I’ve started musing about a new project this month which I am calling OpenFarm (I even bought the domain! but more on that in a later post).  The basic premise is that I want to make precision agriculture free for farmers.  In this article I will explain what precision agriculture is, why it needs to be free, and how the scientific community can work in a more collaborative way to get farmers the insights they need to produce more with less.

As background, I have been spending the last six weeks spinning up a portfolio of interesting business models around satellite technologies, and have become particularly interested in the applications to agriculture.  Satellites are interesting for agriculture because they can see many farms at once, and can see things that the naked eye cannot (for example, using infrared light as a proxy to determine the health of a field).

The term precision agriculture has emerged to describe the practice of using satellite imagery and other remotely-sensed data to guide the more precise application of farming inputs, such as seeds and fertilizers.  Precision agriculture is powerful.  Farmers can greatly improve the yields of their crops, while reducing the cost of inputs by ensuring they are applied only where and when they are needed.

Good measurement is the first building block of precision agriculture. Many companies have popped up to offer farmers yield mapping services, which use satellite imagery to define “management zones” on a farm (e.g. by soil quality, historical yield, etc.).  Precision agriculture becomes really powerful when GPS is added to the mix.  GPS-guided tractors can know when exactly when they are in a management zone, and use a variable rate controller to adjust the rate at which fertilizers and other inputs are being applied.

The problem is that precision agriculture is painfully unaffordable, particularly for smallholder farmers.  Business models built on satellite imagery are bound to be unaffordable. There are two main providers of satellite imagery (DigitalGlobe and Astrium) who, due to the substantial cost of their satellites (upward of $400m) charge upward of $20k for a single high resolution image.  This is changing.  Significant standardization in satellite designs and launch services mean that smaller, but nonetheless capable satellites can be launched for closer to $200k, a 1000x decrease.  Even better, companies like Planet Labs, who were recently funded to launch a satellite fleet, want to make their data open to the world.

With the cost of measurement declining, there is new opportunity to get farmers the insights they need to make their land more productive, and this task should not be taken lightly. Smallholder farmers of cash crops in the poorest nations are critically dependent on their yields. In a good year they can eat. In a bad year they cannot. The problem is exacerbated by volatility in the prices of food commodities. In the 2008 Global Food Crisis, many riots ensued as the price of grains soared and the poorest people, many of them farmers, went hungry.

In the long run, this problem (often referred to under the umbrella term “Food Security”) will begin to affect us all. Human population growth is outpacing the growth of global food production, and some estimates suggest that by 2050 there will not be much more farmable land to go around.  We simply need to get better at producing more out of our farmable land, and precision agriculture offers one way to do this while conserving inputs.

So what does this all mean? It is clear that historical farming practices are relatively crude. Lord Kelvin famously stated “if you cannot measure, you cannot improve it.”  I believe the first step is open platform where farmers can access rich information about their farm.  This would include satellite imagery and climate data, but more importantly would provide a platform to capture on-the-ground observations, such as soil samples, pictures of unusual looking crops, occurrences of pests and disease, etc.  OpenFarm would use this information to generate intuitive maps for farmers, highlighting the most and least productive zones, potential problem areas, and basic recommendations to improve yields.  The platform would promote open compatibility with GPS devices and technologies that can be used to take action against the new information, and would close the loop by monitoring the yields that result.

Where it gets really interesting is when the system hits a critical mass of users that are regularly contributing their data, taking action, and recording results.  At this point, much more powerful data processing techniques such as machine learning can be used to build an intelligent recommendation engine, and produce tailored action plans for farmers. Just as Facebook can now identify your face in a photo, OpenFarm would be able to identify the pattern that indicates an infestation is imminent on a farm, and be able to tell the farmer precisely where to cull the crop (and all this for free!).

This is all much more complex than described in this article. For starters, not all crops are well understood and easy to measure.  Wheat, soy and corn have shown the greatest progress (to the point that companies like Reuters now use satellites to produce crop forecasts for commodity traders), but crops like coffee and cocoa have a ways to go.  The scientific community could contribute to OpenFarm by helping to shape crop models and the algorithms that are used provide insight to their respective farmers.

I have a ways to go to make OpenFarm a reality, and am starting to think through the best crop and geography for a pilot project. Suggestions are welcome 🙂