Seeing Leaves in a New Light



  Page 2

As anyone who has tended a garden knows, even relatively minor changes in climate and temperature can have a pronounced effect on plant life. Less understood, perhaps, is that en masse plants can in turn have a sizable impact on the climate. Regionally, an increase in plant growth can cool surface temperatures and give rise to more rain and cloud cover. On a global scale, a rise in plant growth can lower the amount of atmospheric carbon dioxide, which is one of the most abundant of the greenhouse gases, and potentially cool the atmosphere.


Photographs of Leaves


For many years biologists and Earth scientists have known of these interactions, but they have never been able to assess to what degree plants influence climate. The problem has always been how to observe this influence. In order to understand the interaction of vegetation with the climate, researchers must have long-term, worldwide measurements of plant growth. So far scientists have met with limited success. While they have obtained qualitative records of relative plant density across the world using satellite imagery, they still cannot quantitatively assess the degree to which vegetation changes from one month to the next. For years, labs around the world have been working towards this goal.


Although it’s obvious that climate determines what types of plants live in a given area, plants themselves have an effect on climate. Lush vegetation absorbs sunlight, cools the Earth’s surface and increases humidity. Current climate models don’t always account for these effects, so scientists are developing new datasets based on satellite data to use in improved models. (Photographs courtesy Philip Greenspun)


Map of spring greening based on Leaf Area Index


Remote sensing specialists at the University of Boston and NASA may now be onto a solution. Using a measurement known as Leaf Area Index, they have found a way to quantify plant growth on a global scale with satellite imagery. Their method can pinpoint when leaves begin to grow in a region, when they fall off, and how dense they become at peak growing season. With the aid of instruments such as MODIS aboard NASA’s Terra satellite and the AVHRR instrument aboard NOAA’s polar satellites, the researchers may soon be able to forecast the ways in which plants impact our weather and global climate.

next How Plants Can Change Our Climate

The data used in this study are available in one or more of NASA's Earth Science Data Centers.

  The two images above show the changes in Leaf Area Index (LAI) that occured during the spring of 2002 in Switzerland and eastern France. White corresponds to an LAI of 0, while green, dark green, and blue represents increasingly dense vegetation. Topographic shading emphasizes the barren peaks of the Alps. (Data provided by Boston University Climate and Vegetation Research Group. Images by Robert Simmon)


How Plants Can Change Our Climate

  Page 1Page 3

“Plant growth can have a considerable effect on the climate,” says Wolfgang Buermann, a geographer at Boston University. He explains that there are several ways in which plants can alter the temperature of the Earth’s atmosphere. Through the process of photosynthesis, plants use energy from the sun to draw down carbon dioxide from the atmosphere and then use it to create the carbohydrates they need to grow. Since carbon dioxide is one of the most abundant greenhouse gases, the removal of the gas from the atmosphere may temper the warming of our planet as a whole.

Plants also cool the landscape directly through the process known as transpiration. When the surrounding atmosphere heats up, plants will often release excess water into the air from their leaves. By releasing evaporated water, plants cool themselves and the surrounding environment. “It’s like sweating. When you sweat you cool the surface of your skin,” says Buermann. Over a forest canopy or a vast expanse of grassland, large amounts of transpiration can markedly increase water vapor in the atmosphere, causing more precipitation and cloud cover in an area. The additional cloud cover often reinforces the cooling by blocking sunlight.

  Diagram of Plant Transpiration

Because of these processes, many researchers believe plants may have a sizable impact on global climate in the future. As humans continue to generate carbon dioxide and other greenhouse gases, the Earth’s surface will likely warm at a faster rate than it has in a thousand years. According to the Intergovernmental Panel on Climate Change (IPCC), the Earth is likely to warm another 1.4 degrees to 5.8 degrees by the end of this century (IPCC 2001). Needless to say, such big changes in the climate would likely alter vegetation growth all over the world. Many researchers hypothesize that the changes in vegetation could either serve to worsen or put a damper on global warming. If, for instance, the increased temperature and carbon dioxide levels of the Earth cause vegetation worldwide to flourish, plants could draw down more carbon dioxide and thus reduce the impact of the greenhouse effect. If, on the other hand, global warming causes widespread drought, then the loss of vegetation may result in even higher surface temperatures.

To model and then understand the ways in which vegetation interacts with the climate, scientists will need to maintain an accurate record of the Earth’s vegetation well into the future. For this purpose, for roughly the past twenty years, researchers have employed multi-spectral remote sensing satellite instruments such as the Advanced Very High Resolution Radiometer (AVHRR) instrument aboard NOAA’s polar-orbiting satellites. As is the case with most remote sensing satellite instruments, AVHRR houses a number of separate types of light detectors, which acquire images of different bands (colors) of light reflected off of or emitted from the Earth’s surface and atmosphere, including blue, green, red, near-infrared, and even thermal infrared energy. From these satellite data, scientists can produce images of the Earth showing a single band of light or a combination of bands. With a resolution on the order of 1 square kilometer per pixel and up, AVHRR images are not well suited for viewing details of the planet’s surface any smaller than a farm, but they are extremely useful for mapping and monitoring vegetation on a global scale.


As plants ‘breathe’ and ‘perspire’ they help cool the atmosphere. Plants consume carbon dioxide—a significant greenhouse gas—in the process of photosynthesis. The reduction of carbon dioxide in the atmosphere has an indirect cooling effect. Plants also cool the atmosphere because they release water vapor when they get hot, a process similar to sweating. The diagram at left shows the microscopic structure of a leaf, and the processes of photosynthesis and transpiration. (Illustration courtesy P.J. Sellers et al.)

To learn more about the role of plants in the hydrosphere, read The Water Cycle. To learn more about plants’ consumption of carbon dioxide, read The Carbon Cycle.

Graph of Chlorophyll Absorption

The amount and extent of vegetation, however, cannot be discerned from the raw satellite images alone. To extract information about vegetation from the satellite data, scientists must manipulate the images. The preferred method for years has been the normalized difference vegetation index (NDVI). Developed in 1979 by a NASA researcher, NDVI is a measure of the green, leafy vegetation density or the lushness of vegetation. [For more details see Measuring Vegetation (NDVI & EVI)] NDVI is produced by observing the discrepancy between the visible and near-infrared sunlight that reflects off of vegetation. As can be seen through a prism, many different wavelengths make up the spectrum of sunlight. The pigment in plant leaves, chlorophyll, strongly absorbs the visible light in the solar spectrum for use in photosynthesis. The cell structure of the leaves, on the other hand, strongly reflects near-infrared solar light. By measuring the difference between these two wavelengths of light in remote sensing data, scientists can get a relative measure of vegetation. If the difference is large, an area is likely to be densely vegetated, and if the value is small, the vegetation is likely to be sparse.

next A New Way to View Vegetation
back Seeing Leaves in a New Light

  The graph at left shows how efficiently the chlorophyll pigments in plants absorb light. The difference in absorption between visible and near-infrared light (longer than 0.7 µm) forms the basis for the measurement of Normalized Difference Vegetation Index. (Graph courtesy Compton Tucker, NASA GSFC)


A New Way to View Vegetation

  Page 2Page 4

“The problem is that NDVI is just a qualitative measurement of vegetation density, but it is not a quantitative measure of how much vegetation is present,” says Buermann. “Many scientists don’t trust NDVI measurements enough to use them in climate models.” In other words, NDVI can give researchers an idea of how lush the vegetation is in one area of the world relative to another, but it cannot tell them quantitatively how much vegetation there is in that one spot. Only by observing NDVI in an area over a long period of time and comparing it to other regions around the world, can researchers get a fix on what the normal vegetation density is for a region. Many in the science community feel that NDVI values are too inexact to use in climate models to determine, for instance, how much carbon plants draw down each year in a given area.

Buermann explains that there is a better measurement for plant density and growth known as leaf area index (LAI). LAI assigns a quantifiable value to the amount of vegetation on the ground. Simply put, LAI is the leaf area per unit ground area as seen when looking down on vegetation. One can imagine looking down on a tree canopy from a platform that stands high above the treetops. A tree canopy would have a leaf area index of one if every square inch of the ground below the tree canopy were overshadowed by exactly one leaf in the tree canopy. If exactly two leaves blocked the view of the ground then the tree canopy would have a leaf area index of two. Of course, most trees have layers and layers of leaves obstructing patches of the land unevenly. So a broadleaf deciduous forest (one that loses its leaves in the fall) typically will have a leaf area index of 3 or above in the summer, and evergreen conifer trees will have an LAI range of between 2 and 3.5 year round.

Horticulturists and biologists have used leaf area index to measure leaf density and vegetation health since the early part of the last century. Measurements were usually made on a local scale on the ground or from an airplane. For roughly the last decade, remote sensing scientists have attempted to measure LAI on a global scale using satellite data. The difficulty has always been one of accurately calculating the amount of leaves on millions of trees and shrubs from images taken from many miles above the ground. Previous attempts have yielded only very rough measurements of LAI directly from NDVI values.

Showing LAI versus NDVI

Buermann says he and his colleagues have uncovered a way to obtain more accurate values. The Boston team has developed a computer program that takes satellite data and other information gathered about the Earth’s surface and transforms them into values for leaf area index. “Essentially, what we try to do is simulate how the light is reflected off of the vegetation canopy,” says Buermann. Given ground cover and soil type, their computer simulation calculates what light from the sun would look like after it hits the vegetation and the ground and is reflected back up through the leaves, the atmosphere, and into space. Using this computer simulation, the scientists can compute LAI values by observing near-infrared and visible light from satellite data.

Together with researchers form the University of Arizona, Georgia Tech, and Ames Research Center, the scientists first employed their method on the AVHRR satellite data gathered from 1981 to 1990. They created global data sets/maps that displayed the average LAI values over the globe for each month over the ten-year period derived from the infrared and visible light readings from the data. “We then needed to show that LAI computed from satellite data are consistent with observed data,” says Buermann. Where they could, the researchers compared the satellite LAI values to existing LAI values obtained by ground-based measurements over the same period. Most of these records were of farmland in the Midwestern Plains States and of temperate and boreal forests of North America. The LAI values matched up well for both types of terrain (Buermann et al. 2001a).

  Leaf Area Index (LAI) is related to, but not directly proportional to, Normalized Difference Vegetation Index (NDVI). In addition, different vegetation types (broadleaf evergreens versus needleleaf evergreens, for example) and soil types exhibit different relationships between the two parameters. The graph at left compares NDVI to LAI. Unfortunately, the relationship between NDVI and LAI is not unique (multiple values of NDVI correspond to a single LAI value), a problem which is being addressed by the current generation of satellite instruments, such as the Moderate-Resolution Imaging Spectroradiometer (MODIS). (Graph courtesy Wolfgang Buermann, Boston University Climate and Vegetation Research Group.)

Area Index, March 1991

Unfortunately, these ground-based records weren’t very extensive. To test their LAI values thoroughly in this manner, the researchers would require LAI measurements for all types of terrain and all regions of the world. But conducting ground-based surveys of leaf area index worldwide would have been incredibly costly and time consuming. As an alternative, the team compared LAI values to known changes in plant growth around the world. If the LAI values were representative of vegetation density, then they should move in concert with the changes in seasons, geography, and rainfall patterns.

Buermann and his colleagues went about verifying, for instance, that LAI values of broadleaf evergreen trees in equatorial rainforests resulted in some of the highest LAI values and barren deserts resulted in some of the lowest values. They verified that seasonal changes in vegetation in the northern latitudes resulted in smooth seasonal changes in LAI. They even compared the LAI measurements to El Niño events in the 1980s. As expected, in areas that experience greater rainfall during El Niño, such as the west coast of Central America and South America, LAI values were higher. In areas that experience less rainfall, such as Australia, the LAI values were lower (Buermann et al. 2001a). “We had good agreement in semi-arid, tropical, and subtropical areas where change in vegetation and precipitation run together,” says Buermann.

next Modeling the Earth’s Vegetation
back How Plants Can Change Our Climate

  Scientists used the long record of NDVI data acquired by NOAA’s polar-orbiting weather satellites to create a long-term LAI dataset. The image at left shows LAI values from March 1991. An animation shows average monthly LAI from July 1981 to June 1991. The MODIS instruments aboard the Terra and Aqua satellites will extend this dataset with more accurate LAI measurements. (Data provided by Boston University Climate and Vegetation Research Group. Image and animation by Robert Simmon)


A New Way to View Vegetation

  Page 2

LAI palette

Scientists used the long record of Normalized Difference Vegetation Index data acquired by NOAA’s polar-orbiting weather satellites to create a long-term Leaf Area Index (LAI) dataset. The image at left shows LAI values from March 1991. An animation shows average monthly LAI from July 1981 to June 1991. The Moderate-Resolution Imaging Spectroradiometer (MODIS) instruments aboard the Terra and Aqua satellites will extend this dataset with more accurate LAI measurements. (Data provided by Boston University Climate and Vegetation Research Group. Image and animation by Robert Simmon)

large animation (640 by 480 pixels, 30 frames per second, 14.4 MB Quicktime)

back A New Way to View Vegetation



Modeling the Earth’s Vegetation

  Page 3

With partially validated LAI values in hand, the researchers then wanted to know if their values would work in a climate model. More specifically, the scientists wanted to incorporate these data into a computer model of the climate to observe how changes in plant growth from year to year influence near-surface temperatures. To get the clearest results, the scientists created LAI global data sets showing minimum and maximum values over a 12-month period. To form their minimum LAI data set, they examined their ten years of data and found all the lowest LAI values in every region of the world for each month in the calendar year. They then put these values together into one twelve-month data set. By doing so, they essentially lumped all the droughts and abnormally dry spells around the world over a ten-year period into one year-long set of maps. They created the maximum LAI data set in much the same way, except all the abnormally wet spells were lumped into one data set (Buermann et al. 2001b).


Comparison of Maximum and Minimum Leaf Area

The researchers ran each data set through a computer simulation of the climate that could compute the effects of plant growth on surface temperatures, cloud cover, and rainfall. The model runs produced results that were much more in line with field observations than those obtained from non-satellite-derived LAI values. “We discovered that the [increase in vegetation] between the wet and dry simulation caused over a 1°C difference in the near-surface temperatures in many areas,” says Buermann. Generally, more transpiration occurred in the wet scenario. The transpiration cooled the surface both directly by the release of water to the atmosphere and indirectly by giving rise to additional cloud cover, which effectively blocks incoming solar radiation (Buermann et al. 2001b).

Buermann’s team is now using their process to investigate whether global warming over the past two decades has affected plant growth in the Northern Hemisphere. A study released last year by scientists at Boston University and NASA’s Goddard Space Flight Center suggests that yearly plant growth has increased in northern latitudes over the last twenty years due to an elongated growing season, possibly brought on by global warming (Zhou et al. 2001). Since the team used NDVI values to measure plant growth, there is still much speculation over whether the increase is due to a longer growing season, the presence of more carbon dioxide in the air, or simply due to more rainfall. Buermann and his colleagues plan on retrieving LAI measurements of the region using the same satellite data. They then hope to determine what is causing the increase in plant growth. “With LAI, we’d be able to measure precisely when the leaves start to grow, the density of the leaves during the growing season, and when they fall off,” says Buermann. “We can even tell how much carbon the plants draw down each year.”

In the future, the team will continue to take measurements of the LAI values of the Earth’s vegetation using the MODIS instrument aboard NASA’s Terra and Aqua satellites. The MODIS instrument is a lot like the AVHRR in that it follows a near-polar orbit and gathers global-scale data. MODIS’ measurements, however, are taken at a more moderate resolution (500 meters per pixel) over the entire Earth’s surface almost every day. The MODIS instrument also has many more light sensors than AVHRR. MODIS will allow the Boston team to produce more accurate LAI readings of the entire Earth every eight days. With such data, they will continually be able to monitor the Earth’s vegetation and work towards a full understanding of the role vegetation plays in climate change.

1. Intergovernmental Panel on Climate Change, 2001: Summary for Policymakers, A Report of Working Group 1 of the Intergovernmental Panel on Climate Change, Cambridge, UK, 2-17.

2. Buermann, W., Y. Wang, J. Doug, L. Zhou, X. Zeng, R. E. Dickinson, C. S. Potter, and R. B. Myneni, 2001a: Analysis of multi-year global vegetation leaf area index data set, J. of Geophys. Res., (submitted).

3. Buermann, W., J. Dong, X. Zeng, R. B. Myneni, and R. E. Dickinson, 2001b: Evaluation of the utility of satellite based vegetation leaf area index data for climate simulations, J. of Climate, 14, 3536-3550.

4. Zhou, L., C. J. Tucker, R. K. Kaufman, D. Slayback, N. V. Shabanov, and R. B. Myneni, 2001: J. of Geophys. Res., 106, 20,069-20,083.

back A New Way to View Vegetation


In addition to evaluating Leaf Area Index (LAI) on a month-by-month basis, the Boston University researchers also investigated the total variability of LAI from 1981 to 1991. The scientists combined the maximum and minimum LAI values into a pair of datasets (shown at left). These data were then used in climate models. The models showed a decrease in surface temperature and increase in rainfall (primarily due to increases in transpiration) for regions with maximum LAI, a result which agrees with real-world observations. (Data provided by Boston University Climate and Vegetation Research Group. Images by Robert Simmon)

To learn more about modeling the interactions between climate and biosphere, read Modeling Earth’s Land Biosphere.