|
||||||||||||||||
|
||||||||||||||||
The following HTML text is provided to enhance online readability. Many aspects of typography translate only awkwardly to HTML. Please use the page image as the authoritative form to ensure accuracy. 9
|
||||||||||||||||
 |
 |
Page 73 |
 |
 |
 |
|
| ||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 73
Earth Observations from Space: The First 50 Years of Scientific Achievements
9
Ecosystems and the Carbon Cycle
Research on the biosphere aims to understand and predict how terrestrial and marine ecosystems are changing, how they are affected by human activity or through their own intrinsic biological dynamics, how they respond to climate variations, and in turn how they affect climate. One of the primary goals of ecosystem research is to determine the amount of primary production, which is most commonly expressed in units of carbon incorporated during photosynthesis and estimates the amount of energy available for higher trophic levels. Since the discovery of the importance of carbon dioxide as a greenhouse gas, the estimation of global carbon fixed by photosynthetic processes has become a central quest in global carbon cycle research and an integral part of climate models.
Before the satellite era, few scientists had attempted to estimate these parameters at a global scale. Instead, most research efforts were dedicated to understanding local dynamics because ecosystem processes are highly variable in response to localized environmental changes. Orbiting satellites provide an ideal vantage point for viewing dynamic ecosystems on the land and in the ocean (Box 9.1). This chapter discusses how the remarkable technological advances of the past decades have enabled scientists to compose routinely global maps of terrestrial and marine productivity, assess the role of the ocean in the global carbon cycle, observe long-term ecosystem trends and atmosphere-biosphere coupling, and even study plant physiology from space.
For the first time, remote sensing made direct global observations of photosynthesis, plant growth, and ecosystem phenology possible, leading to the evolution of a global perspective on ecology (Boxes 9.2 and 9.3). Charles Keeling’s continuous measurements of atmospheric carbon dioxide (CO2) concentrations at Mauna Loa, beginning in 1957, showed a seasonal signal in the atmospheric CO2 concentration due to the terrestrial biosphere being a source and sink for carbon during the winter and summer, respectively. Subsequent work showed that atmospheric CO2 was steadily increasing (Keeling et al. 1976) and that it stemmed from fossil fuel burning, catalyzing an interest in obtaining a global perspective of the carbon cycle. In 1982, the National Aeronautics and Space Administration (NASA) held a workshop in Woods Hole, Massachusetts, on global change (Goody 1982) that spurred a subsequent paper by Tilford (1984) presenting the scientific rationale for the Earth Observing System (EOS). These papers called attention to how anthropogenic global changes might impact ecosystems.
TERRESTRIAL PRIMARY PRODUCTIVITY
New awareness of the relationship between microclimate and plant functions in the 1970s and 1980s spurred the development and evolution of field-portable instruments to measure plant physiological processes, photosynthesis, and transpiration, moving these measurements from the laboratory to the field. Despite these newly available field instruments, global observations of ecosystem and larger-scale processes did not become available until the advent of satellite observations because the field measurements were generally restricted to short-term (seconds to minutes) leaf measurements. The capability of assessing plant productivity from satellite radiance measurements (Box 9.2) opened an entirely new front in ecosystem research. Because small-scale point measurements did not lend themselves well to interpolating and creating global maps, synoptic satellite data provided the first direct globally distributed measurements of terrestrial functioning.
The NASA EOS program brought new capabilities for monitoring terrestrial productivity, with near-daily global coverage of a more capable well-calibrated Moderate Resolution Imaging Spectroradiometer (MODIS) that has allowed development of new biophysical measurements with less reliance on simple empirical indices. One of the new products is the direct global measurement at 1-km resolution of leaf area index (LAI)—an important structural property of
OCR for page 74
Earth Observations from Space: The First 50 Years of Scientific Achievements
BOX 9.1
Ecosystems as Seen from Space
Satellite-based studies of the land and ocean ecosystems rely primarily on imaging sensors measuring radiance in the visible and near infrared. These spectral bands were ideally suited to monitor plant biomass and primary production because the chlorophyll a pigment, found in all marine and terrestrial photosynthetic plants, reflects green light while absorbing in the blue and red spectral regions. Because plant leaves contain no molecules with high absorption in the near infrared, they are highly reflective in this region. Therefore, the “greenness” of terrestrial ecosystems can be mapped by employing the ratio of red to infrared bands. However, this ratio does not work for the ocean because water is such a strong absorber in the red and infrared that little or no radiation is reflected out of the ocean at those wavelengths. Instead, the ratio of blue to green bands, after correcting for the atmosphere, has been used to quantify the chlorophyll concentration in the ocean (Box 9.3).
Remote sensing techniques for mapping and studying terrestrial and marine ecosystems have evolved along different paths because of different technological requirements. Compared to the ocean, the land is a bright surface whose features have distinct spectral signatures and generally sharp boundaries. The spatial scale of such features is on the order of tens of meters, thus requiring high spatial resolution, but the features generally change slowly over seasons or longer. In contrast, the ocean is a dark surface with subtle spectral variation that requires high radiometric sensitivity. Reflectance from the atmosphere dominates the signal received by a satellite over the ocean, and this signal must be estimated and removed before the ocean signal can be analyzed. Features in the ocean have spatial scales on the order of tens of kilometers, with fluid boundaries that change on timescales of hours to days. These differences have led to different sensor and mission requirements, but the goals remain similar. Both terrestrial and marine studies have sought to quantify primary productivity and the role of the biosphere in the global carbon cycle.
BOX 9.2
Converting Radiance to Plant Productivity
Jordan (1969) was the first to use a ratio of near-infrared and red radiation to estimate biomass and leaf area index (leaf area/ground surface area) in a forest understory. This study was quickly followed by application of near-infrared/ red ratios to estimate biomass in rangelands (e.g., Pearson and Miller 1972; Rouse et al. 1973, 1974; Maxwell 1976) and was extended by Carneggie et al. (1974) to the Earth Resources Technology Satellite (ERTS-1) observations of seasonal growth, which showed that the seasonal peak in the near-infrared/red ratio coincided with maximum foliage production, thus effectively tracking the phenological cycle.
Rouse et al. (1974) introduced a spectral index, a normalized ratio that reduced illumination differences and other extrinsic effects by dividing the difference of the two bands by their sum, a ratio adopted as the normalized difference vegetation index (NDVI). A landmark paper by Tucker (1979) established linear relationships between vegetation spectral indices (ratios of visible and near-infrared bands) to leaf area and biomass. Following this paper, vegetation indices rapidly became an established method for analysis of plant biophysical properties using laboratory, field, airborne, and Landsat data. Today, nearly 2,000 papers have been published using the NDVI, and nearly 6,000 have used some type of vegetation index to study vegetation. These early studies established that red and near-infrared satellite bands could track changes in plant growth and development.
the plant canopy used to estimate functional process rates of energy and mass exchange, specifically to calculate rates of photosynthesis, evapotranspiration, and respiration (Figure 9.1). For the first time this measurement provides a consistent observational basis to estimate and monitor global productivity. Time series of LAI allow comparison of phenological patterns among six global terrestrial biome types. LAI is defined as the one-sided leaf area per unit of ground area and is produced by R.B. Myneni, Boston University. An algorithm is used to convert red and near-infrared band reflectances to global maps of LAI with modifications for the six biome types, taking into account the directional Sun and
OCR for page 75
Earth Observations from Space: The First 50 Years of Scientific Achievements
FIGURE 9.1 February (top panel) and August (bottom panel) 2006 global 4-km monthly composites of leaf area index, computed from the Moderate Resolution Imaging Spectroradiometer (MODIS; Mod15, collection 4). SOURCE: R.B. Myneni, Boston University, http://diveg.bu.edu.
view factors and measurement uncertainties. Prior to today’s satellites, this key biophysical variable was painstakingly evaluated at the scale of small field sites by dropping a pin or line through the canopy and counting the number of leaves that were contacted. With the development of red and near-infrared indices such as normalized difference vegetation index (NDVI) in the 1980s, it became possible to correlate these ground measurements with index values, allowing the extension of direct measurements to larger regions.
Today, with MODIS, this observation has become more precise by its extension to a biophysical measurement. Satellite monitoring of the dynamics of Earth’s vegetation is essential to understanding global ecosystem functioning and response to climate variability and climate change. This new observational perspective has led ecologists to see ecosystem processes in an integrated temporal and global context.
MARINE PRIMARY PRODUCTIVITY
Approximately half of all global primary production occurs in the ocean, almost entirely due to microscopic single-cell algae known as phytoplankton. In the presence of ample sunlight and nutrients, phytoplankton reproduce rapidly and biomass can double in a day. As the cells grow and reproduce, carbon dioxide dissolved in the surface ocean is converted to organic matter, which is then consumed by
OCR for page 76
Earth Observations from Space: The First 50 Years of Scientific Achievements
BOX 9.3
Global Marine Biomass from Ocean Color Remote Sensing
The ability to derive global maps of chlorophyll a concentration (milligrams per cubic meter) in the upper ocean from ocean color sensors was a groundbreaking achievement for the oceanographic community (Figure 9.2). This biomass estimate can then be related to primary productivity and the marine carbon cycle. Although clouds prevent ocean color sensors to see the entire ocean surface on each orbital pass, a global picture of the distribution of photosynthetic plant biomass emerges from averaging data over several consecutive days or weeks.
The first ocean color sensor was the Coastal Zone Color Scanner (CZCS), an experimental proof-of-concept mission operating on the Nimbus 7 satellite between 1978 and 1986. The CZCS demonstrated that it is possible to detect subtle changes in the color of the ocean and relate these to the concentration of chlorophyll a, the light-harvesting pigment found in all plants. In particular, chlorophyll a concentrations are quantified by empirical algorithms relating spectral band ratios (blue to green) to the concentration of chlorophyll in the ocean (Clark 1981, Gordon and Morel 1983, O’Reilly et al. 2000). A major requirement is that the spectral radiance measurements made by the satellite be corrected to remove the effect of the atmosphere, which comprises more than 90 percent of the top-of-atmosphere signal. This was a major technological breakthrough after the launch of the CZCS (Gordon et al. 1980). Contrary to its name, the sensor was better at estimating biomass in the open ocean than in the coastal zone. Phytoplankton and dissolved organic matter are the primary sources of optical variability in the open ocean (so-called Case 1 waters [Morel and Prieur 1977, Gordon and Morel 1983, Siegel et al. 2002]), whereas in coastal regions, mixtures of organic and inorganic materials affect the ocean color. The problem of differentiating and quantifying individual constituent concentrations in the coastal ocean remains a challenge today.
The ocean color technology pioneered by the CZCS has since been improved and incorporated into modern space instruments. The first modern global ocean color sensor was Japan’s Ocean Color and Temperature Sensor (OCTS) launched in August 1996 aboard the Advanced Earth Observing Satellite (ADEOS). The U.S. Sea-Viewing Wide Field-of-View Sensor (SeaWiFS) followed in August 1997, shortly after the ADEOS experienced structural damage after only 9 months in orbit. SeaWiFS is owned by Orbital Sciences Corporation, with a guarantee from NASA to buy data for the scientific research community. Ocean color data continue to be acquired by the Moderate Resolution Imaging Spectroradiometers (MODIS) aboard the Terra and Aqua satellites launched in 1999 and 2002, respectively, and by a number of other ocean color instruments operated by other countries (Table 9.1).
FIGURE 9.2 Map of chlorophyll a concentration (milligrams per cubic meter) in the upper Atlantic Ocean derived from data obtained by the Sea-viewing Wide Field-of-view Sensor (SeaWiFS). SOURCE: SeaWiFS Project, NASA Goddard Space Flight Center, and GeoEye.
OCR for page 77
Earth Observations from Space: The First 50 Years of Scientific Achievements
TABLE 9.1 Past and Present Satellite Sensors with Ocean Color Capability
Sensor
Satellite
Country
Dates
Spatial Resolution (m)
No. of Bands
Comments
CZCS
Nimbus-7
United States
November 1978-July 1986
825
5
Proof-of-concept
MOS
IRS P3
Germany-India
March 1996-March 2004
523
13
Requires ground station
OCTS
ADEOS-1
Japan
August 1996-June 1997
700
8
+ Four thermal IR bands for SST
SeaWiFS
SeaStar
United States
August 1997-present
1,100
8
Commercial data, free to researchers
OCI
ROcSAT-1
Taiwan
December 1998-July 2004
800
6
Latitude coverage 35° N-35° S
OCM
OceanSat-1 (IRS P4)
India
May 1999-present
360
8
+ Scanning microwave-SST
MODIS
Terra Aqua
United States
December 1999-May 2002
1,000
9
+ 27 other bands for land, atmosphere, SST
MERIS
Envisat
EU
March 2002-present
250 LAC
1,000 GAC
15
LAC data require ground station
NOTE: Nations such as Japan, Taiwan, and India have invested in ocean color as a valuable source of information for their fishing fleets (Laurs et al. 1984, Butler et al. 1988). This has also met with some success in the United States where it has been argued that the satellite information makes fishing more efficient, thus saving fuel and other limiting resources.
CZCS = Coastal Zone Color Scanner; GAC = Global Area Coverage; LAC = Local Area Coverage; MERIS = Medium Resolution Imaging Spectrometer; MODIS = Moderate Resolution Imaging Spectroradiometer; MOS = Maritime Observation Satellite; OCI = Ocean Color Imager; OCTS = Ocean Color and Temperature Sensor; SeaWiFS = Sea-Viewing Wide Field-of-View Sensor; IRS = Indian Remote Sensing Satellite; ADEOS = Advanced Earth Observing Satellite; SST = Sea Surface Temperature.
FIGURE 9.3 Global annual NPP (in grams of carbon per square meter per year) for the biosphere, calculated from the integrated CASA-VGPM (Vertically Generalized Production Model) model. Input data for ocean color from CZCS sensor are averages from 1978 to 1983. The land vegetation index from the AVHRR sensors is the average from 1982 to 1990. SOURCE: Field et al. (1998). Reprinted with permission from AAAS, copyright 1998.
OCR for page 78
Earth Observations from Space: The First 50 Years of Scientific Achievements
zooplankton, fish, and other animals in the “food chain.” Because of its rapid growth and many consumers, phytoplankton biomass or chlorophyll concentration varies on short timescales, yet the extent of a “patch” of accumulated biomass is on the order of 10-100 km.
Satellites have allowed scientists to routinely estimate phytoplankton productivity on an annual basis for the first time, enabling them to detect a trend in decreasing phytoplankton productivity associated with warming of the surface ocean at mid- to low latitudes. Because a phytoplankton bloom and its associated productivity are such large-scale yet short-lived phenomena, there is simply no way to survey large enough areas of the ocean to capture their dynamics using ships to map phytoplankton biomass and productivity.
Prior to the introduction of satellite observations, estimates of oceanic primary production depended on relatively few labor-intensive ship-based incubations using the 14C technique that had become the standard method for measuring primary productivity in the ocean (Steeman-Nielsen and Jensen 1957). To estimate global annual oceanic production (gigatons of carbon per year), the mean integral productivity was first estimated for the different oceans and depth ranges using relatively few measurements made in each domain. These were then multiplied by the area of the ocean domain and 365 days per year to derive annual oceanic primary production. Due to the vastness of the ocean and high spatial and temporal variability, ship-based global mapping was infrequently attempted and could not realistically capture the interannual variability. Even with the development of fluorescence-based estimates of marine primary productivity, which could be obtained from instruments towed behind ships, obtaining global coverage would still require years. It has long been recognized that ship-based sampling methods suffer from significant undersampling in both space and time (McCarthy 1999). Consequently, the best quantitative global estimates of both biomass and productivity are derived with the use of satellite observations that provide the necessary frequency of global coverage.
To estimate primary productivity from satellite measurements, it is assumed that the productivity is proportional to the phytoplankton biomass. Consequently, measuring biomass is the first critical step in estimating marine primary productivity from space. Chlorophyll a, the ubiquitous light-harvesting pigment found in all green plants, has long been a standard measure of phytoplankton biomass (Box 9.3, Figure 9.2, Table 9.1). This is largely because chlorophyll can be measured rapidly and easily owing to its fluorescent and absorption properties.
Early estimates of oceanic primary productivity derived using satellite data provided a relative static picture in that they represented average annual productivity (Platt and Sathyendranath 1988, Antoine et al. 1996, Behrenfeld and Falkowski 1997, Field et al. 1998). One of the most thorough estimates was that of Longhurst el al. (1995), who estimated global ocean net primary production using Coastal Zone Color Scanner (CZCS) data and models of the subsurface chlorophyll distribution and Photosynthesis-irradiance (P-I) relationships defined for 57 biogeochemical provinces.
GLOBAL MARINE AND TERRESTRIAL PRIMARY PRODUCTION
Net primary productivity (NPP) is influenced by climate and biotic controls that interact with each other. Field et al. (1995) predicted global terrestrial NPP on a monthly time step using the Carnegie-Ames-Stanford Assimilation (CASA) model, incorporating a set of ecological principles and satellite and surface data. Several authors have used satellite data to estimate global net primary production, combining both terrestrial and oceanic models. Within a few years they used a linked ocean-terrestrial model that combined an 8-year Advanced Very High Resolution Radiometer (AVHRR) record and a 6-year CZCS data record with a biogeochemistry model to estimate global land and ocean NPP (Field et al. 1998, Figure 9.3). This study found that the contribution of land and ocean to NPP was nearly equal but that there was striking variability in NPP at a local level. Based on the spatial variability in the satellite data, their model predicted strong differential resource limitations for terrestrial and ocean habitats.
Behrenfeld et al. (2001) used the Sea-Viewing Wide Field-of-view Sensor (SeaWiFS) data to estimate terrestrial and ocean primary production during the transition between El Niño and La Niña conditions in 1997 to 1999. They found that the ocean exhibited the greatest effect, particularly in tropical regions where El Niño-Southern Oscillation (ENSO) impacts on upwelling and nutrient availability were greatest. Terrestrial ecosystems did not exhibit a clear ENSO response, although regional changes were substantial. These studies clearly demonstrate the invaluable contribution satellite observation of NPP make to the fundamental understanding of climate change impacts on the biosphere.
THE OCEAN CARBON CYCLE
Satellite observations afford the only means of estimating and monitoring the role of ocean biomass as a sink for carbon. In particular, the fundamental question of whether the biological carbon uptake is changing in response to climate change can only be addressed with satellite measurements. It requires not only ocean color measurements (phytoplankton biomass and productivity) but also coincident space-based observations of the physical ocean environment (circulation and mixing) and land-ocean exchanges through rivers and tidal wetlands, as well as winds, tides, and solar energy input to the upper ocean. Observing linkages between the physical and chemical environment and the biology of the ocean is a significant achievement of observations from space. Continuity of this record is critical. Understanding the consequences of the CO2 increase and its effect on terrestrial and marine
OCR for page 79
Earth Observations from Space: The First 50 Years of Scientific Achievements
ecosystems will require global-scale long-term observations from carefully calibrated satelliteborne sensors.
Early carbon cycle models that were used to investigate sources and sinks of anthropogenic CO2 ignored the effects of marine productivity, which was thought to be in equilibrium on annual timescales. Since marine productivity is not limited by carbon, it was reasoned that increases in CO2 would not affect oceanic productivity. More recently, modelers have investigated how marine productivity might be affected indirectly by climate change through its effect on oceanic and atmospheric circulation patterns.
Because phytoplankton life cycles are orders of magnitude shorter (days versus years or decades) than those of terrestrial plants, phytoplankton may respond to climate influences on ocean circulation, mixing, and the supply of nutrients and light much more quickly than plants in terrestrial ecosystems. Given that oceanic primary productivity is estimated to be roughly half of all global primary productivity, the oceanic component of the carbon cycle will respond more quickly to climate changes.
For example, there are vast areas of the Pacific and Southern Oceans, where phytoplankton productivity might be limited by iron (Martin et al. 1994). In contrast to the other limiting nutrients, which are supplied primarily by the deep ocean, atmospheric dust deposition is one of the main sources of iron to the open ocean. Paleorecords indicate that the Southern Ocean responded with increased productivity during colder periods when iron atmospheric deposition was enhanced due to the expansion of arid regions. This led to the notion that these areas in the Pacific and Southern Oceans could be stimulated to draw down large amounts of atmospheric CO2 if they were provided with iron. Several experiments conducted in the late 1990s and early 2000s proved conclusively that iron does limit production in these regions (Coale et al. 2004). Iron is supplied to the open ocean by atmospheric transport (dust deposition), by lateral advection of waters from the continental margins, and by upwelling of deep iron-rich waters. Long-term monitoring of the ocean phytoplankton will reveal whether climate change will affect these iron supplies potentially fertilizing the Southern Ocean or the Pacific.
With 10 years of continuous ocean color data (since 1997), we now have the ability to observe year-to-year variability in global oceanic primary production and begin to assess longer-term trends in ocean carbon uptake. Behrenfeld et al. (2006) describe a steady climate-driven decrease in oceanic NPP related to the warming of permanently stratified ocean waters at mid- to low latitudes over the past 8 years. This period of decreasing NPP followed the rise in NPP between the El Niño and La Niña phases. Satellite observations afford the only means of estimating and monitoring the role of the ocean biomass as a sink for carbon.
LONG-TERM ECOSYSTEM RECORD REVEALS ATMOSPHERE-BIOSPHERE COUPLING
Although early studies established that red and near-infrared satellite bands could track changes in plant growth and development (Box 9.1), the large number of Landsat images (~5,000) required to assemble a global database, combined with computational requirements and frequent cloud cover, have prevented analysis of complete global or time series of Landsat data sets. Launched in 1978, the Coastal Zone Color Scanner showed that ocean productivity could be observed using visible and near-infrared bands; however, CZCS measurements were saturated over land and thus unusable.
The Advanced Very High Resolution Radiometer on the National Oceanic and Atmospheric Administration’s (NOAA) polar-orbiting weather satellites has obtained a continuous record of daily global observations since 1978, acquiring both red and near-infrared bands. Because AVHRR was not designed for observing the terrestrial biosphere and the 1- to 8-km scale of AVHRR pixels was significantly larger than theoretical understanding of ecosystem processes, scientists were initially skeptical about whether biospheric patterns and trends could be observed. However, scientists have managed to overcome technical problems such as maintaining calibrations, screening clouds, and adjusting for different observational angles. Thanks to the pioneering efforts of Compton Tucker, the daily AVHRR data set now spans more than 25 years and is the longest continuous global record available of terrestrial productivity, phenology, and ecosystem change for monitoring biospheric responses to climate change and variability. Although AVHRR was not designed for climate monitoring, continuing improvements in calibration and reanalysis have produced a consistent record for monitoring and assessing past and future biospheric responses resulting from climate change and variability and anthropogenic activities.
Initial studies using AVHRR followed seasonal and annual trends in ecosystem production and vegetation phenology at regional and continental scales (Tucker et al. 1985, Townshend et al. 1985) and at the global scale (Justice et al. 1985). In the early 1990s some key papers introduced the use of remote sensing data to ecology (Roughgarden et al. 1991, Ustin et al. 1991) and stressed the need for ecologists to focus on global ecological problems (Mooney 1991). These ideas led to the resurgence in ecosystem research and modeling of biogeochemical processes and significant advances in understanding the Earth as a system.
By the mid-1990s, global ecosystem and biogeochemical models used satellite data to establish variable vegetation composition and abundance (e.g., Biome BioGeochemical Cycles [BGC], Running and Hunt 1993; CASA, Potter et al. 1993). The concept of resource limitations as the controlling mechanism determining NPP was established in the late 1980s (Chapin et al. 1987). This placed a premium on direct
OCR for page 80
Earth Observations from Space: The First 50 Years of Scientific Achievements
satellite observations of vegetation conditions to provide more realistic estimates of NPP. Previous estimates used uniform rates of NPP for each land-cover type and assumed that NPP is proportional to reflected net shortwave radiation.
The relationship between vegetation indices and the physiological processes of photosynthesis and absorbed photosynthetic radiation (APAR) were formalized in theoretical analyses by Piers Sellers (1986). These developments led to a seminal paper by Tucker et al. (1986) in which it was shown that changes in the planetary NDVI (greenness) were strongly correlated with daily dynamics of terrestrial IPAR (intercepted photosynthetically active radiation) and atmospheric CO2 concentrations. There is a strong negative correlation between NDVI and atmospheric CO2 such that NDVI is high when CO2 concentrations are low and low when CO2 concentrations are high (Figure 9.4). This temporal pattern in ecosystem photosynthesis and respiration demonstrates the dynamic coupling between the biosphere and the atmosphere.
In the past decade, NDVI data from AVHRR have become a critical component in monitoring climate change (Fung et al. 1987, Sellers et al. 1994, Angert et al. 2005), assessing changing length and timing of the growing season (e.g., Justice et al. 1985, Myneni et al. 1997, 1998; Box 9.4, and Figure 9.5), and monitoring the state of the biosphere (Anyamba et al. 2001) and other ecosystem phenomena. Long-term records of NDVI have revealed its increase in response to a warming climate during the 1980s and early 1990s, but this trend has leveled off most recently (Angert et al. 2005).
STUDYING PLANT PHYSIOLOGY FROM SPACE
To estimate actual NPP in the presence of environmental stressors, researchers developed methods to remotely estimate regulatory plant biochemicals. The first advance was the development of the “photochemical reflectance index” (PRI) by John Gamon and colleagues (Gamon et al. 1992) to better predict radiation use efficiency. This index has had extensive use for noninvasive studies of leaf photosynthesis by plant physiologists, although at the image level it appears more related to carotenoid content. The PRI has led to a range of other studies to quantify plant pigments and develop methods for assessing them. These advances follow increasingly specific knowledge of spectroscopy of plant properties and how this information can be retrieved from satellite sensors.
A radiative transfer model, developed by Jacquemoud
FIGURE 9.4 Weighted NDVI data plotted against time and latitude zone. Note the highly seasonal effects in the northern latitudes, the influence of deserts in the 20°-30° N latitude zone, the generally constant response in equatorial areas, and the influence of the low proportion of land area south of 30° S. SOURCE: Reprinted with permission from J.E. Pinzon (SSAI-NASA/GSFC) and C.J. Tucker (NASA/GSFC).
OCR for page 81
Earth Observations from Space: The First 50 Years of Scientific Achievements
BOX 9.4
Increasing Growing Season
Myneni et al. (1997) published a groundbreaking paper using daily satellite data over a 9-year period to show increases in the length of the growing season in the boreal region. They used a time series of NDVI, a measure of the photosynthetic activity of vegetation canopies, derived from the daily AVHRR satellite data, and showed an increase in length of the growing season in the boreal region (north of 45°) of 12 days (8 days in spring and 4 days in autumn) from 1981 to 1991. They demonstrated that this extension of the growing season and enhanced amplitude of NDVI over the summer were likely correlated with warmer spring and autumn temperatures over the region. This result partially corroborated an estimated 7-day extension of the growing season that was inferred from atmospheric CO2 measurements. Uniquely, their analysis detected significant spatial variation in the distribution of enhanced NDVI, with western and eastern Canada and southern and central Alaska having large increases in contrast with little change in other areas, such as central Canada and Siberia. Monitoring the spatially variable increase in biospheric activity over the circumpolar region was only possible because of the availability of polar-orbiting satellites.
Furthermore, scatterometer data from satellites provide further evidence that the growing season has lengthened in the Arctic region over the past 20 years. Figure 9.5 shows the progression of the spring 2000 thaw in Alaska. Similar measurements made since 1988 show that the thaw in the Arctic has been advancing by almost 1 day a year. These observations could not have been made without satellites since melting occurs rapidly across the Arctic during the period of melt and the timing varies between years, depending on weather conditions.
FIGURE 9.5 Progression of the spring thaw in Alaska during the year 2000 with snow and ice (blue), ice and slush with bare ground (yellow), and water and bare ground (red). A series of SeaWinds scatterometer measurements on the QuickScat satellite, which are sensitive to water in frozen and liquid states, were used to make these images. SOURCE: Kimball et al. (2006). Reprinted with permission from the American Meteorological Society, the American Geophysical Union, and the Association of American Geographers, copyright 2006.
and Baret (1990), has rigorously demonstrated the potential to retrieve several plant biochemicals from reflectance and transmittance data and is in wide use today. As summarized by Ustin et al. (2004), the list of plant biochemicals has become longer with studies of chlorophyll fluorescence (Zarco-Tejeda et al. 2000a, b), canopy water content (Gao and Goetz 1995, Zarco-Tejeda et al. 2003), and canopy nitrogen content (Kokaly and Clark 1999). Many of the more recent advances are based on new imaging spectroscopy technology using NASA’s Airborne Visible/Infrared Imaging Spectrometer (AVIRIS), an aircraft instrument operated by the Jet Propulsion Laboratory since 1987. NASA has flown one hyperspectral imager in space, the Earth Observing-1 Hyperion, which was launched in 2000 as an engineering test
OCR for page 82
Earth Observations from Space: The First 50 Years of Scientific Achievements
bed yet has continued to operate to today. This technology has significant promise for continued advances in detecting biochemical properties of interest but also for using the high dimensionality of the data to improve land-cover and land-use classifications. Several recent studies have used NASA’s AVIRIS and other hyperspectral imagers to map invasive weeds with high specificity (Figures 9.6 and 9.7; see also Box 9.5, Williams and Hunt 2002, Underwood et al. 2003, Asner and Vitousek 2005).
In a span of slightly more than 25 years, NASA instruments and the research supported by the agency have evolved from primitive correlative studies to physically based accurate analyses. Understanding has advanced rapidly with the synergistic advent of new sensor capabilities such as increased signal to noise ratios, with simultaneous higher spatial and spectral resolution, radiometrically stable instruments, accurate geolocation of images due to advances in satellite pointing control and Global Positioning System (GPS), and development of atmospheric radiative transfer models allowing retrieval of accurate reflectance data. Computer advances have allowed more complex analytical methods to be developed that better match the spatial and spectral patterns in the data. The extensive research funded by NASA through the Earth Observing System program and the scientific advances in understanding our home planet over the past two decades represent a major achievement of the space program.
FIGURE 9.6 A map of invasive species in the Hawaiian rainforest, measured using NASA’s AVIRIS data and impacts of invasive species and plant functional types on biogeochemical cycles. SOURCE: Modified from Asner and Vitousek (2005).
OCR for page 83
Earth Observations from Space: The First 50 Years of Scientific Achievements
BOX 9.5
Detecting Invasive Plant Species
All global ecosystems, with the possible exception of Antarctica, are impacted by invasive species that are substantially changing their functional and structural integrity. Invasion of natural ecosystems represents a serious threat to global biodiversity. Factors attributed to the spread of these species include climate change, land use, land conversion, resource extraction, and habitat fragmentation, combined with international transport. Substantial economic costs are associated with these changes, from loss of agricultural production and increased wildfire frequency to loss of recreational potential. Costs in the United States alone are estimated to exceed $120 billion per year (Pimentel et al. 2005). Recent advances in imaging spectroscopy, a technique to measure a detailed spectrum for all pixels in the image have allowed mapping of individual species and plant communities based on their spectral characteristics. Underwood et al. (2003) used this data to map invasive species in native shrublands along the central coast of California at Vandenberg Air Force Base. Figure 9.7 shows the distribution of invasive species and native plant communities at 3-m pixel resolution for part of the base along the Pacific Coast shoreline. This information is being used by land managers to improve efficiencies in eradication and containment programs. Data of the quality required for mapping individual plant species must currently be acquired by airborne hyperspectral imagers. NASA’s suborbital sciences program has led to the development of this cutting-edge technology and has supported the research required to use it effectively, as shown in the figure.
FIGURE 9.7 Distribution of three invasive species—iceplant, jubata grass, and blue gum—in two native shrub ecosystems—coastal sage scrub and Burton Mesa chaparral—on the central coast of California. The map was produced from a mosaic of flightlines acquired from airborne NASA AVIRIS data, a 224-band imaging spectrometer measuring from the visible through the solar infrared (400-2,500 nm) and measured at a nominal 3-m pixel resolution. SOURCE: Underwood et al. (2006). Reprinted with kind permission of Springer Science and Business Media, copyright 2006.