Preserving Geospatial Metadata During Raster Processing
Raster geospatial data finds application in various areas such as geographic information systems (GIS), remote sensing, satellite images, aerial mapping, environmental monitoring, and spatial analysis. Unfortunately, during raster processing, one may accidentally lose or distort some of the vital metadata that includes CRSs, geotransforms, pixel resolution, NoData value, and image acquisition details.
This metadata plays a key role in ensuring the accuracy of spatial operations and preserving data integrity.
What Is Geospatial Metadata?
Geospatial metadata is the data that describes features and geospatial characteristics of the raster. Unlike actual pixel values, geospatial metadata allows us to interpret and place the raster correctly.
Examples of geospatial metadata include:
Coordinate Reference System (CRS): Describes how raster coordinates correspond to geographical location.
Geotransform: Describes geographic origin, size, and orientation of the raster.
Pixel size: Describes raster dimensions.
Ground Sampling Distance (GSD): The actual area represented by each raster pixel.
NoData: Value that represents the pixel that contains invalid information.
Band information: Information about spectrally separated bands of the raster.
Acquisition information: Can include date, sensor name, and more.
Metadata tags: Other technical and/or raster-specific data.
Thus, for example, a GeoTIFF can contain not only the raster pixel values but also geographic information required for correct placement of the raster on the map by GIS software.
Why Is Metadata Preservation Important?
Typical raster operations include reprojection, resampling, clipping, mosaicking, compression, and format conversion. Each raster operation has the potential to modify metadata.
When metadata is destroyed, the raster could have valid pixel values, but it will be impossible for the raster to line up with any other spatial data layers properly.
Preserving metadata helps with:
Correct Spatial Referencing
CRS and geotransform are what determine the geographic location of the raster pixels. Without that information, it will be placed incorrectly by your GIS software.
Correct Spatial Analysis
Resolution, extent, alignment, and NoData definition are all crucial components of raster analysis. If the metadata is incorrect, the analysis results will be incorrect.
Interoperability
Correctly defined metadata makes it possible to work with a raster dataset across multiple applications like GDAL, QGIS, ArcGIS Pro, Python, or cloud geospatial systems.
Reproducible Workflows
Metadata is a great source of valuable information about the creation and processing history of the dataset.
Common Raster Processes That Affect Metadata
The various raster processes will affect metadata in different ways depending on the process itself.
Reprojection
In this process, the raster data is transformed from one CRS into another. This means that the output raster will have to get the new CRS together with the geotransformation.
For example, when raster data is being converted from geographic coordinate system like EPSG:4326 to the new projected coordinate system, the spatial reference and the pixel dimension and resolution will be affected.
Just copying the CRS metadata to the output raster will be erroneous.
Resampling
Resampling changes the raster's pixel size or dimensions. Methods such as:
Nearest neighbor
Bilinear interpolation
Cubic convolution
Average
Mode
can affect raster values and spatial resolution.
After resampling, the output metadata should accurately describe the new pixel size and dimensions.
Clipping
Clipping creates a geographic subset of a raster. The output raster should maintain the correct CRS while modifying its extent and geotransform parameters.
Mosaicking
Mosaicking is the combination of multiple raster datasets into one. Care should be taken regarding metadata when rasters have different:
CRS
Resolution
Extents
NoData values
Band structure
A good mosaic process should align the spatial properties of datasets first before mosaicking them together.
Format Conversion
Conversion from one raster format like GeoTIFF to MrSID, or other raster formats (COG and JPEG2000), may determine which metadata attributes are possible.
For example, conversion from a georeferenced raster to a raster format that does not support the relevant spatial metadata could lead to loss of geographic data.
Preserving Metadata with GDAL
GDAL (Geospatial Data Abstraction Library) is one of the most widely used tools for processing raster and vector geospatial data.
A basic raster conversion can be performed using:
gdal_translate input.tif output.tifWhen processing data, it is important to inspect the source and output metadata:
gdalinfo input.tifThe output can provide information about:
CRS
Raster dimensions
Pixel size
Geographic extent
NoData values
Raster bands
Metadata tags
After processing, running gdalinfo on the output file can help verify whether critical metadata has been retained.
For more complex workflows, GDAL tools such as gdalwarp can handle reprojection and resampling while creating appropriate output georeferencing information.
gdalwarp -t_srs EPSG:3857 -r bilinear input.tif output.tifHere, the output CRS is explicitly defined, while bilinear resampling is used to generate the output raster.
Preserving Metadata with Python
Python provides several libraries for working with geospatial raster data, including Rasterio, GDAL, Xarray, and rioxarray.
Rasterio allows metadata to be accessed through a raster dataset profile.
import rasterio
with rasterio.open("input.tif") as src:
profile = src.profile
data = src.read()
print(profile)When creating a new raster, the source profile can be updated for the characteristics of the processed dataset.
profile.update(
width=new_width,
height=new_height,
transform=new_transform
)
with rasterio.open("output.tif", "w", **profile) as dst:
dst.write(data)This approach helps preserve important properties while explicitly updating metadata that changes during processing.
Metadata Preservation in Raster Compression
Compression can significantly reduce raster storage requirements, particularly for large satellite and aerial imagery datasets.
However, compression workflows should be designed so that geographic metadata remains available after compression.
Lossless and lossy compression have different effects on raster pixel values, but both should be evaluated for:
CRS preservation
Geotransform preservation
NoData handling
Band metadata
Image dimensions
Pixel resolution
Auxiliary metadata
For geospatial workflows, reducing file size should not come at the expense of spatial reference information.
GeoTIFF and MrSID Metadata
GeoTIFF is widely used for georeferenced raster datasets because geographic information can be stored alongside raster data.
MrSID extends the GeoTIFF format with an organization that enables efficient access to raster data.
When creating a MrSID, workflows should verify both the file's internal organization and its geospatial metadata. Important properties include:
CRS
Transform
Resolution
Tile structure
NoData value
Compression settings
Tools such as GDAL can be used to inspect the resulting file and verify that the expected spatial information remains intact.
It is imperative to protect geospatial metadata in raster operations for reasons of ensuring spatial accuracy, interoperability, and reliability of analysis. Processes like reprojection, clipping, resampling, mosaicking, compression, and format transformation may change some of the attributes of the raster; hence, it is imperative to verify the metadata manually.
The use of programs like GDAL and Rasterio alongside proper metadata verification will guarantee that the processed rasters are geographically accurate.
For firms dealing with huge satellite, aerial, and other geospatial rasters, it is necessary to treat the metadata as an essential part of the raster and not just additional information.
For more information or any questions regarding the LizardTech suite of products, please don't hesitate to contact us at:
Email: info@geowgs84.com
USA (HQ): (720) 702–4849
(A GeoWGS84 Corp Company)
Comments