Our Blog

Information, analyses and news for agribusiness go-to-market

    Solving Agriculture Problems with Geospatial Analysis & Data Science [Case Study]

    Posted by FMiD Team on Jan 24, 2019


    Solving Agriculture Problems

    Big Data can be intimidating. But at its most fundamental level, it's very simple and practical: data provides information that helps us solve agriculture problems.

    I often hear questions from customers like “what makes your data different?” or “where does your data come from?” The answer is powerfully simple: our data is built on 904 million acres of land and 34 million farm fields that are linked to 2.4 million farm owners and operators, creating a database with over 95 septillion datapoints.

    While numbers that large boggle the mind, they are actually critical to the problems we’re solving for our own customers today.

    Identify high-value prospects with our on-farm grain storage identification.  Click here to learn more about how it can help you.

    For example, several of our customers asked us about on-farm grain storage data. Because of the accuracy, comprehensiveness, detail and currency of our database, we were able to solve for that problem. Here are the steps we took to do that.

    Listening to customers to find actual problems to solve

    Our product ideas come from our customers. We listen. This is key to solving problems in agriculture – because you shouldn't try to solve a problem that doesn’t exist.

    As farmers continue to manage more complex and sophisticated operations, they’re going to spend less time listening to sales pitches and more time talking to people who can actually solve the problems they face. This makes access to data and information about the farm operation critical.

    In our case, we had many clients who asked us if we had data on on-farm grain storage. This led to further discussions and the definition of a market need and whether we could solve it.

    The USDA doesn’t have this data, and surveys, while popular, are limited and fraught with inaccuracies. We needed a creative approach that met the standards of data accuracy that we demand from every other part of our database.

    Solving Agriculture Problems with on-farm grain identification

    Because land is linked to growers, we can tie the grain bins we identify directly to an individual grower who operates that land. 


    Getting creative in solving agriculture problems

    After looking at various ways of using satellite imagery and LiDAR, we determined that Object-Based Image Analysis (OBIA) could analyze satellite images to find grain bins. This inevitably, though, led to a trade-off. To cover the states where grain is farmed, ensure those images were high resolution and update that data as changes took place would have resulted in an impractical cost.

    So we had to get creative.

    Over 90 percent of on-farm grain storage is in only 20 states. We were able to reduce that area of interest further by excluding land area that couldn’t contain a grain bin, like roads, bodies of water, urban areas and vegetative cover based on 10-meter imagery and NDVI.

    After we narrowed down to a workable area of interest, we then had to determine the right resolution for our images. OBIA with 1-meter imagery from the National Agriculture Imagery Program (NAIP) didn’t meet our standards for accuracy and coverage, so we switched to 40-cm Digital Globe imagery, which did provide us with the quality and coverage we needed. We then turned to crowd-sourcing to review each image to select possible grain bins to confirm they were actually grain bins, create outlines around the bin and place a point to indicate the shadow tip.

    Those images then underwent an extensive QA review. Our QA process leverages both human review and machine learning to flag potential images that were inconclusive and need to undergo additional review.

    After reviewing over 43,000 images containing over 100 million acres, we were able to identify and record dimensions on 1.1 million grain bins. Using the shadow tips, we were able to calculate height based on the solar position on the date the image was collected. We confirmed those height calculations using LiDAR data.


    solving agriculture problems with on-farm grain identification

     Combining on-farm grain data with location services helps customers find the grain bins near their current location. 

    Delivering tangible results to our customers

    The result is that we can deliver identification of on-farm grain storage and provide both the count of bins and the storage capacity of each bin. In total, we have over 1.1 million grain bins representing approximately 15 billion bushels of grain storage.

    Layering this data on top of our grower data tied to the land helps us tell customers how many bins and bushels of storage a given grower has. From there, we can deliver contact information for marketing and sales to connect with and, hopefully, start new business relationships with these growers.

    Using our data combined with some good old-fashioned problem-solving helped deliver results for our customers. This is the power of what’s possible with data, when you move beyond the mind-boggling numbers and into what the data can actually do to help grow your business.

    Click here to read the guide and learn more.

    Click to read our guide to on-farm grain storage data.