Unless farmers fix their data management practices first, there is no chance of successfully deploying the new technologies under the precision farming.
All new technologies hitting agriculture are based on analyzing data. But there is an old truth of all analysis: crap in, crap out.
Then came artificial intelligence. Machine learning. Big data analytics. Unstructured data. The promise was that the algorithms will figure it all out themselves. Just throw the data at it, and it will spew out knowledge.
But then the old truth prevailed: crap in, crap out.
So I am very sorry to be the messenger of bad news, but unless you fix your data management practices first, you have no chance of successfully deploying any of the new technologies under the precision farming umbrella.
The problems of data management are not specific to agriculture, they are generic to all industries, and they typically include the following:
The solution to the data problems is not that hard, it just requires a bit of determination and structure:
This may look easy in theory. It's really hard in real life. My own farming business, which I run with two partners, is entering its 8th season, and we still don't have our data management fully under control. We operate 50 hectares, divided into 32 fields, with 5 different crop categories (apples, plums, pears, strawberries, raspberries), and a bunch of varieties of those crops. We have estimates and approximations, but we just can’t conclude which of the combinations has been the better business for us. We also struggle to distribute seasonal workers’ hours in those fields and varieties. And when our accounting system shows a certain cost of pesticides, we can’t easily distribute that cost on fields and crops.
In the realm of data analytics, the timestamp is one of the most important attributes. When did something happen? If you want to save the observation of the first bloom for future reference, and you take a picture of it, it’s not worth a lot without knowledge of when it happened, right? The rest of the world is happy using the Gregorian calendar for this. Farmers are not.
Plants don’t really care a lot about dates. They care about temperature. The plant’s calendar is measured in temperature days. We, therefore, need to store all data with two timestamps, one of them for us humans to know when something happened, and the other one for the plants to know when something happened. We call this the phenological timeline, and storing this for all data points at the farm is going to make a world of difference when future analytics is deployed.
As far as precision agriculture is a journey, data management needs to be the main focus at this stage of the journey. It may seem boring, but the rewards are there waiting. From good data comes many opportunities. The first of which, we´ll explore in the next blog post.
This is part three in a seven-part series on a farmer’s journey to precision agriculture.