How to change the world from space
Over the past half-century, environmental agencies, governments, and private companies have launched hundreds of satellites specifically purposed for public or commercial use. As a result, billions of square miles of the earth’s surface have been surveyed by satellites — in some cases each day — to comprise a deep, unexplored archive of our world. Much of this data is free or relatively inexpensive, and ready to be connected with emerging computer processing and AI technology.
We are in an era of Open Source Geospatial Intelligence: with expanding availability of new tools and resources, we no longer need to be a geospatial scientist or satellite imagery analyst to apply space-derived data to research questions . Changing the world from space is a challenge, but it is a straightforward one accessible to all of us.
Today, we are no longer limited by the amount of data we can access from space, or the tools to analyze that data; but rather our own ability to imagine the potential of leveraging space-derived solutions into our world’s greatest problems.
Then what, specifically, can we do with satellite imagery? Improve the environment? Increase crop yields? Track terrorist kidnappings? Identify illegal fishing? Study nuclear proliferation? Accelerate early warning on unfolding conflicts?
Yes. Satellite imagery can be used to do all of the above, and more. In fact, the use cases and applications of satellite imagery analysis remain vastly under-discovered as more and more satellites are launched, and more advanced processing and analytics technologies emerge to extract information from that imagery. The question is not what can we do, it is “where do we start?”
Every important satellite imagery analysis project is challenging, but here are ten straightforward steps to get started:
1. Begin with the objective
How do you want to change the world? Identify the specific problem, then you can tailor the data and the methods to curate a solution. Identify your region of interest, choose your ultimate objective, then develop an approach. Importantly — do not begin the process with a particular type of technology you’d like to use. Instead, concentrate on the real-world problem that must be solved. For example, avoid the research statement: “I want to use thermal imagery and AI-driven analytics to find earthquakes as they occur all over the world”. Instead, it is imperative to be agnostic of sensor and method agnostic at this stage of your process. Reframe the question to focus only on the ultimate goal: “I want to find earthquakes as they occur all over the world.”
2. Build a foundation
Research the specific region and discipline associated with your objective. Are you interested in accurately predicting earthquakes? Ensuring a nation is honoring its cease-fire agreement? Increasing crop-yields in a developing nation? First, research the specific region and the discipline of your project. agricultural patterns, military order of battle, supply chain operations; there is a wellspring of available information and research that will give you critical background and context to begin your journey. Build a foundation of context and understanding before you download your first image. It will make a world of difference in your project.
3. Seek professional help
Geospatial analysis is an analytical method in which many different forms of earth observation data — including satellite imagery, geo-enabled environmental information, even geo-located posts from social media — are used to build visualizations that allow us to better understand geographic patterns, trends, events, and changes on earth. It’s a unique tradecraft that takes a short time to learn, but a lifetime to master. When changing the world from space, consider hiring a professional geospatial analyst with academic time-in-grade to undertake your project. There are tens of thousands of formally-trained geospatial analysts who specialize in extracting insight from satellite imagery. Many organizations — like the International Union of Geodesy and Geophysics, the American Geophysical Union, or one of many of university departments around the world can connect a problem with a particular expert that can help you design a viable approach to your solution while ensuring you don’t “reinvent the wheel”.
4. Choose your sensors
There are many types of satellites collecting dozens of different types of data over the earth’s surface. If you want to study broad environmental or land-use changes over time, you should look to the environmentally purposed satellite constellations operated by the United States’ National Oceanographic and Atmospheric Administration or the European Space Agency’s many environmentally-purposed space missions including its Copernicus-Sentinel satellites. These space missions have websites from which you can download this data, and nearly all of it is free.
However, If you want to analyze harder-to-see patterns like shipping or aircraft comings-and-goings, military exercises, endangered animal migration, etc., you’ll need higher-resolution imagery that provides more granular detail. High-resolution imagery is available for purchase, with single images beginning at about $200 per shot.
Generally, environmentally-purposed satellites will not give you nearly enough image detail to detect any object smaller than a football field. Conversely, high-resolution satellite imagery will be vastly too expensive to conduct environmental analysis over large swaths of the earth’s surface. Choose satellite data that will be most suited to effectively and economically address your question.
5. Choose your method and approach
Every geospatial question requires its own approach, but geospatial analytics methods and software already exist that can be applied — with some tailoring — to your own research. Large-scale environmental questions often require vast swathes of imagery which in turn require automated processing programs to separate the signal from the noise.
Conversely, security-related questions, like how a nation is conducting nuclear proliferation, require far less imagery in terms of ground-space, but there are currently no reliable programs to automatically detect nuanced activities and events of this nature. Machine Learning, crowdsourcing, automated object counts, traditional cartography, or legacy imagery intelligence are all viable analytical methodologies; but only one or a few of these methods may be appropriate to address the question at hand.
6. Get to work
You’ve identified your question, gathered background information and expertise, chosen the satellite datasets, and developed a sound approach. The fun part is over. Now it’s time to aggregate the satellite imagery, parse the files into the appropriate software, plug numbers, look at pictures, compare pixels, take notes, and explore the data.
You will experience the triumph of discovering something new, and the frustration of inevitable setbacks, which will likely include a software failure that occurs just before you’ve saved many hours’ worth of work. You may observe something on an image, and it will faintly remind you of something you have seen weeks ago, and realize you must retrace your steps to better identify and detect this new pattern. Inevitably, there will be moments when you no longer believe you can carry out the task.
Don’t give up: patterns and meaning will soon begin to arise. When you start to see something meaningful, you may have to iterate. Explore, experiment, and emotionally prepare yourself to return to square one to start the process over again with new eyes.
7. Don’t lose sight of your original question
Don’t become so consumed with the analytical journey that you forget your original objective, the reason you started the project in the first place. Maintain sight of the “big picture” even as you deal with the thousands of analytical details.
8. Apply scientific method
Often researchers come up with a conclusion first and work backwards from there. However sound your hypothesis may be, you cannot predict what the data is going to tell you. The imagery-derived data drives your journey. Instead of cherry-picking information to support your hypothesis, holistically develop and calibrate your hypothesis based on all the data and context you amassed and aggregated for the project. As the project continues, iterate and test the analysis to fine-tune the methods and results.
9. Report, then solicit feedback
One of the most difficult aspects of this journey is succinctly disseminating your information to the relevant audiences; regardless of their level of literacy in geospatial intelligence. Build visualizations that will resonate with key stakeholders, compelling them to act on the new information you gleaned during this process. All the while, solicit and accept feedback from peers, experts, and stakeholders. If someone detracts from a theory during your research or even after you publish, many instinctively are inclined to “die on their hill”. Try instead to set your main objective as finding a truth rather than proving the original hypothesis or burnishing a reputation.
10. Take your victory lap
Congratulations — through your geospatial journey you have discovered valuable information; results that will make a difference in regional security, agricultural health, awareness of environmental scarcity, or disaster response. In short, you have changed the world from space! Accept your accolades, then determine how you can expand the utility of your project or adapt it to other regions.
In reality, space-enabled solutions building requires teamwork, time, effort, even a bit of audacity. But today more than ever before, individuals and companies are equipped to support your endeavor with data and expertise. As with all emerging tech, learning curves abound. Practitioners must be prepared to treat new geospatial projects as a “pilot” experiment, often building a new solution by compiling established datasets and methods. And with emerging technology comes a lack of tradecraft standards, so you must double your efforts to calibrate and integrate accuracy into your solution. But by leveraging emerging processing and analytics capabilities and applying this ten-step process to your initial solution design, you are well on your way to change the world.
About the author: Allison Puccioni has been an imagery analyst for over 25 years, working within tech, government, media, and academic communities. Allison worked in the U.S. and international intelligence and defense community between 1991 and 2006, then earned her master’s degree in International Policy. In 2008, Allison established the commercial satellite imagery analysis capability for the British publication company Jane’s, publishing Open Source imagery analysis continuously for six years. In 2015, Allison joined Google to assist with the establishment of applications for its commercial small-satellites. Today, Allison is the Principal and Founder of Armillary Services, providing insight on commercial imaging satellites and associated analytics. Concurrently, Allison is an Affiliate of Stanford University’s Center for International Security and Cooperation, managing a multi-sensor imagery analysis team. She is the Geospatial Coordinator for the UN DPPA Innovation Cell.
“Futuring Peace” is an online magazine published by the Innovation Cell of the United Nations Department of Political and Peacebuilding Affairs (UN DPPA). We explore cross-cutting approaches to conflict prevention, peacemaking and peacebuilding for a more peaceful future worldwide.
