Data scientist Pello Múgica Gonzalez partners with Taking Root to create a new approach to improve tree planting projects by using data gathered from outer space.

Well before sunrise, Elsa Gonzales gets on her motorcycle to visit Benito, a tree planting farmer that she trains in the mountains of Northern Nicaragua. Together, they walk the farm and inspect the trees. Elsa notices that grasses are taking over.

Under ideal conditions, trees can reach 2 metres in height in their first 6 months. But if grasses take over, the young seedlings will die. Elsa demonstrates how to clear a 1-metre wide row free of grass so that the trees can grow big and strong. Without this timely inspection, the trees would have died. The last 6 months of work caring for the seedlings in the nursery, preparing the land, and planting the trees would have been for nothing.

Elsa is an extension agent (or community technician), whose job is to support farmers improve yields and access better markets. Extension agents are the interface between every farmer cooperative, non-profit organization, or company that works with smallholder farmers. Like millions of other extension agents around the world, Elsa spends her days travelling around the countryside visiting and supporting farmers.

What’s different about Elsa is that she visited Benito’s farm that morning as opposed to one of the other 200 farmers that she supports because of an alert she received on her phone. She’s part of a software pilot project designed to increase the effectiveness of extension agents.

Data-driven decision making is making its way into many industries but doing so for agricultural extension agents in remote regions of developing countries is extremely difficult. This is because it’s really hard to isolate the effect of an extension agent’s visit to a farm from all the other variables that could have influenced outcomes. For example, did Benito’s farm end up over-performing because of Elsa’s visit, or was it because he is a great farmer? Or was it the fertility of his farm or the favourable weather? Even if you could collect all of this information, the cost of doing so could easily become cost prohibitive.

To tackle this problem, Taking Root, an organization that grows trees with smallholder farmers in Nicaragua with a bent for data-driven solutions teamed up with Pello Múgica Gonzalez (no relation to Elsa Gonzalez), a data scientist from Belgium with a passion for international development.

Big data to help small farmers

Given enough data to analyze, answers can be found. The real challenge is collecting lots of good data cost-effectively. To solve this problem, Taking Root developed a mobile app called Farm-Trace and started asking its extension agents to simply record what type of intervention they did every time they visit a farmer and to take photo evidence of the fact. This information is then supplemented with the time and location of the activity given the phone’s GPS and clock. And because the location of all the farms are geo-fenced using the same app, they confirmed that the visit really took place.

Pello then collected and analyzed satellite images of every farm within Taking Root’s program starting from the moment Taking Root’s extension agents started recording their visits to the present.

Satellite data is much more than just imagery because the sensors capture other wavelengths’ (e.g. infrared and near-infrared) that are not visible to the human eye. These wavelengths are extremely useful for detecting healthy vegetation and to do various analyses such as landcover classification.

Pello Múgica Gonzalez looked at the change in land cover using the satellite data over a 1-year time period and looked for patterns in the data related to the number of visits each farmer received, the specific agent who visited them, and the type of training that they received.

After analyzing the data, he found that the timing, the type of training and the person doing the training all helped explain the difference they could observe from outer space. Understanding what works, when and where can really improve the effectiveness of farmer extension services. He can then use the data to develop predictive algorithms that recommend fixing problems before they even arise.

“Growing a tree in rural regions of Nicaragua is a low-tech low-cost process. You dig a hole, place a seedling in it and fill the hole with dirt. Every few months, you weed around the trees until it is large enough to fend for itself. Technological solutions rarely make sense in this context and usually just make things more expensive. What was really cool about this approach was the invisible technological layer to the whole thing and how the extremely high tech merged so seamlessly with the extremely low tech”, explains Pello.

“In the next year, we plan on completely automating the process of acquiring and processing new satellite imagery on a continuous basis and developing a machine learning algorithm that looks for correlations between changes in individual farms and the trainings that they receive. We want extension agents to receive recommendations in as near real-time as possible so that we can really stretch the impact we get out of our limited budgets.

Given that trees are indispensable in limiting the impacts of climate change and the fact that there are a billion farmers that could be growing more trees on their farms, it’s imperative that we find really cost-effective ways of doing so”, explains Dr Kahlil Baker, Taking Root’s executive director.

Monitoring the change in vegetation from outer space can be a great solution for global-wide projects. On the level of local reforestation programs, Mugica Gonzalez’s study exposes even greater value. By combining satellite-derived data with data that was collected on the ground, more in-depth analyses of reforestation projects can be executed. It allows decision makers to detect dysfunctionalities and opportunities that can have a systematic impact on the long-term.

Development organizations do all sorts of interventions without even knowing if those interventions actually help. Using a data-based approach for reforestation, monitoring and evaluation become a lot easier. Mugica Gonzalez’s research illustrates the opportunity for reforestation projects to evaluate their efforts and increase their impacts by using the power of data.

Click HERE to read more about Taking Root’s technology platform and how it works.