Taking a space-age approach to food security

LONDON — With the world’s population expected to grow to almost 10 billion by 2050, that’s an awful lot of mouths to feed. But satellite technology can play a key part in meeting this ever-growing need, through the rise of high-resolution imagery.

“There’s an opportunity to try and do things in a way we couldn’t even dream of just a few years ago,” said Nagaraja Rao Harshadeep, lead environmental specialist and global lead for watersheds at the World Bank. “A real paradigm shift is happening in terms of the kinds of activities that we can support now using a lot of these new technologies.”

Not only have recent years seen the development of much cheaper, smaller satellites in lower-Earth orbits, but mobile devices have also boosted the ability to transfer information between tech providers and farmers. The European Space Agency’s launch of the Sentinel-2 satellites as part of the Copernicus program in the past few years has also been a big step forward in high-resolution, publicly available imagery.

Among the companies seeking to drive things forward is Planet Labs, in which the World Bank invested $20 million in 2015 through its sister organization the International Finance Corporation. The company has so far put 200 of its shoebox-sized Dove satellites in orbit and last year reached the stage where it had enough satellites to image all of Earth’s landmass every day.

That could be key for agriculture. “It allows you to not only understand change in the present tense, but to make better-informed decisions about the future based on historical patterning,” said Andrew Zolli, vice president of global impact initiatives at Planet. “Things are proceeding at breakneck pace.”

The challenge now, he believes, is to gain fuller insight into what healthy and unhealthy change looks like, and then train machine-learning algorithms to detect the difference between the two. There is a need, for example, to more easily detect when plants are not getting enough or too much water, facing flooding or droughts, and whether there are invasive pests.

Irregular plots

What complicates all this on smallholder farms in the developing world is that plots are often relatively small, with many different crops growing on them and irregular boundaries, compared with the frequently large, neat, monocropped systems of the developed world. Smallholder farmers may, for example, mix maize and coffee in the same plot, said Bruno Sánchez-Andrade Nuño, vice president for impact at Argentina-based satellite company Satellogic: “The signals of those plants are going to interfere in the images.”

So although we’re in a “revolutionary period” where we’ve got all the satellite imagery we need to solve the problem, that alone cannot give all the answers, said Zolli: “What we don’t have, and where the bottleneck is, is really good, high-quality, gold-standard labelled ground data.”

This, he said, is where nongovernmental organizations can come in, with revenue and value to be gained from continuous ground data collection that they can charge for and that they are uniquely positioned to pull together. That can, in turn, be fed back into the satellites so they can be used to better detect patterns in areas that have not seen such intensive data collection.

Zolli thinks there will be an evolution from big data to what he terms “big indicators” — highly detailed global indicators that track change in Earth’s most important systems in real time.

Meanwhile, he pointed to work by a team at Stanford University. For a study they published in the journal Proceedings of the National Academy of Sciences of the United States of America in early 2017, Stanford’s Marshall Burke and David Lobell collected ground data from thousands of smallholder plots in Kenya, in collaboration with nonprofits such as One Acre Fund and Innovations for Poverty Action, a research organization that surveyed farmers for the study. They combined this with high-resolution satellite imagery, concluding from their results that this could be used to make predictions of smallholder productivity about as accurately as traditional survey-based methods.

That suggests there is promise for satellites to accelerate this type of learning, though Zolli concedes that this all poses an enormous challenge. “It’s a moonshot-scale activity,” he said — but he thinks if things are done right, it can ensure that many millions of people are fed, increasing yield while reducing risk and lowering the impact of agriculture.

Seeking answers

Satellogic is also making tracks in this area, having launched eight satellites in low-Earth orbit between April 2013 and February this year, with four more pending in the next few months. These use high-resolution spectral fingerprinting to distinguish types of crop, stages in their growth cycle, and events such as drought or flooding.

Despite the name, said Nuño, Satellogic sees itself as more of an “answers” company than merely a satellite company. In other words, it sees more value in using its data to identify and solve the biggest challenges in smallholder agriculture than just acting as a satellite tech company.

“We need to figure out in this market of many clients and smaller profits, what is the thing that makes the most sense and helps the most people,” he said.

Meanwhile, for funded projects, once farmers have awareness of a service, one of the challenges is making sure it then remains financially viable beyond just the duration of the project, said Ruud Grim, coordinator for the Geodata for Agriculture and Water Facility at the Netherlands Space Office. To do that, he estimates a service would need to reach at least 100,000 farmers. One of the easiest ways to reach this scale, he said, is by using an “aggregator” — a company already in the agribusiness industry with ready distribution channels that can add the satellite-based service as a complementary offering.

Other ways that those in the industry suggest to help fund services that farmers will hope to access for a low price or for free are methods such as freemium content — whereby basic features are free but premium content is charged for — and cross-subsidization of services.

Sharing knowledge

Another challenge often cited is a need for more insight into the realities of what’s happening on the ground, and for the sharing of these learnings.

Among the projects it backs, the Bill & Melinda Gates Foundation ran its STARS remote-sensing study in Africa and Asia partly as a “learning project” to help assess a range of public and commercial imagery products in developing novel applications of relevance to smallholder farming, said Stanley Wood, a senior program officer for agricultural development at the foundation.

One learning is that there could be a further role for the foundation to play in increasing the availability of open calibration and ground-truth data that both public agencies and private sector companies can then use to create better remote-sensing products. Wood pointed to the Radiant Earth project that the foundation is already funding to push toward open access to geospatial data to aid such innovation.

The World Bank has also introduced tools to help spread awareness, such as the Open Learning Campus it launched in 2016 as a repository of global development learning.

There is still, of course, the problem that apart from awareness of satellite-based services, people in the most remote areas may lack consistent connectivity, though this is always improving.

Despite that, NGOs can also play a part in helping get usable, appealing applications to farmers on the ground.

“NGOs play a critical role in making accessibility to satellite imagery and aerial photography a reality for smallholder farmers via simple, tailored applications,” said Carmelle Terborgh, a member of the nonprofit and global organizations team at imagery and remote sensing company Esri. “They can develop solutions with our platform that run on mobile devices anywhere and at any time.”

“The data revolution has occurred,” she added. “Now we need an application revolution to make simplified applications that rely on advanced analytics in the cloud — and deliver just the right content at the right time to the right location where a farmer is needing it.”