What’s Missing from Ag Data at Scale: The Farmer Behind the Field

There’s a story I tell when explaining why data freshness matters in agriculture.

A couple of years ago, the Francis Scott Key Bridge in Baltimore collapsed. For weeks afterward, if you pulled up the satellite view on Google Maps, the bridge was still there — an intact structure spanning the Patapsco River. The traffic view, however, showed it was gone — because Google Maps doesn’t rely on satellite imagery alone to tell you how to navigate. It layers real-time signals on top: reported incidents, live traffic data, user updates. The satellite view tells you what the land looks like, while the traffic layer tells you what’s happening on it.

That distinction closely parallels a core problem in agricultural data. In this Google Maps example, the satellite view showed something that no longer existed — a collapsed bridge. Someone relying on that view to navigate might have attempted to drive onto a non-existent bridge. The same problem plays out with agricultural data every day. From a yield-data view, you might see a 100-acre corn farm and reach out to the operator on file. But if that operation changed hands and the data hasn’t caught up, you’re contacting the wrong person.

Many data sources tell you what farmland looks like from an aggregated, static view. DTN adds the equivalent of Google Maps’ real-time traffic layer: the signals that show what’s actually happening on the land and who controls it today, so customers across the ag value chain can act on reliable information.

In this post — the first in a series on agricultural data quality — I’ll explain what limits the utility of many agricultural data sources, why it matters for decision-making, and what a more complete approach looks like. Later in the series, I’ll go deeper on specific models and methodology and host a live Q&A. But this is the starting point: why a satellite image of a field is only the beginning of what you actually need to know.

Most ag data stops at the satellite view

Remote sensing products can tell you what’s on a field: what crop is planted, how healthy it is, when it’s harvested, and what residue it leaves behind. Freely available sources like USDA’s Cropland Data Layer provide national crop classification at 30-meter resolution. These are valuable inputs, but they answer only one question: what’s there? They cannot tell you who farms the field, whether that operator is still active, what their total operation looks like, or where their grain goes.

Platforms that collect data through grower enrollment have deep field-level records, but only for participating farmers. The data quality for enrolled users is high, while the coverage is not. If you need to understand a full supply region, a competitor’s draw area, or a grower population you don’t already have a relationship with, enrolled platforms are not built for that.

Public data aggregators — repackaging USDA county statistics, census data, and crop maps — provide useful baselines. But county averages cannot tell you which specific operations in that county have grain to sell this week or which accounts your sales team should call first.

The missing piece in all three cases is the operator-to-field linkage. That linkage is a continuously evolving agricultural knowledge graph that makes it possible to identify, for any piece of agricultural land in the United States, who farms it, what their full operation looks like, and how confident that association is. It’s what turns a field on a map into a business opportunity.

Here are some examples of what this operator-to-field linkage enables. A yield forecast tied to a specific operator tells an origination team not just that this individual farm is trending toward 190 bushels per acre, but how much on-farm storage capacity that farm has and how likely they are to move grain with harvest around the corner. A carbon-intensity score tied to an operator’s inputs and practices tells a sustainability team which suppliers in their sourcing region are already farming in ways that support their low-CI targets, without requiring anyone to enroll or self-report. A gross income estimate per operator tells a lender or input company how to segment their market by commercial scale, not by zip code.

Without the operator linkage, each of those signals exists in isolation, such as a yield number with no name attached, a carbon score that can’t be traced to a specific supplier relationship, or a storage estimate with no seller to call. The linkage is what makes them actionable.

Data quality is about decision readiness

Since the operator-to-field connection makes agricultural data actionable, data quality has a practical meaning: how current, how complete, and how well-supported is that connection? In practice, it breaks down across three dimensions:

Freshness: Change is constant in agriculture. Parcel boundaries shift; operators retire, sell, or hand off land to family members; and a dataset accurate at collection drifts materially within a single crop year. High-quality data needs to reflect how operations look today, not how they looked several seasons ago.

Coverage: Does data exist where decisions are being made? Eighty-five percent county coverage of operators and acreage may sound strong until you realize the 15% gap is concentrated in the geography where your origination team is working. Our most recent update improved acreage coverage across 10.5% of Corn Belt counties — not because the underlying satellite data changed, but because we improved how we were interpreting and linking it to operators.

Confidence: How well-corroborated is each record? An operator assignment verified across multiple independent signals — government payment records, parcel ownership data, and historical DTN records all in agreement — is a meaningfully different tool than a record carried forward with no recent validation. In Illinois corn alone, roughly 77% of operators sit at our highest confidence tier. About 15% sit at middle tiers, and 8% sit at lower tiers where the evidence is thinner, and the data should be used differently as a result.

Together, these dimensions determine whether a customer can actually act on the data — what we call decision readiness. The goal isn’t a single accuracy number; it’s knowing what level of confidence is warranted.

We’re prioritizing signal transparency

The customers who most effectively use our data are the ones who understand where it’s strongest and where it’s still improving — not because they’ve lowered their expectations, but because they’ve matched the data to the decision.

A yield estimate in the core Corn Belt in August — late in the growing season, backed by months of satellite vegetation data — is strong enough to inform a basis move or shape a procurement plan. The same model in May, early in the season and in a peripheral geography, is more of a directional read — useful for thinking through scenarios and setting early expectations, but not something you’d want to treat as a firm number before committing inventory or locking in contracts.

The difference matters practically. Using a May estimate the same way you’d use an August estimate means making confident decisions on data that hasn’t had the time to converge yet. That’s not a failure of the model; it’s a mismatch between what the data is saying and how it’s being used.

Operator data works the same way. A record verified across government payments, parcel ownership records, and DTN  Farm Market iD data — a proprietary operator-to-field matching system with over 15 years of training history — is decision-ready for outreach. A record with only historical carryover and no recent corroboration should inform prioritization, not drive a contract conversation. Communicating the difference in data confidence and building that transparency into the platform is something we’re actively working toward.

The goal isn’t to claim perfect fidelity; it’s to be specific about where the data is strong, where it’s still improving, and what that means for the decisions it supports.

Every downstream intelligence product depends on one layer getting the basics right

The operator-to-field connection is a foundational layer in DTN data platform. To understand why, it helps to see how the platform actually works.

Construction begins with data ingestion: government payments, parcel ownership, crop input and protection survey data, satellite imagery, agronomic support data, contact information, and grain bids. Each dataset can be leveraged individually to find value. Together, they allow DTN to construct a powerful digital twin of agricultural activity across the United States.

DTN has built and refined this data architecture over 15 years. Recent investments in continuous data ingestion, more sophisticated matching algorithms, and expanded feedback loops have made the data-driven insights more reliable. The foundation layer is also the one most likely to drift from reality as operators change and land changes hands, which is why DTN data quality work is concentrated there.

This is also why keeping the foundation current matters more than any single model improvement. An origination team that can rank their entire draw area by estimated available grain and knows which operators to call first — or a sales agronomist who walks into a grower conversation already knowing the full scope of that operation — is only possible if the underlying operator-to-field data reflects how things actually are today, not a year ago.

DTN is the only commercial provider that links field-level data – crops, inputs, storage, emissions – to the specific operators who farm those fields, across the entire United States without requiring grower enrollment. Built on this data foundation, the DTN Ag Hub platform tells you not just what’s on the land, but who controls it, how much they’re producing, and whether the data behind that answer is strong enough to act.

You may also like:

What’s Taking Grain Buyers Away from Growers?

How Ag Retailers Can Prioritize Growth in a Tight Ag Economy

The Most Important Decisions in Agriculture Happen Between Forecast Updates