Skip to main content

Trends in the Optical Commercial Remote Sensing Industry - Part 1

By GeoConnexion - 11th June 2014 - 08:45

In the first of a two-part article, Kumar Navulur, Fabio Pacifici and Bill Baugh examine the potential and expectations of optical satellite remote sensing as the geospatial industry enters its fourth eraâ©

Over the last decade, significant progress has been made in developing and launching satellites suited for earth observation, with instruments in both the optical/infrared and microwave regions of the spectra. Commercial availability of optical very high spatial resolution spaceborne imagery began more than 10 years ago with the launch of IKONOS and QuickBird, which led to an increasing interest in satellite data for mapping and precise location-based service applications. Since then, a large amount of data has been acquired, including images from newer and more complex platforms such as WorldView-1, WorldView-2, GeoEye-1, and the more recent Pleiades-1A and Pleiades-1B. â©

Currently, the potential global capacity of very high spatial resolution imaging satellites is greater than 1.8 billion square kilometers per year, which corresponds to more than 12 times the land surface area of the earth. This capacity could potentially increase to more than 2.4 billion square kilometers per year (about 16 times the land surface area of the earth) in the near future. â©

Despite the vast amounts of data collected, commercial imagery providers are finding that imagery alone does not meet all customers’ real needs. Users in many domains require information or information-related services that are focused, concise, reliable, low-cost, timely, and which are provided in forms and formats specific to a user’s own activities. â©

The commercial remote sensing industry is on the verge of an information revolution, as new satellites are developed that offer increased resolution, improved accuracies, and faster access to imagery and derived information. These trends are further aided by technology improvements in processing speeds, cloud computing, delivery mechanisms, and new information extraction techniques that will make the imagery and derived information more economical and accessible. â©

As shown in Fig.1, the evolution of the geospatial industry can be illustrated as four different eras, each characterized by its ground-breaking emphasis, namely resolution, accuracy and precision, speed, and analytics. Satellite resolution was leveraged in a way to support basic geospatial needs, where a premium was placed on the detail within the scene. â©

For years, the industry rode the “one-meter-resolution” standard that has since been surpassed by resolutions well under a half-meter. Accuracy and precision became relevant as both government and commercial enterprises focused on building maps to facilitate urban planning, infrastructure deployment, and voice-guided turn-by-turn navigation systems. Speed became a critical aspect as an expanding number of users wanted and expected on-demand, rapid access to data required for emergency planning and response, risk assessment, and monitoring. And now, as the fourth era unfolds with the expectation of both information and insight derived from the imagery, the geospatial industry is well positioned to deliver capabilities that include custom site monitoring, change detection analysis, and active monitoring of “hot events” around the world, such as natural disasters, social unrest, or man-made crises.â©


The designing and launching of more sophisticated space sensors has led to increasingly finer spatial, spectral, and temporal resolutions of data. Sensors with spatial resolutions of meter or sub-meter resolution allow the detection of small-scale objects, such as elements of residential housing, commercial buildings, transportation systems, and utilities. â©

Sensors with spectral capabilities provide additional discriminative features for objects that are spatially similar. The temporal component, integrated with the spectral and spatial dimensions, can provide critical information, such as vegetation dynamics. Finally, newer classes of satellites have high-performance camera control systems capable of rapid re-targeting, allowing the collection of dozens of images over a single target, each with a unique angular perspective.â©

Spatial resolution refers to pixel size with respect to the smallest feature that can be detected from space. The late 1990s saw the launch of the first sub-meter resolution satellite, IKONOS. Since then, satellites have been trending toward higher and higher resolutions. DigitalGlobe currently operates some of the highest spatial resolution commercial satellites with resolutions up to .41cm. â©

In the coming years, several commercial providers expect to launch satellites with 1 m resolution or better. For example, the Indian Cartosat-3 is planned to collect imagery at 25 cm resolution. Fig.2 illustrates the refinement in spatial accuracy using platforms with 1 m, 50 cm, and 30 cm resolution. For example, cars can be detected with some level of uncertainty (depending on their size) with 1 m resolution imagery, whereas 50 cm resolution allows for the delineation of cars’ windshields. Cars’ side mirrors can be detected only with 30 cm imagery, clearing a path for automated computer vision techniques permitting car model identification. It is also worth noting that yellow lines in the parking area appear clearer at 30 cm resolution, while they are barely visible at 1 m resolution.â©

Spectral resolution refers to the number of spectral bands available on a satellite. Each of the spectral bands is designed for specific applications and can range from visible, to near infrared (NIR), to short wave infrared (SWIR), to thermal bands. â©

Commercial satellites primarily have four bands in the visible and NIR bands (VNIR). DigitalGlobe’s WorldView-2 satellite was designed with eight spectral bands in the VNIR region, with the additional bands being much narrower in width (40 to 50 nm) as compared to 100 nm or broader in typical VNIR bands. Fig.3 illustrates the “walk-through” from the longest to the shortest wavelengths, of the eight spectral bands of WorldView-2 over a coastal region. Fig.3(a) shows the scene in true colour. As displayed, different features appear with different band combinations. For example, wave refraction patterns and submerged aquatic vegetation appear clear with combinations of the NIR bands, whereas structural features are visible using shorter wave visible bands, such as coastal and blue channels.â©

Radiometric resolution refers to the bits of information in the imagery. Radiometric capabilities have greatly increasing in recent years, with sensors evolving from 8-bit, to 11-bit, to 14-bit capabilities. These increased capabilities determine the quality of the images and, subsequently, the ability to extract information from them accurately and in automated fashion.â©

Temporal resolution refers to the frequency that a satellite, or constellation of satellites, can collect imagery over a given area of interest. With the increased agility provided by technologies such as controlled moment gyros, today’s satellites can take images further from nadir, greatly improving collection efficiency and allowing rapid collection of point targets. Improved temporal resolution all serves to increase area collection capability due in part to technologies permitting forward and backward scanning. Figure 4 illustrates the satellite agility of the five satellites in the DigitalGlobe`s constellation.â©

DigitalGlobe’s constellation of satellites has intra-day revisit anywhere across the globe and it is capable of collecting over 3 million square kilometers of imagery every day. The company’s archive has complete coverage of most nations and urban areas have imagery as fresh as three months old.â©

Angular resolution refers to the agility of a satellite system to collect off-nadir as well as stereo imagery. Satellites are capable of shooting high off-nadir images that can be used to measure heights of objects such as buildings or oil tanks. Multiple images over an area of interest, collected either in one pass or multiple passes, can be used to create accurate 3D models of cities as well as highly accuracy elevation models. â©

Fig.5 illustrates the process of automatically generating a realistic 3D model, from the planning of the collection as shown in (a), to the extraction of 2 m resolution Digital Surface Model (DSM) and Digital Terrain Model (DTM) illustrated in (b), to the final city model as shown in (c).â©


As location-based systems become an integral part of life, high accuracy and precision are two aspects needed to ensure that imagery and derived information can be used for actionable intelligence. â©

Imagery’s positional accuracy has been steadily improving with errors around 23 m in the early 2000s to 3 m today. Increased accuracy is primarily due to more stable orbits and innovative post processing techniques that reduce the error margins. There are several technologies that enable efficient registration of data to a base map, both imagery as well as vector base layers. This practice is referred to as “second generation ortho” where a new image is registered to a base map that is, in turn, used for maintenance and updates of geospatial databases aligned to the base map. â©

The coming years will see accuracies getting better with increased spectral resolution. Precision, on the other hand, refers to relative accuracy of images collected over time. This is an important aspect to consider when creating and maintaining multi-year geospatial databases. Fig.6 illustrates the concepts of accuracy and precision. As shown, newer platforms such as the WorldView series of satellites have an average accuracy of 4 m which can be compared to the performance of precision aerial imagery.â©

Kumar Navulu, Fabio Pacifici and Bill Baugh are all with DigitalGlobe (

Read More: Aerial Imaging LBS – Location Based Services Satellite Imaging

Subscribe to our newsletter

Stay updated on the latest technology, innovation product arrivals and exciting offers to your inbox.