precise coordinate points (vector points). These points are then used to edit and correct the boundary lines of the parcel polygon. The final product is a Land Parcel Map (LPM), which is a clean, legal vector shape linked directly to the registration database.

---

5. Crucial guidelines for Field Surveyors

• Snapping is essential: When editing vector boundaries, always enable vertex-snapping in your GIS software. Without snapping, you will introduce tiny gaps or overlapping polygons, which can lead to legal property disputes during subdivision.
• Understand resolution limits: If you are overlaying vector boundaries on a raster image with a 50cm resolution, a single pixel represents half a meter. Tracing lines based on this image can easily introduce errors of up to 1-2 meters on the ground. Always check the ground sampling distance (GSD) of your raster basemap before digitizing.
• Never trust raster lines blindly: Trees, shadows, and roof overhangs in aerial photos can obscure actual ground parcel bunds. Always rely on actual physical stone measurements (vector points) and field verification rather than guessing from raster images.

6. Summary Rule

For ownership records, legal boundaries, and area calculations, always use Vector models. For visual ground context, continuous elevation, and remote sensing verification, use Raster models. An effective GIS system combines both, overlaying precise vector cadastral structures over rich raster visual backdrops.