Blog: The right imagery for the job
by Charlotte Bishop is Remote Sensing Projects Manager with NPA Satellite Mapping, CGG (www.cgg.com/npa). There has never been such an abundance of satellite imagery and elevation data. It can, however, pose challenges in knowing where to start in selecting the right imagery for the job. NPA Satellite Mapping’s Charlotte Bishop traces the evolution of the market and reviews the options now on offer
When NPA Satellite Mapping was founded 45 years ago it initially focused on the application of satellite imagery for geological exploration before growing to become a leading independent supplier of satellite data and derived solutions to a global client base across a range of market sectors including oil & gas, mining, engineering, environment and defence. It is therefore a fitting moment to look back and consider the numerous ‘revolutions’ that have occurred within the satellite imagery market and how these changes impact the decisions we can now make in selecting and exploiting the optimum data set.
Most specialists would agree that the first revolution was the launch of Landsat-1, initially known as ERTS-1, in 1972, which was the first commercially available satellite mission.
The series is now on its eighth satellite and, with plans for the launch of Landsat-9 in 2023, continues to highlight the value of its medium spatial resolution multi-spectral imagery that is now publicly available. To this day, Landsat still provides the backbone to numerous remote sensing applications thanks to its unrivalled historical archive and fixed 16-day revisit period.
The next revolution was the launch of the first operational Synthetic Aperture Radar (SAR) satellite mission following the success of SEASAT in the 1970s. SAR had the advantage of providing all-weather, day/night imaging capabilities. ERS-1, operated by ESA (European Space Agency), was launched in 1991 and its data acquisition strategy resulted in a significant global archive that continues to be exploited as part of historical assessments. Long-term C-band radar continuity was subsequently provided by ERS-2 (1995), ENVISAT (2002) and, more recently, Sentinel-1A/1B (2014/2016).
At this early stage, the skies were largely dominated by lower spatial resolution satellite missions but diversity was creeping in with the availability of both optical and radar systems. It wasn’t until 1999, and the launch of IKONOS-2, that the first commercial Very High Resolution (VHR) optical satellite successfully reached orbit and imaged the world at a resolution better than a 10 m pixel.
A ground-breaking satellite of its age, it acquired imagery at 80 cm (panchromatic) and 3.6 m (multispectral) spatial resolution and became a workhorse for detailed mapping for the next 16 years. IKONOS-2 marked the start of many similar missions including Quickbird-2, Worldview, Kompsat, and Pleiades, to name but a few.
Another revolution, occurring in parallel with IKONOS-2, focused on increasing the spectral resolution (more spectral bands) of satellites to widen the range of information that could be discriminated from an image. ASTER, launched by NASA in late 1999, was the first satellite with this increased spectral range. Building on the Landsat spectral resolution, it had increased capabilities in the visible, short-wave and thermal portions of the electromagnetic spectrum and is, to this day, unmatched by any other spaceborne multispectral sensor.
Not all the advances were in the optical domain. TerraSAR-X and COSMO-SkyMed, launched within days of each other in June 2007, were the first X-band SAR missions capable of acquiring high spatial resolution data. This technology not only led to daily SAR acquisition capabilities as part of the COSMO-SkyMed constellation but also the generation of the WorldDEMTM global elevation product with a 12m grid derived from TerraSAR-X and its twin, TanDEM-X.
Relaxing the rules
For optical satellites, the next challenge was to improve spatial resolution still further. Worldview-3 was the first – and currently the only – satellite capable of collecting imagery at 30-cm spatial resolution. Launched in 2015, it prompted a relaxation of US Government restrictions by giving the commercial market access to 25 cm resolution imagery (the previous limit being 50 cm). Coupled with the increased spectral capabilities of its visible and shortwave sensor, it offers the balance of highly detailed mapping and increased feature extraction capabilities.
Although not a direct technological advancement, open access data is a revolution in itself and one that has brought significant benefits. The United States Geological Survey was the first to make all of its data freely available in the late 2000s leading, in turn, to an exponential increase in the usage of its data. Subsequently, ESA, with its extensive archive of SAR data, followed suit, setting the baseline for Sentinel. This satellite constellation, with various capabilities, is now providing data quickly and robustly to support the European ‘Copernicus’ programme. Making data open access not only makes it more accessible, but also encourages research and development. This, in turn, encourages new products and services into the marketplace.
The current revolution is the boom in constellations of small satellites such as Planet’s Doves. These missions, typically flown by innovative new operators, seek to improve temporal collection to never-before-seen levels at a fraction of the cost of the larger commercial missions, while also challenging how we ingest and use that data in new and novel ways.
These historical and contemporary satellite imagery revolutions now put us in an enviable position, with many missions, offering numerous, varied capabilities. However, this diversity of options makes for a complex landscape where determining the optimal image purchasing solution can be difficult. It invariably involves a compromise between multiple factors, both technical, economic and intended use. However, some key considerations can help with the selection. These include the following:
Image timing and date
For some applications, the most recent image may be the most important requirement. In others, a time-series may be required whereby satellite constellations can be exploited for both historical and ongoing monitoring. This can be particularly important in areas that are variable due to seasonal change or undergoing extensive development. For such applications, using a single image may not provide a truly representative view or provide sufficient context for analysis.
A 30-cm image will provide an extremely high level of detail but the tradeoff is that it covers only a small area and can be a costly option. Depending on the required mapping scale, a slightly lower resolution, even down to 1.5 m, can provide sufficient detail to derive a 1:10,000 map with the advantage that it covers an area three times the size of a 30-cm image.
Deciding whether to use an optical or SAR sensor, or a combination of both, comes down to a number of factors. An optical sensor, which relies on the reflectance properties of surface features, only operates during the daytime and its applicability can be severely limited in some areas due to cloud cover and haze. SAR, an active sensor, is capable of imaging through clouds during both day and night. In certain situations, SAR can provide some ground penetration, which may also provide useful information on shallow subsurface features. The ‘textural’ information from SAR and the ‘spectral’ information from optical has specific value for different types of projects, and combining them can provide additional insight beyond that available from either used in isolation.
Today, we can do far more with satellite data than we ever imagined possible 45 years ago, so what tomorrow could bring is a very exciting prospect. With its extensive applied satellite remote sensing experience, long-term relationships with satellite operators, and independent supplier status, NPA Satellite Mapping offers a simple, impartial entry point into the increasingly complex world of satellite imagery. Selecting the right imagery is the critical first step to unlocking actionable information.