Will 2017 be Remembered as a Stellar Time for Earth Observation?



Joe Francica, Managing Director Location Intelligence Solutions with Pitney Bowes discusses the growth and direction of Earth Observation satellites.  

As the mornings become darker and the nights draw in, it’s a fantastic time of year to spot some of the thousands of satellites that orbit the Earth.  Dawn or dusk are the best times to view them, and in theory you should have to wait no more than 15 minutes before one comes into sight to the naked eye. Data collected by the United Nations Office for Outer Space Affairs (UNOOSA) in 2016 reveals that there are over 4,200 satellites orbiting the planet, an increase of 4.39% over the previous year1. Around 1,459 of these are currently operational.

Satellites launched by country

The map shown below from the UCS Satellite Database compares the global development of satellites launched, by country between 1966 and 2016. It’s a fascinating view. In 1966, satellite launches were confined largely to the US, Canada, Alaska, parts of Europe and Russia. By contrast, in 2016 just a few countries remained which hadn’t launched satellites. Now, countries are accelerating their satellite programmes and pushing boundaries. This February, for example, the Indian Space Research Organisation (ISRO) became a world record holder with its launch of 104 satellites in just one rocket – the Polar Satellite Launch Vehicle (PSLV-C37).

In October of this year, satellites were launched from countries including Japan and Venezuala. QZS-4 (MICHIBIKI-4) originates from Japan, with the aim of improving GPS coverage. VRSS-2 is an Earth observation satellite launched by Venezuala2 with data collected to be used to help security forces, emergency service, health professionals and the agricultural industry.

The data-collecting mission

Of the satellites currently in operation:

  • 713 are used for communications;
  • 160 are used for technology demonstration and development
  • 105 are used for navigation and global positioning and
  • 67 for space science

In addition, 374 are currently used for Earth observation and science. Those that depend on in-car satellite navigation systems to get you from A to B might find it surprising that more than three times the number of satellites are used for Earth observation than for navigation and global positioning. The sheer number of Earth observation satellites demonstrates the far-reaching importance of the data they collect, and the value and deep insight which can be extracted from this data.

A pivotal point in the history of Earth observation

We’ve reached a pivotal point in the history of Earth observation, with key movement in the market and some fascinating, groundbreaking merger and acquisition activity, the significance of which should not be underestimated.

An article published three years ago outlined the collaboration between Harris, Draper Laboratory and Dynetics, named OmniEarth LLC. OmniEarth planned to launch a constellation of 18 satellites. At the time, it was an ambitious but achievable business plan. Since then however, OmniEarth’s business model has shifted towards analytics gleaned from other sources’ data, and in April this year it was acquired by EagleView. Lars Dyrud, president and chief executive of OmniEarth, was quoted as saying at the time of the acquisition, “By gaining access to EagleView’s world-class Pictometry image library and product infrastructure, the OmniEarth team will be able to accelerate its development of advanced analytic solutions.”4

Continuing this run of movement in the market, earlier this year Google sold Terra Bella (formerly, Skybox Imaging) to Planet, with the intention of capturing daily images of Earth, while in October MacDonald, Dettwiler and Associates (MDA) completed its acquisition of DigitalGlobe, a major global player in high resolution Earth imagery and information, creating a new brand, Maxar Technologies.

What we’re seeing in the market is the move from just capturing data to analytics. While some businesses in the industry are combining different areas of expertise, such as advanced data capture capabilities and analytics to deliver a comprehensive solution, others are specialising in individual elements of this chain. In a model that shares characteristics with the TV industry, one business might collect or create content, and another might package and publish information in a way that delivers critical insight.

Deep analytics, machine learning and critical insight

Another reason behind the pivotal point in the industry is that organizations are realizing not just the importance of the data collected, but the importance of analytics that reveal proximity patterns and trends that might otherwise go unnoticed. Data are captured and processed in near real-time. Coupled with big data platforms such as Hadoop that have the capability to incorporate geoprocessing,  deeper analytics are now possible.


Machine learning uses algorithms to generate insights from structured and unstructured data. It sifts through large data sets which might include images, text, voice, video and location to identify correlations, patterns and trends. From this, predictive analytics can be employed to determine the path of violent storms, mitigate traffic congestion or look at buying patterns from billions of financial transactions that impact inventory and distribution. Machine learning now supports accurate predictions before with high volume and velocity data processing and software solutions, but machine learning takes it to the next level.

Estimates of per capita consumption in four African countries. Stanford researchers used machine learning to extract information from high-resolution satellite imagery to identify impoverished regions in Africa. (Image credit: Neal Jean et al.)

Estimates of per capita consumption in four African countries. Stanford researchers used machine learning to extract information from high-resolution satellite imagery to identify impoverished regions in Africa. (Image credit: Jean, N., Burke, M., Xie, M., Davis, W. M., Lobell, D. B., & Ermon, S. (2016). Combining satellite imagery and machine learning to predict povertyScience353(6301), 790-794.)

In addition, some government agencies are willing to share data as part of an open data policy. For example, the European Space Agency is sharing data collected from Earth observation satellites, Sentinel 1 and Sentinel 2, and with the objective of broadening the availability and application of remotely-sensed imagery.

Data in near real-time combining with automation

Other contributing factors to the paradigm shift in the industry include the ability to capture near real-time data with new levels of automation. Both factors are identified in a report5 from MDPI entitled “A workflow for automated satellite image processing”, in which the authors talk about the world’s population increasing to 9.6 billion by 2050, and the pressure this places on food production. Earth observation data is crucial, and automation has a key part to play. “As we are entering the big data era,” says the report, “the need to establish operational image workflows that produce actionable information in a trusted, robust, and stand-alone fashion is arising.” This signals a path toward automated image processing, and the ability of platforms to process very high spatial resolution data. When it comes to the power of near real-time data, we can look at EagleView as a standout, which hopes to deliver products that empower executives at the highest level of either government or business to act on the information immediately.

Industry transformation now will protect our future

It’s no exaggeration to say that as the Earth observation industry moves through its current trajectory, so does its capacity to generate insight which has a major, life-changing effect on our planet.

Deep analysis of near real-time data on precipitation, ocean salinity levels, the Earth’s atmosphere, soil acidity, data on hazards and disasters could have an enormous impact on the preservation and protection of our planet and our population.


1 Data cited on pixelatics.com

2 Data from ucusa.org

3 Data from n2yo.com

4 Information from Space News

5 A Workflow for Automated Satellite Image Processing


Like this article and want more?

Enter your email to receive the weekly GIS Lounge newsletter: