Skip to main content
U.S. flag

An official website of the United States government

Eyes on Earth Episode 8 – Assessing America’s Cropland

Right-click and save to download

Detailed Description

Every year, the USDA’s National Agricultural Statistics Service (NASS) uses data from satellites like Landsat to estimate crop types and crop yields in the United States. The result is the Cropland Data Layer (CDL), which offers an annual look at more than 100 crop categories across the country. In this episode, Dave Johnson with NASS discusses how Landsat can identify different crops, providing a valuable economic tool for agriculture.




Public Domain.


YOUNG: Hello everyone. Welcome to this episode of Eyes on Earth, a podcast that focuses on our ever changing planet and on the people from across the United States and the world who use remote sensing to help monitor and study the health and well-being of our planet. Iím your host, Steve Young. Todayís guest is Dave Johnson, a senior geographer with the US Department of Agricultureís National Agricultural Statistics Service or NASS. He is also a member of the Landsat Science Team. Welcome, Dave.

JOHNSON: Hi. Thanks for having me.

YOUNG: So, what can satellite sensors tell us about the health and well-being of crops on the surface of the Earth?

JOHNSON: The satellites can tell us a lot about whatís going on, on the surface of the Earth and they can do it quite efficiently and in a rapid manner versus you or I driving around in a car on a given afternoon. We use the satellites to estimate the crop production. That's one of the major goals and mandates of the National Agriculture Statistics Service. Most of that information comes through surveys where we randomly contact farmers and ask how much theyíve planted, how much theyíve harvested, what theyíre growing, where theyíre growing it, in addition to livestock and other types of questions. But the satellite remote sensing, from sensors like Landsat gives us a different view of whatís going on and one where we donít have to necessarily bother the farmers who, they are busy like we all are.

YOUNG: So, you put out a Cropland Data Layer, what is it? Is Landsat used to make it? Tell us a little bit about it.

JOHNSON: So the Cropland Data Layer is a, itís a, we call geospatial product. So, it is a map product that effectively at the end of the year is our snapshot of what was being grown in what field. So we are effectively determining with a fairly high degree of accuracy what fields were planted as corn, what fields were planted as soybeans, cotton, wheat, rice, you name it. It is a snapshot of what we saw over the summer, really throughout the whole country.

YOUNG: And Landsat?

JOHNSON: The Cropland Data Layer certainly leverages as much Landsat data as we can grab. We love Landsat, it has been around for years. We effectively download and integrate as much Landsat data over the summer. And more recently, have also been integrating a lot of data from the European system called Sentinel 2. That has really improved the revisit rates. So now we really have 2 or 3 times more data than we did even just 2-3 years ago.

YOUNG: How does a satellite 400 miles up in the sky look down and say, well, thatís a corn field and thatís a soybean field?

JOHNSON: So, the satellites are high up in the sky and thereís two ways we use them to determine or identify what kind of crop was being grown within a given field. The first way is they collect data not only in the visible part of the electromagnetic spectrum, that area where we use our eyes and we can see a green field versus a black road or red barn. The satellites also observe in the longer wavelengths, they tell us something about the chlorophyll content of those plants, the plant vigor, the plant health. They could also tell us a little bit about the temperature of those plants. And, different crops do respond differently. Soybeans is very very verdant, youíd say in the longer wavelengths, itís very bright in the near infrared. Whereas wheat would be less bright in the near infrared. So that's the first way we do that. So, the second way we do that is the crop timing tells us something. A crop like winter wheat planted say, down in Kansas, is planted in the fall, itís very lush and verdant in April and May. It is harvested in what most of us would consider the middle of the summer. So it greens up and it senesces. So we know that if we are observing the ground in August in Kansas, and we see something green on the ground effectively, it cannot be winter wheat. So, something like corn and soybeans, they are planted in similar times but corn tends to be planted earlier. So if we see a real green signature of reflectance from the satellite earlier in the season, it's more likely that is corn, letís say in South Dakota, than soybeans which tends to be later in the season. So, the combination of both what we see in terms of reflectance and the timing of that reflectance. And furthermore, a crop like alfalfa tends to be green all spring and summer and even into the fall although it is cut multiple times. So if we observe something that looks very green all year round that may be an alfalfa field or hay or thatís probably not a crop that you would expect to not be green. So, itís a combination of what we see in the electromagnetic frequency bands and also the timing of the crops and when the images are collected.

YOUNG: How accurate are you?

JOHNSON: The accuracy of the Cropland Data Layer varies by the commodity type and the area within the country. You know, using corn and soybeans as theyíre the major crops grown in the United States and over the corn belt, the Cropland Data Layer is very accurate for those. Probably at least 95% correct. So meaning if you look at the map and we have a field labeled as corn, there is a 95% chance or higher that that is indeed a corn field. And the same with soybeans. Now a commodity that is much smaller in nature, we have much less information about how it is grown, letís say onions for example, actually those are going to be much less and we are only 50% confident in identification of those fields. But it also depends on the geographic scope and where you are. But in the corn belt the ìIî states of Iowa, Illinois, Indiana, we feel like the remotely sensed information is giving us estimates or indications as we like to call them, as nearly as accurate as the survey information. Now, we donít rely on just one or the other. We want all of the information so that we can get the very best and certainly there will be areas where the surveys just outweigh, there is just so much information on a given year. Maybe we didnít get good satellite information because of clouds. But maybe there is an area where the survey information was thin or the conditions changed rapidly and satellite imagery like from Landsat was able to give us a better snapshot of whatís occurring in that area.

YOUNG: Would the only people interested in the Cropland Data Layer be those people who are planting and harvesting crops? Or, do groups outside of agriculture have interest in it?

JOHNSON: The Cropland Data Layer has a wide variety of users. And, certainly anyone who is interested in understanding what was being grown where over the last decade, we have that snapshot. On the ag side people might be interested in understanding crop rotation patterns. The cropland data layer can help you understand what areas maybe have gone out of production or have come into production. We have people using the Cropland Data Layer to help model where wildlife will be so maybe certain wildlife prefer certain crop type areas. People have also used it for range land style studies, to try to understand urban extent or encroachment or how the crop fields are expanding into grassland areas. Or if cropland areas are diminishing, being returned to a more native state. Letís say you are trying to site an ethanol plant and you are trying to understand where is the best place, you really care about where the corn fields are. So, with the Cropland Data Layer you can literally say I want to site my plant here and draw a boundary and calculate whatís the estimated corn area within that. If you are trying to understand watershed impacts and what the ability of nutrients and how that loading impacts the river systems, the Cropland Data Layer is much more effective than course county level statistics. It really allows you to isolate down to the watershed level, what is growing where. People might use the Cropland Data Layer for atmospheric style studies. Different crops transpire different amounts of water, so if someone is trying to understand what is going into the atmosphere, they might need a high level depiction of whatís on the ground versus whatís vegetative whatís not vegetative, whatís corn versus soybeans. I could go on down the list, itís a broad base and itís not just one sector. Itís the government types, the academic types, the commercial types, the educational groups or it could be a family farmer who's just curious whatís being grown where in their neighborhood and really donít have the ability to get out there quickly. 

YOUNG: Is there a certain website or something you go to see the Cropland Data Layer or to even see your NASS statistics?

JOHNSON:  If you Google NASS dot usda that will certainly take you to the agencyís website. In terms of the Cropland Data Layer, we have a website coined CropScape. If you google CropScape, that will highlight the website. Within CropScape, weíve set it up that not only can you visualize the data but you can actually do, some would say, basic analysis, some would say it is actually fairly sophisticated. So you can go in there and try to understand how the crops have changed through the years, you can do overlay analysis within the website, you can draw polygons, if you want to calculate what the area of corn is over a certain plot of land or maybe a watershed you have interest in, you can download the data through there if you have your own specialized software. Or, if you have a cartographic tool so if you want to print out what the crops are doing in your county, you can do that. We often times donít know what the uses are or what people want to show at the end of the day so weíve tried to make it flexible.

YOUNG: Weíve been talking to Dave Johnson about his work using remote sensing to estimate crop acres planted and yields across the United States. Itís been a fascinating conversation, Dave.

JOHNSON: Thank you.

YOUNG: We hope you come back for the next episode of Eyes on Earth. This podcast is a product of the US Geological Survey Department of the Interior. Thanks for joining us.

Show Transcript