Eyes on Earth Episode 12 – Plant Health via Satellite (NDVI)
A farmer at the foot of a corn stalk can tell how well the plant is faring. That same farmer might survey his entire field for crop health. But assessing the health of crops or forests at regional, national, and international scales requires remote sensing, most often via satellite. In this episode of Eyes on Earth, we talk to Jesslyn Brown about the Normalized Difference Vegetation Index (NDVI), a tool that uses the broader electromagnetic spectrum to estimate plant health.
John Hult: Hello, everyone. My name is John Hult, and Iím your host for todayís episode of Eyes on Earth, a podcast of the U.S. Geological Surveyís Earth Resources Observation and Science Center.
Or EROS. Which is an acronym. Anyone who spends any amount of time out here learns pretty quickly that we use a lot of acronyms.
For example, weíre the ground station for the Landsat series of satellites, and when we talk about them, we talk about their sensors. You know, the MSS, the TM, the TM plus, the OLI, and the TIRS.
Now Iím not going to explain what all those are. My point here is that we use acronyms to simplify communication about the tools, concepts, and programs that we use every day.
One of our goals here at Eyes on Earth is to explain a few of those terms, so that you too can partake of the alphabet soup of Earth science. Todayís flavor is NDVI ñ the Normalized Difference Vegetation Index.
I know, I know.
But NDVI is super interesting. Itís essentially a measure of plant health. How green are the crops? The forests? The grasslands? Satellites can track that every day, everywhere in the world. Drones equipped with near-infrared cameras can track it, too.
NDVI can help us answer all sorts of questions about how droughts, floods, wildfires or seasonal weather patterns affect our food, our forests, and our lives.
With us today to explain NDVI is Jesslyn Brown. Jess, welcome to Eyes on Earth.
Jess Brown: Hi, John. Good to be here.
JH: So what is NDVI, what is this concept? Where did it get the name? Who came up with this idea?
JB: So NDVI, or Normalized Difference Vegetation Index, is one of a family of indicators, also known as vegetation indices. These are calculated from data collected from satellite sensors as a way to simplify looking at the Earth.
JB: So early in the days of working with satellite data, there were some realizations in looking at data that were collected in different parts of the electromagnetic spectrum. The red and the near-infrared part of the spectrum have specific and sort of predictable responses to vegetation.
JH: Right, and when you talk about near infrared, youíre talking about part of the electromagnetic spectrum that we canít see.
JB: Beyond what we can see.
JH: Itís beyond what we can see
JB: Thatís correct. So we see visible light is in the blue, green and red. And just beyond red, a little bit longer wavelengths is the near infrared.
JH: These wavelengths are reflected, and we donít necessarily see them, but the near infrared is reflected by Ö vegetation, by plants.
JB: The healthier and denser and greener, honestly, the vegetation is, the higher that reflectance becomes. So dead vegetation reflects at a lower amount of reflectance than green.
JH: Right, so when weíre looking at this, weíre looking at the reflectance that we canít see, thatís beyond our ability to see, and we are coming up with this index. Itís essentially a math problem, isnít it? That gives you this index?
JB: So this is the really geeky part, or one of the really geeky parts. The red part of the spectrum is highly absorptive. Green vegetation absorbs red light, thatís why it doesnít look red, it looks green. When you get into the near infrared, you see this high spike in reflectivity. The vegetation index was a way to simplify the information from two bands and utilize them together.
JH: Two bands, the near infrared and the red.
JB: And the red, thatís right. So the first vegetation index was the simple ratio, which is just a difference of near IR and the IR.
JH: One minus the other?
JB: Right. But this needed to be normalized so that the measurements could be compared through time and across space. Normalization is accomplished through the sum of the red and the near IR. So really the math problem, as you call it, the algorithm, is the difference between the near IR and the red over the sum of the near IR and the red. NDVI. However, all of these are estimations of the Earthís surface. And vegetation indices in particular are not actually tied to any biophysical parameter per se.
JH: Biophysical parameter, you mean Ö
JB: I mean biomass or something called LAI, which is leaf area index. Vegetation cover density.
JB: Those are all things that can be estimated with a VI, but theyíre not absolutely measured.
JH: So the VI itself doesnít actually tell you how much biomass there is, how many leaves there are, but itís a way to Ö
JB: Itís a way to estimate it. And itsí a way to look at it through time.
JH: Each kind of plant, do they all have the same reading? Itís a number, right? Itís between negative one and one Ö do all plants have the same number if theyíre healthy?
JB: Oh, gosh no.
JH: Do only plants have an NDVI number? Does water have one, for example?
JB: Yeah, so NDVI over water tends to be very low, close to zero or negative. Snow also tends to be very low or negative. Bare soil, like a desert, or rock, tends to be very low. We refer to this in our field as a qualitative indicator rather than a quantitative indicator. So it has information. Low NDVI. We were talking about low NDVI. Low NDVI can be indicative of low vegetation or no very little vegetation, or senescent vegetation, and what I mean by that is, after the end of fall and after the leaves have fallen from the trees and in the grass, you will see a low NDVI compared to mid-summer when the vegetation canopy is fully green, healthy, that kind of thing. So thatís that ability to track phenology. So whatís kind of lovely and simple about the NDVI is that when it rises, vegetationís getting denser, this is looking at the same location, and when it falls, vegetationís either dying off or getting less dense. That makes a lot of sense. So I think early folks working in remote sensing were attracted to this very logical pattern of NDVI across time. It was easy to communicate, it was easy to understand.
JH: We have talked in the past Jess, about how NDVI is sort of like an ingredient. You donít just get an NDVI and thatís the end point. What do you do with NDVI beyond just estimating how healthy the plants are?
JB: All right, John. Good question. So NDVI is ingested into a variety of operational applications, and what I mean by that is, you know, providing information on a regular basis on a topical area thatís of interest to people. One of those areas is famine early warning systems. The NDVI is calculated over the whole continent of Africa, and actually a few different areas after that, and provided to the U.S. Agency for International Development, USAID, to help them track and even possibly predict famine events in a continent like Africa or countries like Ethiopia or the Sudan where peopleís lives are going to be in danger if they donít get access to food in a prompt way.
JH: So this reading helps guide food aid and really save lives. This is part of what theyíre looking at.
JB: Yeah. NDVI is just one ingredient of helping to figure out where to provide aid. But because we have satellite sensors we can provide this information over broad areas and then combine it, letís say, for example, a different indicator of rainfall or an indicator of evapotranspiration, you know how much moisture is being sucked out of the atmosphere if you will, and put all those pieces of information together and say ìyeah, this area is in trouble.î And they also need to look at things like economic indicators, as well. So NDVI feeds into something called FEW NET, the Famine Early Warning System Network, and when youíre interested in figuring out if thereís a famine or a drought going on, you really want to know what happened many years leading into this year. How does this year compare to what youíd expect in a normal year, for example. NDVI gives you a way, a building block for calculating a normal.
JB: What is normal for this time of year? What is normal for June 1? And what does it look like in comparison to normal? So again, where math is used to calculate a percentage or some sort of relationship to a normal NDVI signal, or a usual NDVI signal.
JH: Right. And those anomalies are part of the mixture, as you say, of information that goes into making a determination about what youíre going to do.
JB: Thatís right. Yeah, so some other uses for it: Itís been used in health and epidemiology. As you said drought detection, land degradation over time. All of this kind of involve this time component in the study. Deforestation and just plain change detection and monitoring, so has the area really changed?
JH: Right. And weíre actually recording in 2019, and there are plenty of things going on in farm country in 2019. We can actually see from NDVI, looking at last year to this year, those signatures look vastly different because itís been a very wet year, there have been all kinds of problems. Thatís reflected, right? You can actually go and see what the difference is?
JB: Well, thatís one of the cases where you would want ancillary or extra information to help you with your determination, because extra saturated soils, for example what we saw here in this region and just to the south of us this spring where we had massive flooding and delay to the start of season due to some really late, very weird and broad snowstorm and rain events, thatís going to depress the NDVI just like a drought is going to depress the NDVI. So you really need to bring in ancillary information to tell you what is the cause of the problem.
JH: Right, so you can see that the readings are not as good, you can tell that the plants are not as healthy as they were last year, but that alone does not tell you why Ö
JB: Thatís right.
JH: But it does give you and indication of maybe how broad the problem is.
JB: Thatís right. You can see a spatial representation of where that is.
JH: Now how do I see that, if Iím just a regular person, not a scientist. Do they talk about this on the farm report? Can I go and check this out somewhere? Is there a simple way to do this?
JB: So we donít just send it out to the world, NDVI goes out to the world, but there are some ways to get access to that information. You can download your components that you can make an NDVI from from EROS, from our archive. That would take some work to do, and youíd have to have some specialized software to do that. But there is a very nice tool, online tool, called AppEEARS, and that tool will allow you to access NDVI from MODIS for this country, for the U.S. that we have also done some filtering and smoothing on. We do that on a weekly basis. Thereís a little bit of lag time, like a day or two, but then we put that onto AppEEARS. A person can log into AppEEARS and bring up, for their area of interest, an NDVI time series.
JH: Essentially anybody can go to AppEEARS, and thereís a map, and you can draw a square around what you want to look at.
JB: Thatís right.
JH: Itís relatively simple.
JB: You can draw a square, you can extract the data, which is actually data values that you can plot through time using something like Excel.
JH: There are ways to get at this information if youíre curious.
JB: Itís becoming more broadly available, yeah.
JH: Weíve been talking to Jess Brown about NDVI. Itís been a fascinating conversation. We hope you come back for the next episode of Eyes on Earth. This podcast is a product of the U.S. Geological Survey, Department of Interior. Thank you for joining us.