Data-Driven Agriculture and Role of AI in Farming: Example of Applying Fungicide for Wheat

Data-Driven Agriculture and Role of AI in Farming: Example of Applying Fungicide for Wheat

data-driven agriculture
Data-driven agriculture is not only about finding the way inputs affect outputs of farm, but also to digest the complexity of interactions between farm variables. Although this complexity in farming data can overwhelm the human mind, AI algorithms are structured to deal with it effectively. This post highlights the potential of AI in leading to data-driven decisions that are proactive rather than reactive.

What is Data-Driven Agriculture?

Data-driven agriculture means maximizing insights from your data toward your return on investment in a sustainable approach. This means to make the best with your data considering the uncertainties ahead. Agricultural activities and specifically farming are complex processes with a high number of variables involved. Thanks to advances in precision agriculture, different types of data are collected from the farms. A data driven approach provide farmers and growers actionable intelligence extracted from the data toward optimizing their practices.

different items affecting agriculture
Data-driven farming is not only about finding how inputs affect outputs of farm, but also the complexity of interactions between these variables.

How is Data Used in Agriculture?

While precision agriculture produces lots of data, extracting useful insights out of this data is a complicated task. Let’s show this challenge with a simple example. The role of nitrogen fertilizer, water, and herbicide on yield is known and pretty much established. Having all other factors unchanged, you can find the optimum amount of each to maximize the yield.
Effect of different amounts of nitrogen, water, and herbicide on yield
Effect of different amounts of nitrogen, water, and herbicide on yield

However, ‘Having all other factors unchanged’ is an assumption that never happens in farming. In reality, these variables change simultaneously and each one is affected by others. Rain and irrigation could wash down nitrogen and herbicide. Nitrogen and weeds affect the uptake of water by plants. Weeds consume part of nitrogen out of soil, and nitrogen affects the efficiency of herbicide. There are many more factors involved in farming, and as a result, data-driven decision is not a straightforward task even having all the data collected.

Nitrogen, water, and herbicide affecting each other and the yield
Role of Nitrogen, water, and herbicide on yield could not be evaluated without considering their interactions

How does AI Help in Data-Driven Agriculture?

Although complexity in data can overwhelm the human mind, AI algorithms are structured to deal with it effectively. These algorithms are designed to find hidden patterns in data and predict outcomes, which leads to data-driven decisions that are proactive rather than reactive. Training a reliable machine learning model is at the heart of any complex data-driven agriculture. This leads to the next phase which is translating the insight from artificial intelligence to an actionable intelligence where decisions can be made upon it.

Inputs (e.g., nitrogen, water, herbicide) feed into machine learning model to predict output (e.g., severity of disease)
Inputs (e.g., nitrogen, water, herbicide) feed into machine learning model to predict output (e.g., severity of disease)

A Case-Study on Application of Data-Driven Farming

We obtained wheat farms data from CIMMYT, fed part of data into the machine learning model to train it, and used the rest of the data to test the reliability of trained model. The objective is to predict severity of foliar diseases in wheat farm before occurs. Severity was quantified based on extent of damage from none to severe. Our trained model could predict severity of foliar disease with 89% accuracy.
Our trained model could predict the severity of foliar disease with high accuracy.
Our trained model could predict the severity of foliar disease with high accuracy.

Let say we inform the farmer that he or she should expect a slight occurrence of foliar diseases in the coming weeks. Based on this new piece of information, the farmer evaluates whether or not they should pay for fungicides this farming season.

making data-driven decision on the application of fungicide

This is where data-driven decision takes place. It is to make the best of the information you have even though there is uncertainty around that. One side of the equation is that the costs associated with purchasing and applying fungicides. The other side of the equation is the yield loss because of an occurrence of foliar diseases. This can be estimated from historical data, multiplied by dollar value of that loss, multiplied by how likely is that to happen, which comes from machine learning model. Comparing two sides of this equation leads to an informed decision.

Farmers can make an informed decision by comparing two sides of the equation
Farmers can make an informed decision by comparing dollar value of applying fungicide with its probable loss caused by foliar disease occurrence

This was a simple example and farmers need to evaluate more factors before making decisions, but it shows how AI and data driven approach could provide farmers with another piece of information to help them make the best decisions.

There is a gap between extensive data collected from farms and its usability for the farmers. The complexity of agricultural problems cannot be addressed fully by experts where data is too big for human mind to digest. We help farmers and agronomists to better make sense of their vast data and add to their expertise.

Get in touch with us today

To discuss how actionable intelligence from AI could benefit the farming system you manage.
Sign up for our Newsletter