Editor’s Note: W. Brant Howard is CEO of CompassData Inc.
Satellite and aerial sensors play leading roles in the collection of data for geospatial applications. However, the accuracy of the raw or unprocessed imagery often fails to meet end user requirements. Enhancing the raw imagery and deriving complex products with accurate geometry can be achieved by processing with ground control points (GCPs). Accurate and consistent GCPs are vital to the success of many applications of remotely sensed data.
GCPs are typically captured using GPS receivers to survey coordinates of photo-identifiable points on the ground, such as features on tennis courts or high-contrast marks on concrete or asphalt. Coordinates are reported in latitude, longitude, and elevation—or northing, easting, and heights. The accuracy of the points depends on hardware, software, the processing approach, and the experience of the production team.
For geospatial applications, the most common use of GCPs is to georeference an optical satellite or aerial image to the correct position on the Earth’s surface during ortho-processing, which corrects for terrain displacement and perspective projection. The goal of an accurate image map is for every point in the processed data set to match its corresponding point on the ground with a determined level of accuracy. Other data sets such as Light Detection and Ranging (LiDAR) and Interferometric Synthetic Aperture Radar (IFSAR) surface models are also georeferenced to the terrain with GCPs.
In addition to georeferencing, GCPs are commonly used to verify geospatial data accuracy, orthorectify imagery, fuse multiple raster datasets, and calibrate new imaging sensors.
Why are GCPs Important?
Most remote sensing devices and the platforms that carry them have been calibrated to collect georeferenced data. But for many geospatial applications, the accuracy of this georeferencing is not sufficient and must be improved. This improvement is achieved after acquisition through processing with high-quality GCPs.
A raw satellite image, for example, with an advertised native horizontal accuracy of 3.5 to 5 meters may have its relative accuracy improved to 50 centimeters after being georeferenced. Processing with ground control will become even more critical as higher-resolution imaging satellites are launched.
Additionally, many UAVs and small satellite constellations entering the market do not carry inertial measurement units capable of supporting highly accurate images or videos. Post- or near-real-time processing with GCPs will be required to improve data accuracy.
Collecting Quality GCPs
For organizations engaged in mapping operations worldwide, consistency is just as critical as accuracy when it comes to GCPs. Orthoimages created for an area in the United States, for instance, must be processed with GCPs meeting the same specifications as those used to generate similar products in Europe, Asia, South America, or the Middle East.
To ensure GCPs are uniformly accurate and consistent, the collection process should be standardized and replicable worldwide. Here are important tips:
- Point collection should be performed with professional GPS receivers by surveying professionals who follow standardized, documented procedures.
- The ideal GCP is captured on a ground feature that is identifiable in a satellite or aerial image. The feature should be permanent, on flat ground with good contrast, and with no overhead obstructions.
- Any permanent GCPs can be used multiple times in the future, compared to temporary plastic panels.
- Metadata must be collected along with the GCPs, including station diagrams and ground photographs showing the exact location of the surveyed point.
- The horizontal and vertical accuracy of the GCP coordinates must meet the requirements of the end-user application. Typical GCP accuracy ranges between two and 10 centimeters in relation to a specific coordinate system.
- Quality GCPs can be saved and used repeatedly for multiple purposes in the future. GCPs can be purchased from commercial archives.
- GCPs collected in one coordinate system, datum, and epoch can be transformed into others, including future epochs.
- Small projects typically require 10 to 20 GCPs, while large projects can easily require hundreds of GCPs depending on total area and the spatial resolution of the sensor.
For the GEOINT Community, which often uses unclassified commercial imagery and digital elevation model products, high-quality, unclassified GCPs are also available from commercial sources. Processing geospatial data with these GCPs enables coalition partners to deliver accurate products to the warfighter in the field without the burdens associated with sharing classified data sets.
The most important thing to keep in mind is to rely on experienced field surveyors to collect GCPs using procedures that are standardized and applied consistently regardless of geographic location. This ensures the GCPs will be accurate and can be used for a variety of applications.