The postgis_raster extension adds support for raster (grid-based) spatial data to PostGIS. It enables you to store, analyze, and process raster data such as satellite imagery, elevation models, and other gridded datasets directly in your PostgreSQL database. Your Nile database arrives with the postgis_raster extension and its dependency postgis already enabled.

Understanding Raster Data

A raster consists of a matrix of cells (pixels) organized into rows and columns where each cell contains a value representing information such as:

  • Elevation data (DEM - Digital Elevation Model)
  • Satellite imagery
  • Temperature maps
  • Land use classification
  • Any other grid-based spatial data

A raster can store multiple layers of data, each layer is called a band. For example a band can represent elevation data, another band can represent temperature data. In satellite imagery, each band typically represents a different wavelength of light (Red, Green, Blue).

Quick Start

Let’s walk through some common operations with raster data.

Creating a Raster Table

-- Create a table for storing elevation data
CREATE TABLE elevation_data (
    id INTEGER PRIMARY KEY,
    name VARCHAR(100),
    acquisition_date DATE,
    -- Raster column with spatial reference system EPSG:4326
    rast raster
);

Loading Raster Data


-- Insert a simple 3x3 raster representing elevation data
INSERT INTO elevation_data (id, name, acquisition_date, rast)
SELECT 
    1,
    'Sample DEM',
    '2024-01-01',
    ST_AddBand(
        ST_MakeEmptyRaster(3, 3, -74.0, 41.0, 0.01, -0.01, 0, 0, 4326),
        '32BF'::text,  -- 32-bit float
        1.0, 0.0       -- pixel value 1.0, nodata value 0.0
    );

-- Update raster values (example elevation data in meters)
UPDATE elevation_data
SET rast = ST_SetValues(
    rast,
    1,                -- band number
    1, 1,            -- starting pixel
    ARRAY[
        [100.0, 120.0, 130.0],
        [110.0, 125.0, 135.0],
        [105.0, 115.0, 140.0]
    ]::double precision[][]
)
WHERE id = 1;

Basic Raster Operations

Query pixel values at a specific point:

-- Get elevation at a specific coordinate
SELECT ST_Value(rast, 1, ST_SetSRID(ST_MakePoint(-74.0, 41.0), 4326))
FROM elevation_data
WHERE id = 1;

Calculate statistics for a raster:

-- Get summary statistics for the raster
SELECT 
    ST_SummaryStats(rast) AS stats,
    (ST_SummaryStats(rast)).min AS min_elevation,
    (ST_SummaryStats(rast)).max AS max_elevation,
    (ST_SummaryStats(rast)).mean AS mean_elevation
FROM elevation_data
WHERE id = 1;

Resample a raster to different resolution:

-- Resample to 50% of original resolution
SELECT ST_Resample(
    rast,
    scalex := 0.02,    -- New pixel width
    scaley := -0.02,   -- New pixel height
    algorithm := 'NearestNeighbor'  -- Resampling algorithm
)
FROM elevation_data
WHERE id = 1;

Raster Analysis

Calculate slope from elevation data:

-- Generate slope (in degrees) from elevation raster
-- Returns a new raster where pixel values represent slope angles
-- Use ST_SummaryStats() to view actual degree values

SELECT ST_Slope(
    rast,
    1,        -- Band number
    '32BF',   -- Pixel type
    'DEGREE'  -- Output units
)
FROM elevation_data
WHERE id = 1;

Generate contour lines from elevation data

-- Generate contour lines from elevation data
-- Returns a MultiLineString geometry where each line represents locations of equal elevation
-- Interval of 10 meters means lines will be drawn at 100m, 110m, 120m elevation etc.

SELECT ST_Contour(
    rast,
    1,        -- Band number
    10.0      -- Contour interval
)
FROM elevation_data
WHERE id = 1;

Raster Properties

-- Get raster metadata
SELECT 
    ST_Width(rast) AS width,
    ST_Height(rast) AS height,
    ST_NumBands(rast) AS num_bands,
    ST_PixelWidth(rast) AS pixel_width,
    ST_PixelHeight(rast) AS pixel_height,
    ST_SRID(rast) AS srid
FROM elevation_data
WHERE id = 1;

Raster Manipulation

-- Clip raster to polygon
SELECT ST_Clip(
    rast,
    ST_GeomFromText('POLYGON((-74.01 41.01, -73.99 41.01, -73.99 40.99, -74.01 40.99, -74.01 41.01))', 4326)
)
FROM elevation_data
WHERE id = 1;

-- Reproject raster to different coordinate system
SELECT ST_Transform(rast, 3857)
FROM elevation_data
WHERE id = 1;

Best Practices

  1. Storage and Indexing:

    • Use appropriate pixel types for your data
    • Create spatial indexes on raster columns
    • Consider tiling large rasters
  2. Performance:

    • Use appropriate chunk sizes for large rasters
    • Optimize raster resolution for your use case
    • Consider using out-db raster storage for very large datasets
  3. Data Quality:

    • Validate raster data before loading
    • Handle NODATA values appropriately
    • Use appropriate resampling methods

Common Use Cases

  • Digital Elevation Models (DEM)
  • Satellite imagery analysis
  • Land use/land cover mapping
  • Temperature and climate modeling
  • Watershed analysis
  • Viewshed analysis
  • Terrain analysis
  • Environmental monitoring

Limitations

  • Large raster datasets can consume significant storage
  • Processing time increases with raster size
  • Memory usage can be high for large raster operations

Additional Resources

Was this page helpful?