- Email: [email protected]
New tools: potential medical applications of data from new and old environmental satellites
Acta Tropica 79 (2001) 35 – 47 www.parasitology-online.com
Review
New tools: potential medical applications of data from new and old environmental satellites Oscar Karl Huh *, John B. Malone Coastal Studies Institute and Pathobiological Sciences, Louisiana State Uni6ersity, Baton Rouge, LA 70803, USA
Abstract The last 40 years, beginning with the first TIROS (television infrared observational satellite) launched on 1 April 1960, has seen an explosion of earth environmental satellite systems and their capabilities. They can provide measurements in globe encircling arrays or small select areas, with increasing resolutions, and new capabilities. Concurrently there are expanding numbers of existing and emerging infectious diseases, many distributed according to areal patterns of physical conditions at the earth’s surface. For these reasons, the medical and remote sensing communities can beneficially collaborate with the objective of making needed progress in public health activities by exploiting the advances of the national and international space programs. Major improvements in applicability of remotely sensed data are becoming possible with increases in the four kinds of resolution: spatial, temporal, radiometric and spectral, scheduled over the next few years. Much collaborative research will be necessary before data from these systems are fully exploited by the medical community. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Earth; Satellite; Emerging; Remote-sensing; Vector-borne
1. Introduction The population of satellites may be divided into two major categories, low earth orbit (circa 860 km) polar orbiting environmental satellites and high earth orbit (circa 35 000 km) equator positioned Geostationary Operational Environmental Satellites (GOES) (Fig. 1). The low earth orbit satellite sensors provide high spatial resolution * Corresponding author. Tel.: + 1-225-578-2952; fax: +1225-578-2520. E-mail address: [email protected] (O.K. Huh).
less frequently (e.g. 1.1 km twice daily, 30 m each 16 days) whereas the high earth orbiting satellites provide less spatial resolution detail but repeat data on emergency time intervals as fast as every five minutes. The satellite borne sensors, predominantly radiometric imaging devices, fall into active and passive categories. Passive sensors measure the natural radiation reflected or emitted from the earth ‘scene’ whereas active sensors illuminate the scene and measure the returning, reflected fragments of the outgoing signal. With the results of research by countless investigators, the physics of
0001-706X/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 1 - 7 0 6 X ( 0 1 ) 0 0 1 0 1 - 2
36
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
radiometry and the interaction of radiation with earth scene elements have advanced to a point that a great variety of measurements are now possible from radiometers on space platforms. Measurements of surface water and land temperature, vegetation conditions, mineral content of soil and rock, atmospheric conditions, ocean turbidity, cloud height, aerosol distribution, wild fires, upper atmospheric winds, rainfall, detailed global topography of lands and seas, and more, are possible from space (Huh, 1991; Kramer,
1996). Sea surface temperature is routinely measured to within less than a degree Celsius from a radiometer aboard a satellite 860 km above the earth. Many kinds of measurements, crude and inaccurate at inception, are evolving into precise and accurate tools that are important to environmental scientists. Images from space convey an immediate impression of areal patterns and structures of the earth, often with great clarity. They provide contextual and quantitative information around the
Fig. 1. The orbits: (a) the NOAA satellite polar orbit, a 101 min period, sun oriented orbital plane yielding revisit at approximately the same hour of each day; (b) geostationary orbit 24 h period, an earth synchronous orbit on the equatorial plane, revisit time is on command, as the satellite is continuously pointed at the fixed earth scene.
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
37
Fig. 2. The electromagnetic spectrum, with AVHRR bands delineated. Transmittance dependant on spectral band and airmass type for atmospheric transmittance conditions.
globe not previously possible before the birth and development of aerospace remote sensing systems. For most scientific uses, extraction of measurements and quantitative information from the imagery is important. Medical earth scientists have long been limited to fragmented and partial views of the conditions at the earth’s surface, and by the time and costliness of field sampling projects. It is the purpose of this paper to introduce the recent and planned advances in the field of satellite remote sensing of earth environments and the potential for future utilization of this growing resource by the medical community.
2. Resolution of the data Resolution has always been a problem in the exploitation of remotely sensed measurements. Each of the four kinds of resolution involved
must be understood since resolution of a sensor is critical to fidelity of its measurements for whatever the application. It is useful to examine what is involved in the concept of ‘resolution’. Spatial resolution: The dimension of the earth scene instantaneous field of view (IFOV) and resulting picture element, referred to as the ‘pixel’ size on the ground. Pixel size can be changed by sub-sampling, IFOV is fixed by sensor optics and satellite altitude. Temporal resolution: This is known as the revisit time interval, or the time interval of repeat sampling. Spectral resolution: This refers to the width of the spectral band within the electromagnetic spectrum that the instrument detector ‘sees’ as well as the number of different narrow bands that the sensor can detect (Fig. 2). Some sensor measurements sample the radiant energy in a series of narrow bands over a large range of
38
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
the solar spectrum 0.4– 1.2 mm. For example, low resolution would be a single broad-band detector whereas high resolution would have more very narrow spectral bands with 30– 100 or more narrow bands within that range. Radiometric resolution: This refers to the quantization interval, or how much of a change in radiant energy intensity there must be to shift from one to another quantization interval. For example, 6 bit data means the data are quantized from 0–63 levels whereas 10 bit data is much more finely resolved into 1024 levels. Critical to the use of any of the generated data sets is the establishment of sensor calibration. Huge sums of money have been appropriated for sensor calibration so that absolute values are achieved or approached and the measurements can be trusted. Some systems have built in calibration systems on-board to assure data quality. Other sensor systems have relied on empirical correlations, many with surprisingly valuable results. All conceivable applications have special resolution requirements. No matter how excellent a measurement is made, if it is not repeated within a reasonable time it is a data artifact and not useful for application. This is why so many experimental satellites have been launched without applications development. Only the long-term operational satellite systems, such as the National Oceanic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR), the Geostationary Operational Environmental Satellite (GOES), Landsat and SPOT, have had strong utility, although all have inherent compromises in terms of available data. For application to infectious diseases (Beck et al., 2000), a complex suite of detectable environmental factors are important, all of which can be observed to a greater or lesser degree from spaceborne platforms. Temperature, air, soil and surface water 1. Diurnal temperature maximum 2. Diurnal temperature minimum 3. Diurnal temperature difference 4. Sea/water/land surface temperature
Water, including soil moisture, standing water and atmospheric water vapor Condition of vegetative canopy over the earth Structure and dynamics of the lower atmosphere plus composition and dimensions of airborne particulates (aerosols) contained Topographic and mineralogy, i.e. terrain relief and bedrock/soil types
3. Earth surface temperature Remote sensing of the earth’s surface temperature from space can be done with a specific portion of the electromagnetic spectrum. Only the 3.5–4 mm and the 8–14 mm wavelength, ‘atmospheric window regions’ will emit infrared energy from the surface into space (Fig. 2). For other parts of the spectrum the atmosphere is opaque. Radiometric measurement of the earth surface temperature using the infrared portion of the electromagnetic spectrum has been the subject of research and development for over 20 years. Remote sensing of water surface temperatures and terrain temperatures have important similarities and differences. Measurements of both require clear, cloud free skies. Even under clear sky conditions, correction must be made for atmospheric attenuation of the signal. Measurements of temperature remotely are based on measurement on radiation and Planck’s law on black body radiation: Tbb =
C2
log 1+
B C1
where Tbb is the black body radiometric temperature, B the spectral radiance in units of W m − 2 mm − 1, L the wavelength, and C1, C2 are the physical constants. However, there are no perfect black bodies and both surface and atmospheric effects must be added to the factors that affect radiometric temperature measurements. The following expressions are formulated to give us the workable estimates of land surface temperature (LST) and sea surface temperature (SST). The following expression includes Planck’s
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
function plus the effects of surface emissivity and atmospheric absorption/emission: Ts =
C2 C1Te ln 1+ − La −Lr L
where Ts is the surface temperature, L the surface radiance, emitted energy, T the atmospheric transmission, e the surface emissivity, La the atmospheric radiance, and Lr the reflected portion of downward radiance. Both surface emissivity and atmospheric effects can have substantial effects on the calculations of surface temperature. As noted by Goetz et al. (2000) emissivity over land surfaces ranging from e =1.0 to 0.99 can result in a temperature measurement offset of 1– 2°C. Water surfaces, typically close to a black body in nature, have a consistent e value of 0.97. After passing through the atmosphere, earth surface emitted radiation has been reduced by interaction with the tri-atomic molecules, i.e. water vapor (predominantly), ozone, carbon dioxide and small amounts of others. Temperature can be measured in the thermal infrared portion of the spectrum. For example, two channels of the NOAA AVHRR sensor show progressively reduced intensity as the water vapor in the atmospheric column increases. The family of transparency curves associated with atmospheric transparency in the atmospheric ‘window’ region of the infrared spectral range is shown in Fig. 2. The difference in transparency between the AVHRR channel 4 (10.5– 11.5 mm) and channel 5 (11.5 –12.5 mm) progressively increases with increasing water vapor in the atmosphere. This relationship is the basis for the atmospheric correction. The temperature difference between channel 4 and 5, along with the channel 4 value plus empirically derived coefficients, are used in an equation of the following form: Ts = T4+1.626(T4− T5) −1.14 where Ts is the temperature of earth’s surface, T4 the radiometric temperature from AVHRR Channel 4 (10.5–11.5), T5 the radiometric temperature from AVHRR Channel 5 (11.5– 12.5).
39
This derived temperature is not corrected for emissivity of land surface types or completely atmospherically corrected, resulting in a temperature error between surface values and satellite measured values. This may be calibrated out with selected ground measurements to offset the errors in the measured surface temperature field. Water bodies in particular can be readily mapped into smoothly defined temperature domains. Surface temperatures in a thermal inertia approach to identifying soil moisture domains was successfully applied in the delta of the Nile (Malone et al., 1994; Huh and Malone, 1995; Malone et al., 1997). By combining mid-afternoon AVHRR surface radiation temperatures with predawn early morning overpass surface radiation temperatures, they were able to create a temperature difference map. This map, confirmed by a field sampling program, suggested that AVHRR thermal difference maps reflect regional hydrologic conditions that can be used as a predictor of environmental risk of schistosomiasis for application to control program management.
4. Remote sensing of vegetation conditions Remotely sensed measurements of visible and near infrared radiation can be used to obtain rapid, time-lapse estimates of plant canopy conditions. The reflectance of green foliage (chlorophyll rich) in various wavelengths of incident solar radiation is shown in Fig. 3. Green leaf reflectance is low in the visible wavelengths 0.5–0.7 mm and the reflectance increases rapidly in the near infrared portion of the spectrum 0.7–1.0 mm (Yates et al., 1984). Note that the healthy and vigorous vegetation has high reflectance in the near infrared range and low reflectance in the visible solar radiation range (Fig. 3). This is due to the fact that vegetation uses photo-synthetically active radiation for photosynthesis and near-infrared radiation is not involved in this process. The sensor response curves (not shown) for both NOAA AVHRR channels 1 and 2 are nearly identical to the Landsat Multispectral Scanner (MSS) bands 5 and 7. Their similarity has meant that Landsat ‘vegetation index studies’
40
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
can be compared directly with the inexpensive AVHRR vegetation imagery from the operational NOAA satellites. The AVHRR has much coarser spatial resolution (1.1 km vs. 0.80 km) but a much better temporal resolution ranging from 3 to 12 hrs. Various mathematical combinations of measurements from these two wavelength bands have been sensitive indicators of the presence of green vegetation and are referred to as vegetation indices (Tucker, 1979; Norwine and Greegor, 1983; Tucker et al., 1984; Yates et al., 1984; McGinness and Tarpley, 1985; Hardisky et al., 1986). Two operational indices have been in use, the simple vegetation index (VI), and a normalized difference vegetation index (NDVI), and are computed from the following equations for AVHRR data:
Fig. 3. Spectral reflectance curve for vegetation. Note Blue, Green and Red portions of the spectrum have low reflectance as they are the photosynthetically active radiation and thus absorbed. The near infrared is strongly reflected. The values combined makeup the vegetation index, a sensitive indicator of chlorophyll condition.
V1= CH2− CH1 and NDVI =
CH2− CH1 CH2+ CH1
The NDVI is preferred for global operational vegetation monitoring because it partially compensates for varied illumination conditions, surface slope and view angle/aspect. Clouds, water, and snow have larger reflectances in the visible and near infrared parts of the spectrum, resulting in negative NDVI values. Soils and rock have similar reflectances in these two bands and thus result in vegetation indices near zero. Fig. 3 illustrates how the absorption of visible radiation and the reflectance of near-infrared radiation, when combined, result in a sensitive indicator of vegetation health. There are several problems with this approach to measuring vegetation condition, including atmospheric influences on the measured radiation and interpretation of what conditions or vegetative processes have caused vegetation index variations. When vegetation is present in the AVHRR scene, the NDVI values typically range between 0.1 and 0.6, with the higher values associated with dense and very green plant canopy. Atmospheric effects including Rayleigh (molecular) scattering, aerosol scattering and sub-pixel clouds all affect the measurements by reducing the computed vegetation indices. This simple surveillance tool is useful for detecting stressed conditions of the plant canopy and development of desertification. There are, however, a number of other plant canopy attributes not separable from each other by this continuous monitoring of the biosphere. These include plant water status, parasite or blight stress, green-leaf area, leaf normal distribution, green phyto-mass and chlorophyll density (Asrar et al., 1989). The vegetation indexes are valuable tools, but only one of many in the development of biospheric surveillance. In fact, in desert regions, rainfall is immediately accompanied by growth of a green plant vegetative canopy, making vegetation indices an indicator of regions that received rainfall. Use of the plant canopy discovery began in the late seventies during the use of Landsat MSS multi-spectral sensors for the agriculture and
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
resources inventory through aerospace remote sensing (AgRISTARS), funded by NASA. The visible and near-infrared measurements can be combined with passive microwave radiometry, Synthetic Aperture Radar (SAR) and other sensors to converge more accurately on desired plant canopy assessment.
5. Hyper-spectral remote sensing in the visible and near-infrared parts of the spectrum Hyper-spectral radiometers refer to those sensor systems with extremely high spectral resolution, that is, making measurements in the earth scene of 30 to over 100 spectral bands simultaneously. Examples of these devices, which have been primarily airborne to date, include the airborne visible/infrared imaging spectrometer (AVIRIS), the MODIS airborne simulator (MAS), and the multi-spectral atmospheric mapping sensor (MAMS). These sensors produce an optical spectroscopic analysis of each pixel, generally in the wavelength band between 0.02 and 2.5 mm. Subtle differences between the radiometric values of earth scene elements are measurable and differentiated with this imagery. AVIRIS was the first operational hyper-spectral radiometer and is typical of these devices. They measure the transmitted, reflected, and scattered solar energy from the Earth’s surface and atmosphere in 224 narrow band channels over large regions, at high spatial resolution. The AVIRIS radiance spectra are used to identify, measure and monitor the constituents of the Earth’s surface and atmosphere based on molecular absorption, reflection, and particle scattering signatures.
6. Radar systems: all weather and day – night capabilities Advanced radar systems hold very considerable promise for mapping the earth surface environmental conditions that are of value for applications by the medical community. They are active, earth illuminating, side-looking sensors (Fig. 4). There are two major kinds of imaging radars, real
41
aperture side looking radar (SLR) and SAR. The SAR differs from the SLR in that special aperture synthesis processing and doppler beam sharpening techniques provide for very high spatial resolution in the range of several meters. They produce strip maps of the roughness elements of the scene illuminated using the portion of the transmitted radar pulse that is scattered/reflected back to the antenna. The measurement variables include: wavelength, transmit/receive polarization combinations, incidence angle, and smoothness or roughness of the scene elements. SAR data is available in a number of different modes. For example, the Canadian Radarsat I, has several modes ranging from narrow swath 8 m resolution to the wide swath 100 m resolution and 500 km wide swath (Fig. 4.). The soon to be launched Radarsat II will have multi-wavelength, multi-polarization and resolution modes as options for remote sensing research. SAR readily provides detailed maps of the earth’s surface roughness fine structure, plant canopy conditions, surface water and soil moisture. Both radar data types are self-illuminating (active) sensors usable in all weather and day or night conditions. The radar transmits a pulse of radio frequency energy to the side of the aircraft/ spacecraft. Smooth surfaces act as perfect mirrors and there is no return signal, resulting in a dark featureless zone. When the transmitted pulse strikes a normal surface a strong reflection is produced. If there are spatially irregular relief structures, i.e. rough and hummocky microtopography of surfaces of soil or plant material, a portion of the incident beam is scattered back to the antenna, which has switched to a receiving mode. SAR is particularly useful in mapping land water edges, standing bodies of water, oil slicks, and most terrain features. It is a powerful tool for use in determining the extent of flooding and ponding of runoff waters. Figs. 5 and 6 illustrate examples of 3 cm wavelength and 6 cm wavelength imagery. A major new development in SAR capabilities has been the development of interferometric SAR. It involves interferometric phase comparison of successive or dual/parallel images. With this mission type and data processing capabilities it is possible to detect subtle changes in the Earth’s
42
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
Fig. 4. Side looking radar systems (a) STAR-3i airborne system, by Earth Watch Inc. Colorado, an X-band system, 3 cm wavelength and (b). Radarsat I, Radarsat International Corp. a C-band system, 6 cm wavelength imaging options.
land and ice surfaces over periods of days to years on unprecedented global scales accurate to the millimeter-level under reliable day or night, allweather conditions. It has been recently used to measure ice stream velocities in Antarctica, co-seismic displacement associated with the Landers and other earthquakes in California, and volcanic deformation on Mt. Etna. SAR interferometry can also be used to generate very high-resolution topographical maps. Recently, the entire Antarctic continent was mapped using Radarsat.
7. Current and previously operational satellites Table 1 lists previous and existing operational satellite systems for which there are extensive historical archives and/or for which concurrent data can reasonably be obtained for any project (ERSC., 2000). Satellite sensor data has been most useful for those long-term space programs, often with a series of successively launched satellites that have built archival data sets that enable multi-year comparison of features in the environment.
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
8. Earth observation satellites planned for the future In the next 3 years, a new generation of satellites planned for the US space program, rapidly expanding programs of other nations and interna-
43
tional consortia, and entry to the scene of corporate ventures will all greatly enhance the possibilities for biomedical application of earth observation satellite data (ERSC., 2000). The US will launch MTI, with 15 multispectral channels at 5 m resolution; NEMO (HRST) with
Fig. 5. SAR image from airborne system, 2.5 m spatial resolution, X-band 3 cm wavelength. Interferometric Synthetic Aperture radar (IFSAR), instrument the STAR-3i, flown on a Learjet 6 November 2000. Image shows Mississippi River, various docks and bridges plus Southern University, housing developments and roadways.
44
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
Fig. 6. SAR image from Radarsat I, C-Band image 6 cm wavelength 30 m spatial resolution. Image shows the Mississippi River, Tunica Hills to the NE, agricultural fields, plus standing water from major rainstorm. Abandoned channel of the Mississippi to the south is Raccourci Old River, an oxbow.
hyperspectral data (210 channels, 30 m) and one panchromatic channel at 5 m resolution; the Aqua (EOS PM-1) companion of Terra (EOSAM-1) with a 36 channel MODIS (250– 1000 m) and other advanced sensors. Two satellites are scheduled to replace currently operational satellites of the NOAA series that have provided continuous AVHRR data records since 1979.
The European Space Agency will launch Envisat-1 to provide radar at 30 or 150 m resolution and multispectral data at 300 and 1200 m resolution. XSTAR, by France and Great Britain, will provide 10+ channels of multispectral data at 20 m, France will launch SPOT-5, Australia will launch ARIES (one 10 m panchromatic; 96 hyperspectral channels at
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
30 m) and Japan will launch ADEOS-II (36 channels, 250–1000 m) and ALOS with one radar channel (10 m), four multispectral (10 m) and 1 panchromatic channel (2.5 m). Canada’s Radarsat-2 will provide enhanced SAR data at
45
3+ m and China/Brazil will launch CBERS-2. A joint India/US program will provide new data from the planned IRS-P5 (Panchromatic, 2.5 m) and IRS-P6 satellites (multispectral at 6, 23.5, 80 m).
Table 1 Past and current operational environmental satellites Satellite name
Sensor
Channel no.
Resolution (m) a
Launch
Revisit interval
Geostationary GOES 1–7 GOES 8,10 Meteosat 1–7 GMS 1–5
VISSR GVAR VIS, IR,WV VISSR
5 5 3 2
4000 1000, 4000, 6000 A, 2500, 5000 1250–5000
1978– 1994– 1977– 1995–
15 15 30 30
Polar-orbiting NOAA 6-14 b NOAA 15, 16
AVHRR AVHRR
5 5
1100 1100
1979– 1999,2000
3–4 h 3–4 h
Terra (EOS AM-1)
MODIS ASTER MISR
36 4 4
250, 500, 1000 15, 30, 90 275
1999
daily 4–16 d 9d
Landsat 1-4 Landsat 4-5 Landsat 7
MSS TM ETM+
4 7 8
79 30, 120 Th 30, 15 Pan, 60 Th
1972-1982 1984 1999
18 d 16 d 16 d
SPOT-2, 3 SPOT-4 (France)
HRV HRV; VI
4 9
20, 10 Pan 20, 10 Pan; 1150
1990, 1993 1998
26 d
ERS-1, 2 (European space agency)
AMI ATRS
1 (radar) 4
26 1000
1991, 1995
35 d
RADARSAT (Canada)
SAR
1 (radar)
8–100
1995
B6 d
IRS 1A-1B
LISS-I LISS-II WiFS LISS-III Pan MSMR OCM
4 4 2 4 1 4 8
72.5 36.25 188 23, 70 5.8
1988, 1991
OrbView-2 (Orbimage)
SeaWiFS
8
IKONOS (Space Imaging)
IKONOS
CBERS-1 (China/Brazil) EROS-A1 (ImageSat international)
IRS 1C-1D
IRS P4 (India)
a b
min min, 3 h, 24 h min min
1995, 1997
1999
2d
1130
1997
B3 d
5
4, 1 Pan
1999
B3 d
CCD IRMSS WFI
5 4 2
20 80, 160 Th 260
1999
26 d 26 d 3–5 d
Pan
1
1.5
2000
daily
Data designated as Th =thermal infrared or Pan = panchromatic (visible, B+W), as appropriate. The TIROS 1-10 (Television Infrared Observation Satellite) series began operations in 1960.
46
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
New corporate satellites scheduled for the new millennium include: (1) OrbView-3 and OrbView-4 (Orbimage) with multispectral data (4 channels) at 4 m spatial resolution, panchromatic data (2 channels) at 1 m resolution and hyperspectral data (200 channels) at 8 m resolution; (2) Resource21 (Resource21), multispectral data (6 channels) at 20, 20 and 100+ m; and (3) EROS-A2 and EROS-B1 (ImageSat International), with a single panchromatic channel at 1.5 or 0.82 m resolution. At a meeting on remote sensing and public health held a decade ago in 1990, a participant discussing the potential for a hypothetical ‘MEDSAT’ space launch to address malaria and other vector-borne diseases indicated that nothing short of ‘PUDDLESAT’ would be needed (Rogers, 1990). New data opportunities in the near future at earth surface resolutions of B1 m may provide exactly that capability, plus greatly enhanced data for environmental studies that span nearly the entire electromagnetic spectrum.
9. Conclusions It is the intent of the authors that this review has provided a useful introduction to satellite remote sensing of earth environments for medical applications. Many of the listed operational systems have a long history and the new sensors are the culmination of much research and engineering development. It will be quite easy to find experts with experience using data from these systems and extensive archives are often available, such as for the NOAA, GOES and Landsat series of satellites. Many of the tools described here are so new, neither experience nor theory help us in planning their exploitation. Trial and error, experimentation and a careful examination is needed to define problems and possibilities for new applications. Hyperspectral and SAR systems in particular are quite at their infancy and hold great promise for providing valuable data sets for research on disease control and the environment.
Acknowledgements Particular appreciation is expressed to the State of LA, specifically: (1) the LA Board of Regents who funded our first satellite earth station with the Louisiana Educational Quality Support Fund, enabling us to capture data directly broadcast by the NOAA satellites; (2) the Louisiana Office of Emergency Preparedness whose National Guard officials quickly understood our capabilities, used our decision support during hurricane threats and then funded our GOES receiver, and (3) the Louisiana Innovative Technology Fund (LTIF) for funding an advanced X-band 4.4 m parabolic antenna to access the high resolution data of the newest generation of environmental satellites. These systems serve to support regional environmental research and surveillance, train students with state-of-the-art facilities, conduct event-triggered research, and provide the people of Louisiana with faster, better access to information for crisis management.
References Asrar, G., Myneni, R.N., Kanemasu, E.T., 1989. Estimation of plant canopy attributes from spectral reflectance measurements. In: Asrar, G. (Ed.), Theory and Applications of Optical Remote Sensing, 734pp. Beck, L.R., Lobitz, B.M., Wood, B.L., 2000. Remote sensing and human health: new sensors and new opportunities. Emerg. Infect. Dis. 6, 217 – 227. ERSC., 2000. U wisc www site. Goetz, S.J., Prince, S.D., Small, J., 2000. Advances in satellite remote sensing of environmental variables for epidemiological applications. Adv. Parasitol. 47, 289 – 306. Hardisky, M.A., Gross, M.F., Kelmas, V., 1986. Remote sensing of coastal wetlands. Bioscience 36, 453 – 460. Huh, O.K., 1991. Limitations and capabilities of the NOAA satellite advanced very high resolution radiometer(AVHRR) for remote sensing of the Earth’s surface. Prev. Vet. Med. 11, 167 – 183. Huh, O.K., Malone, J.B., 1995. NOAA satellites and public health risk assessment: epidemiology of schistosomiasis in the Lower Nile Valley. The Earth Observer 7, 31 – 32. Kramer, H.J., 1996. Observation of the Earth and Its Environment, 3rd enlarged ed. Springer, Berlin (960pp.). Malone, J.B., Huh, O.K., Fehler, D.P., Wilson, P.A., Wilenski, D.E., Holmes, R.E., Elmagdoub, A.I., 1994. Temperature data from satellite imagery and the distribution of schistosomiasis in Egypt. Am. J. Trop. Med. Hyg. 50, 714 – 722.
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47 Malone, J.B., Abdel-Rahman, M.S., El Bahy, M.M., Huh, O.K., Shafik, M., Bavia, M.E., 1997. Parasitol. Today 13, 112 – 119. McGinness, D.F, Tarpley, J.D., 1985. Vegetation cover mapping from NOAA/AVHRR. Adv. Space Res. 5, 359 – 369. Norwine, J., Greegor, D.H., 1983. Vegetation classification based on advanced very high resolution radiometer (AVHRR) satellite imagery. Remote Sens. Environ. 13, 69 – 87. Rogers, D.L., 1990. Pers. Comm.
.
47
Tucker, C.J., 1979. Red and photographic infrared linear combinations for monitoring vegetation. Remote Sens. Environ. 14, 219 – 233. Tucker, C.J., Gatling, J.A., Schneider, S.R., 1984. Monitoring vegetation in the Nile Delta with NOAA-6 and NOAA-7 AVHRR imagery. Photogram. Engng Remote Sens. 50, 53 – 61. Yates, H.W., Tarpley, J.D., Schneider, S.R., McGinnis, D.F., Scofield, R.A., 1984. The role of meteorological satellites in agricultural remote sensing. Remote Sens. Environ. 14, 219 – 233.
Review
New tools: potential medical applications of data from new and old environmental satellites Oscar Karl Huh *, John B. Malone Coastal Studies Institute and Pathobiological Sciences, Louisiana State Uni6ersity, Baton Rouge, LA 70803, USA
Abstract The last 40 years, beginning with the first TIROS (television infrared observational satellite) launched on 1 April 1960, has seen an explosion of earth environmental satellite systems and their capabilities. They can provide measurements in globe encircling arrays or small select areas, with increasing resolutions, and new capabilities. Concurrently there are expanding numbers of existing and emerging infectious diseases, many distributed according to areal patterns of physical conditions at the earth’s surface. For these reasons, the medical and remote sensing communities can beneficially collaborate with the objective of making needed progress in public health activities by exploiting the advances of the national and international space programs. Major improvements in applicability of remotely sensed data are becoming possible with increases in the four kinds of resolution: spatial, temporal, radiometric and spectral, scheduled over the next few years. Much collaborative research will be necessary before data from these systems are fully exploited by the medical community. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Earth; Satellite; Emerging; Remote-sensing; Vector-borne
1. Introduction The population of satellites may be divided into two major categories, low earth orbit (circa 860 km) polar orbiting environmental satellites and high earth orbit (circa 35 000 km) equator positioned Geostationary Operational Environmental Satellites (GOES) (Fig. 1). The low earth orbit satellite sensors provide high spatial resolution * Corresponding author. Tel.: + 1-225-578-2952; fax: +1225-578-2520. E-mail address: [email protected] (O.K. Huh).
less frequently (e.g. 1.1 km twice daily, 30 m each 16 days) whereas the high earth orbiting satellites provide less spatial resolution detail but repeat data on emergency time intervals as fast as every five minutes. The satellite borne sensors, predominantly radiometric imaging devices, fall into active and passive categories. Passive sensors measure the natural radiation reflected or emitted from the earth ‘scene’ whereas active sensors illuminate the scene and measure the returning, reflected fragments of the outgoing signal. With the results of research by countless investigators, the physics of
0001-706X/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 1 - 7 0 6 X ( 0 1 ) 0 0 1 0 1 - 2
36
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
radiometry and the interaction of radiation with earth scene elements have advanced to a point that a great variety of measurements are now possible from radiometers on space platforms. Measurements of surface water and land temperature, vegetation conditions, mineral content of soil and rock, atmospheric conditions, ocean turbidity, cloud height, aerosol distribution, wild fires, upper atmospheric winds, rainfall, detailed global topography of lands and seas, and more, are possible from space (Huh, 1991; Kramer,
1996). Sea surface temperature is routinely measured to within less than a degree Celsius from a radiometer aboard a satellite 860 km above the earth. Many kinds of measurements, crude and inaccurate at inception, are evolving into precise and accurate tools that are important to environmental scientists. Images from space convey an immediate impression of areal patterns and structures of the earth, often with great clarity. They provide contextual and quantitative information around the
Fig. 1. The orbits: (a) the NOAA satellite polar orbit, a 101 min period, sun oriented orbital plane yielding revisit at approximately the same hour of each day; (b) geostationary orbit 24 h period, an earth synchronous orbit on the equatorial plane, revisit time is on command, as the satellite is continuously pointed at the fixed earth scene.
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
37
Fig. 2. The electromagnetic spectrum, with AVHRR bands delineated. Transmittance dependant on spectral band and airmass type for atmospheric transmittance conditions.
globe not previously possible before the birth and development of aerospace remote sensing systems. For most scientific uses, extraction of measurements and quantitative information from the imagery is important. Medical earth scientists have long been limited to fragmented and partial views of the conditions at the earth’s surface, and by the time and costliness of field sampling projects. It is the purpose of this paper to introduce the recent and planned advances in the field of satellite remote sensing of earth environments and the potential for future utilization of this growing resource by the medical community.
2. Resolution of the data Resolution has always been a problem in the exploitation of remotely sensed measurements. Each of the four kinds of resolution involved
must be understood since resolution of a sensor is critical to fidelity of its measurements for whatever the application. It is useful to examine what is involved in the concept of ‘resolution’. Spatial resolution: The dimension of the earth scene instantaneous field of view (IFOV) and resulting picture element, referred to as the ‘pixel’ size on the ground. Pixel size can be changed by sub-sampling, IFOV is fixed by sensor optics and satellite altitude. Temporal resolution: This is known as the revisit time interval, or the time interval of repeat sampling. Spectral resolution: This refers to the width of the spectral band within the electromagnetic spectrum that the instrument detector ‘sees’ as well as the number of different narrow bands that the sensor can detect (Fig. 2). Some sensor measurements sample the radiant energy in a series of narrow bands over a large range of
38
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
the solar spectrum 0.4– 1.2 mm. For example, low resolution would be a single broad-band detector whereas high resolution would have more very narrow spectral bands with 30– 100 or more narrow bands within that range. Radiometric resolution: This refers to the quantization interval, or how much of a change in radiant energy intensity there must be to shift from one to another quantization interval. For example, 6 bit data means the data are quantized from 0–63 levels whereas 10 bit data is much more finely resolved into 1024 levels. Critical to the use of any of the generated data sets is the establishment of sensor calibration. Huge sums of money have been appropriated for sensor calibration so that absolute values are achieved or approached and the measurements can be trusted. Some systems have built in calibration systems on-board to assure data quality. Other sensor systems have relied on empirical correlations, many with surprisingly valuable results. All conceivable applications have special resolution requirements. No matter how excellent a measurement is made, if it is not repeated within a reasonable time it is a data artifact and not useful for application. This is why so many experimental satellites have been launched without applications development. Only the long-term operational satellite systems, such as the National Oceanic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR), the Geostationary Operational Environmental Satellite (GOES), Landsat and SPOT, have had strong utility, although all have inherent compromises in terms of available data. For application to infectious diseases (Beck et al., 2000), a complex suite of detectable environmental factors are important, all of which can be observed to a greater or lesser degree from spaceborne platforms. Temperature, air, soil and surface water 1. Diurnal temperature maximum 2. Diurnal temperature minimum 3. Diurnal temperature difference 4. Sea/water/land surface temperature
Water, including soil moisture, standing water and atmospheric water vapor Condition of vegetative canopy over the earth Structure and dynamics of the lower atmosphere plus composition and dimensions of airborne particulates (aerosols) contained Topographic and mineralogy, i.e. terrain relief and bedrock/soil types
3. Earth surface temperature Remote sensing of the earth’s surface temperature from space can be done with a specific portion of the electromagnetic spectrum. Only the 3.5–4 mm and the 8–14 mm wavelength, ‘atmospheric window regions’ will emit infrared energy from the surface into space (Fig. 2). For other parts of the spectrum the atmosphere is opaque. Radiometric measurement of the earth surface temperature using the infrared portion of the electromagnetic spectrum has been the subject of research and development for over 20 years. Remote sensing of water surface temperatures and terrain temperatures have important similarities and differences. Measurements of both require clear, cloud free skies. Even under clear sky conditions, correction must be made for atmospheric attenuation of the signal. Measurements of temperature remotely are based on measurement on radiation and Planck’s law on black body radiation: Tbb =
C2
log 1+
B C1
where Tbb is the black body radiometric temperature, B the spectral radiance in units of W m − 2 mm − 1, L the wavelength, and C1, C2 are the physical constants. However, there are no perfect black bodies and both surface and atmospheric effects must be added to the factors that affect radiometric temperature measurements. The following expressions are formulated to give us the workable estimates of land surface temperature (LST) and sea surface temperature (SST). The following expression includes Planck’s
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
function plus the effects of surface emissivity and atmospheric absorption/emission: Ts =
C2 C1Te ln 1+ − La −Lr L
where Ts is the surface temperature, L the surface radiance, emitted energy, T the atmospheric transmission, e the surface emissivity, La the atmospheric radiance, and Lr the reflected portion of downward radiance. Both surface emissivity and atmospheric effects can have substantial effects on the calculations of surface temperature. As noted by Goetz et al. (2000) emissivity over land surfaces ranging from e =1.0 to 0.99 can result in a temperature measurement offset of 1– 2°C. Water surfaces, typically close to a black body in nature, have a consistent e value of 0.97. After passing through the atmosphere, earth surface emitted radiation has been reduced by interaction with the tri-atomic molecules, i.e. water vapor (predominantly), ozone, carbon dioxide and small amounts of others. Temperature can be measured in the thermal infrared portion of the spectrum. For example, two channels of the NOAA AVHRR sensor show progressively reduced intensity as the water vapor in the atmospheric column increases. The family of transparency curves associated with atmospheric transparency in the atmospheric ‘window’ region of the infrared spectral range is shown in Fig. 2. The difference in transparency between the AVHRR channel 4 (10.5– 11.5 mm) and channel 5 (11.5 –12.5 mm) progressively increases with increasing water vapor in the atmosphere. This relationship is the basis for the atmospheric correction. The temperature difference between channel 4 and 5, along with the channel 4 value plus empirically derived coefficients, are used in an equation of the following form: Ts = T4+1.626(T4− T5) −1.14 where Ts is the temperature of earth’s surface, T4 the radiometric temperature from AVHRR Channel 4 (10.5–11.5), T5 the radiometric temperature from AVHRR Channel 5 (11.5– 12.5).
39
This derived temperature is not corrected for emissivity of land surface types or completely atmospherically corrected, resulting in a temperature error between surface values and satellite measured values. This may be calibrated out with selected ground measurements to offset the errors in the measured surface temperature field. Water bodies in particular can be readily mapped into smoothly defined temperature domains. Surface temperatures in a thermal inertia approach to identifying soil moisture domains was successfully applied in the delta of the Nile (Malone et al., 1994; Huh and Malone, 1995; Malone et al., 1997). By combining mid-afternoon AVHRR surface radiation temperatures with predawn early morning overpass surface radiation temperatures, they were able to create a temperature difference map. This map, confirmed by a field sampling program, suggested that AVHRR thermal difference maps reflect regional hydrologic conditions that can be used as a predictor of environmental risk of schistosomiasis for application to control program management.
4. Remote sensing of vegetation conditions Remotely sensed measurements of visible and near infrared radiation can be used to obtain rapid, time-lapse estimates of plant canopy conditions. The reflectance of green foliage (chlorophyll rich) in various wavelengths of incident solar radiation is shown in Fig. 3. Green leaf reflectance is low in the visible wavelengths 0.5–0.7 mm and the reflectance increases rapidly in the near infrared portion of the spectrum 0.7–1.0 mm (Yates et al., 1984). Note that the healthy and vigorous vegetation has high reflectance in the near infrared range and low reflectance in the visible solar radiation range (Fig. 3). This is due to the fact that vegetation uses photo-synthetically active radiation for photosynthesis and near-infrared radiation is not involved in this process. The sensor response curves (not shown) for both NOAA AVHRR channels 1 and 2 are nearly identical to the Landsat Multispectral Scanner (MSS) bands 5 and 7. Their similarity has meant that Landsat ‘vegetation index studies’
40
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
can be compared directly with the inexpensive AVHRR vegetation imagery from the operational NOAA satellites. The AVHRR has much coarser spatial resolution (1.1 km vs. 0.80 km) but a much better temporal resolution ranging from 3 to 12 hrs. Various mathematical combinations of measurements from these two wavelength bands have been sensitive indicators of the presence of green vegetation and are referred to as vegetation indices (Tucker, 1979; Norwine and Greegor, 1983; Tucker et al., 1984; Yates et al., 1984; McGinness and Tarpley, 1985; Hardisky et al., 1986). Two operational indices have been in use, the simple vegetation index (VI), and a normalized difference vegetation index (NDVI), and are computed from the following equations for AVHRR data:
Fig. 3. Spectral reflectance curve for vegetation. Note Blue, Green and Red portions of the spectrum have low reflectance as they are the photosynthetically active radiation and thus absorbed. The near infrared is strongly reflected. The values combined makeup the vegetation index, a sensitive indicator of chlorophyll condition.
V1= CH2− CH1 and NDVI =
CH2− CH1 CH2+ CH1
The NDVI is preferred for global operational vegetation monitoring because it partially compensates for varied illumination conditions, surface slope and view angle/aspect. Clouds, water, and snow have larger reflectances in the visible and near infrared parts of the spectrum, resulting in negative NDVI values. Soils and rock have similar reflectances in these two bands and thus result in vegetation indices near zero. Fig. 3 illustrates how the absorption of visible radiation and the reflectance of near-infrared radiation, when combined, result in a sensitive indicator of vegetation health. There are several problems with this approach to measuring vegetation condition, including atmospheric influences on the measured radiation and interpretation of what conditions or vegetative processes have caused vegetation index variations. When vegetation is present in the AVHRR scene, the NDVI values typically range between 0.1 and 0.6, with the higher values associated with dense and very green plant canopy. Atmospheric effects including Rayleigh (molecular) scattering, aerosol scattering and sub-pixel clouds all affect the measurements by reducing the computed vegetation indices. This simple surveillance tool is useful for detecting stressed conditions of the plant canopy and development of desertification. There are, however, a number of other plant canopy attributes not separable from each other by this continuous monitoring of the biosphere. These include plant water status, parasite or blight stress, green-leaf area, leaf normal distribution, green phyto-mass and chlorophyll density (Asrar et al., 1989). The vegetation indexes are valuable tools, but only one of many in the development of biospheric surveillance. In fact, in desert regions, rainfall is immediately accompanied by growth of a green plant vegetative canopy, making vegetation indices an indicator of regions that received rainfall. Use of the plant canopy discovery began in the late seventies during the use of Landsat MSS multi-spectral sensors for the agriculture and
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
resources inventory through aerospace remote sensing (AgRISTARS), funded by NASA. The visible and near-infrared measurements can be combined with passive microwave radiometry, Synthetic Aperture Radar (SAR) and other sensors to converge more accurately on desired plant canopy assessment.
5. Hyper-spectral remote sensing in the visible and near-infrared parts of the spectrum Hyper-spectral radiometers refer to those sensor systems with extremely high spectral resolution, that is, making measurements in the earth scene of 30 to over 100 spectral bands simultaneously. Examples of these devices, which have been primarily airborne to date, include the airborne visible/infrared imaging spectrometer (AVIRIS), the MODIS airborne simulator (MAS), and the multi-spectral atmospheric mapping sensor (MAMS). These sensors produce an optical spectroscopic analysis of each pixel, generally in the wavelength band between 0.02 and 2.5 mm. Subtle differences between the radiometric values of earth scene elements are measurable and differentiated with this imagery. AVIRIS was the first operational hyper-spectral radiometer and is typical of these devices. They measure the transmitted, reflected, and scattered solar energy from the Earth’s surface and atmosphere in 224 narrow band channels over large regions, at high spatial resolution. The AVIRIS radiance spectra are used to identify, measure and monitor the constituents of the Earth’s surface and atmosphere based on molecular absorption, reflection, and particle scattering signatures.
6. Radar systems: all weather and day – night capabilities Advanced radar systems hold very considerable promise for mapping the earth surface environmental conditions that are of value for applications by the medical community. They are active, earth illuminating, side-looking sensors (Fig. 4). There are two major kinds of imaging radars, real
41
aperture side looking radar (SLR) and SAR. The SAR differs from the SLR in that special aperture synthesis processing and doppler beam sharpening techniques provide for very high spatial resolution in the range of several meters. They produce strip maps of the roughness elements of the scene illuminated using the portion of the transmitted radar pulse that is scattered/reflected back to the antenna. The measurement variables include: wavelength, transmit/receive polarization combinations, incidence angle, and smoothness or roughness of the scene elements. SAR data is available in a number of different modes. For example, the Canadian Radarsat I, has several modes ranging from narrow swath 8 m resolution to the wide swath 100 m resolution and 500 km wide swath (Fig. 4.). The soon to be launched Radarsat II will have multi-wavelength, multi-polarization and resolution modes as options for remote sensing research. SAR readily provides detailed maps of the earth’s surface roughness fine structure, plant canopy conditions, surface water and soil moisture. Both radar data types are self-illuminating (active) sensors usable in all weather and day or night conditions. The radar transmits a pulse of radio frequency energy to the side of the aircraft/ spacecraft. Smooth surfaces act as perfect mirrors and there is no return signal, resulting in a dark featureless zone. When the transmitted pulse strikes a normal surface a strong reflection is produced. If there are spatially irregular relief structures, i.e. rough and hummocky microtopography of surfaces of soil or plant material, a portion of the incident beam is scattered back to the antenna, which has switched to a receiving mode. SAR is particularly useful in mapping land water edges, standing bodies of water, oil slicks, and most terrain features. It is a powerful tool for use in determining the extent of flooding and ponding of runoff waters. Figs. 5 and 6 illustrate examples of 3 cm wavelength and 6 cm wavelength imagery. A major new development in SAR capabilities has been the development of interferometric SAR. It involves interferometric phase comparison of successive or dual/parallel images. With this mission type and data processing capabilities it is possible to detect subtle changes in the Earth’s
42
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
Fig. 4. Side looking radar systems (a) STAR-3i airborne system, by Earth Watch Inc. Colorado, an X-band system, 3 cm wavelength and (b). Radarsat I, Radarsat International Corp. a C-band system, 6 cm wavelength imaging options.
land and ice surfaces over periods of days to years on unprecedented global scales accurate to the millimeter-level under reliable day or night, allweather conditions. It has been recently used to measure ice stream velocities in Antarctica, co-seismic displacement associated with the Landers and other earthquakes in California, and volcanic deformation on Mt. Etna. SAR interferometry can also be used to generate very high-resolution topographical maps. Recently, the entire Antarctic continent was mapped using Radarsat.
7. Current and previously operational satellites Table 1 lists previous and existing operational satellite systems for which there are extensive historical archives and/or for which concurrent data can reasonably be obtained for any project (ERSC., 2000). Satellite sensor data has been most useful for those long-term space programs, often with a series of successively launched satellites that have built archival data sets that enable multi-year comparison of features in the environment.
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
8. Earth observation satellites planned for the future In the next 3 years, a new generation of satellites planned for the US space program, rapidly expanding programs of other nations and interna-
43
tional consortia, and entry to the scene of corporate ventures will all greatly enhance the possibilities for biomedical application of earth observation satellite data (ERSC., 2000). The US will launch MTI, with 15 multispectral channels at 5 m resolution; NEMO (HRST) with
Fig. 5. SAR image from airborne system, 2.5 m spatial resolution, X-band 3 cm wavelength. Interferometric Synthetic Aperture radar (IFSAR), instrument the STAR-3i, flown on a Learjet 6 November 2000. Image shows Mississippi River, various docks and bridges plus Southern University, housing developments and roadways.
44
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
Fig. 6. SAR image from Radarsat I, C-Band image 6 cm wavelength 30 m spatial resolution. Image shows the Mississippi River, Tunica Hills to the NE, agricultural fields, plus standing water from major rainstorm. Abandoned channel of the Mississippi to the south is Raccourci Old River, an oxbow.
hyperspectral data (210 channels, 30 m) and one panchromatic channel at 5 m resolution; the Aqua (EOS PM-1) companion of Terra (EOSAM-1) with a 36 channel MODIS (250– 1000 m) and other advanced sensors. Two satellites are scheduled to replace currently operational satellites of the NOAA series that have provided continuous AVHRR data records since 1979.
The European Space Agency will launch Envisat-1 to provide radar at 30 or 150 m resolution and multispectral data at 300 and 1200 m resolution. XSTAR, by France and Great Britain, will provide 10+ channels of multispectral data at 20 m, France will launch SPOT-5, Australia will launch ARIES (one 10 m panchromatic; 96 hyperspectral channels at
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
30 m) and Japan will launch ADEOS-II (36 channels, 250–1000 m) and ALOS with one radar channel (10 m), four multispectral (10 m) and 1 panchromatic channel (2.5 m). Canada’s Radarsat-2 will provide enhanced SAR data at
45
3+ m and China/Brazil will launch CBERS-2. A joint India/US program will provide new data from the planned IRS-P5 (Panchromatic, 2.5 m) and IRS-P6 satellites (multispectral at 6, 23.5, 80 m).
Table 1 Past and current operational environmental satellites Satellite name
Sensor
Channel no.
Resolution (m) a
Launch
Revisit interval
Geostationary GOES 1–7 GOES 8,10 Meteosat 1–7 GMS 1–5
VISSR GVAR VIS, IR,WV VISSR
5 5 3 2
4000 1000, 4000, 6000 A, 2500, 5000 1250–5000
1978– 1994– 1977– 1995–
15 15 30 30
Polar-orbiting NOAA 6-14 b NOAA 15, 16
AVHRR AVHRR
5 5
1100 1100
1979– 1999,2000
3–4 h 3–4 h
Terra (EOS AM-1)
MODIS ASTER MISR
36 4 4
250, 500, 1000 15, 30, 90 275
1999
daily 4–16 d 9d
Landsat 1-4 Landsat 4-5 Landsat 7
MSS TM ETM+
4 7 8
79 30, 120 Th 30, 15 Pan, 60 Th
1972-1982 1984 1999
18 d 16 d 16 d
SPOT-2, 3 SPOT-4 (France)
HRV HRV; VI
4 9
20, 10 Pan 20, 10 Pan; 1150
1990, 1993 1998
26 d
ERS-1, 2 (European space agency)
AMI ATRS
1 (radar) 4
26 1000
1991, 1995
35 d
RADARSAT (Canada)
SAR
1 (radar)
8–100
1995
B6 d
IRS 1A-1B
LISS-I LISS-II WiFS LISS-III Pan MSMR OCM
4 4 2 4 1 4 8
72.5 36.25 188 23, 70 5.8
1988, 1991
OrbView-2 (Orbimage)
SeaWiFS
8
IKONOS (Space Imaging)
IKONOS
CBERS-1 (China/Brazil) EROS-A1 (ImageSat international)
IRS 1C-1D
IRS P4 (India)
a b
min min, 3 h, 24 h min min
1995, 1997
1999
2d
1130
1997
B3 d
5
4, 1 Pan
1999
B3 d
CCD IRMSS WFI
5 4 2
20 80, 160 Th 260
1999
26 d 26 d 3–5 d
Pan
1
1.5
2000
daily
Data designated as Th =thermal infrared or Pan = panchromatic (visible, B+W), as appropriate. The TIROS 1-10 (Television Infrared Observation Satellite) series began operations in 1960.
46
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47
New corporate satellites scheduled for the new millennium include: (1) OrbView-3 and OrbView-4 (Orbimage) with multispectral data (4 channels) at 4 m spatial resolution, panchromatic data (2 channels) at 1 m resolution and hyperspectral data (200 channels) at 8 m resolution; (2) Resource21 (Resource21), multispectral data (6 channels) at 20, 20 and 100+ m; and (3) EROS-A2 and EROS-B1 (ImageSat International), with a single panchromatic channel at 1.5 or 0.82 m resolution. At a meeting on remote sensing and public health held a decade ago in 1990, a participant discussing the potential for a hypothetical ‘MEDSAT’ space launch to address malaria and other vector-borne diseases indicated that nothing short of ‘PUDDLESAT’ would be needed (Rogers, 1990). New data opportunities in the near future at earth surface resolutions of B1 m may provide exactly that capability, plus greatly enhanced data for environmental studies that span nearly the entire electromagnetic spectrum.
9. Conclusions It is the intent of the authors that this review has provided a useful introduction to satellite remote sensing of earth environments for medical applications. Many of the listed operational systems have a long history and the new sensors are the culmination of much research and engineering development. It will be quite easy to find experts with experience using data from these systems and extensive archives are often available, such as for the NOAA, GOES and Landsat series of satellites. Many of the tools described here are so new, neither experience nor theory help us in planning their exploitation. Trial and error, experimentation and a careful examination is needed to define problems and possibilities for new applications. Hyperspectral and SAR systems in particular are quite at their infancy and hold great promise for providing valuable data sets for research on disease control and the environment.
Acknowledgements Particular appreciation is expressed to the State of LA, specifically: (1) the LA Board of Regents who funded our first satellite earth station with the Louisiana Educational Quality Support Fund, enabling us to capture data directly broadcast by the NOAA satellites; (2) the Louisiana Office of Emergency Preparedness whose National Guard officials quickly understood our capabilities, used our decision support during hurricane threats and then funded our GOES receiver, and (3) the Louisiana Innovative Technology Fund (LTIF) for funding an advanced X-band 4.4 m parabolic antenna to access the high resolution data of the newest generation of environmental satellites. These systems serve to support regional environmental research and surveillance, train students with state-of-the-art facilities, conduct event-triggered research, and provide the people of Louisiana with faster, better access to information for crisis management.
References Asrar, G., Myneni, R.N., Kanemasu, E.T., 1989. Estimation of plant canopy attributes from spectral reflectance measurements. In: Asrar, G. (Ed.), Theory and Applications of Optical Remote Sensing, 734pp. Beck, L.R., Lobitz, B.M., Wood, B.L., 2000. Remote sensing and human health: new sensors and new opportunities. Emerg. Infect. Dis. 6, 217 – 227. ERSC., 2000. U wisc www site. Goetz, S.J., Prince, S.D., Small, J., 2000. Advances in satellite remote sensing of environmental variables for epidemiological applications. Adv. Parasitol. 47, 289 – 306. Hardisky, M.A., Gross, M.F., Kelmas, V., 1986. Remote sensing of coastal wetlands. Bioscience 36, 453 – 460. Huh, O.K., 1991. Limitations and capabilities of the NOAA satellite advanced very high resolution radiometer(AVHRR) for remote sensing of the Earth’s surface. Prev. Vet. Med. 11, 167 – 183. Huh, O.K., Malone, J.B., 1995. NOAA satellites and public health risk assessment: epidemiology of schistosomiasis in the Lower Nile Valley. The Earth Observer 7, 31 – 32. Kramer, H.J., 1996. Observation of the Earth and Its Environment, 3rd enlarged ed. Springer, Berlin (960pp.). Malone, J.B., Huh, O.K., Fehler, D.P., Wilson, P.A., Wilenski, D.E., Holmes, R.E., Elmagdoub, A.I., 1994. Temperature data from satellite imagery and the distribution of schistosomiasis in Egypt. Am. J. Trop. Med. Hyg. 50, 714 – 722.
O.K. Huh, J.B. Malone / Acta Tropica 79 (2001) 35–47 Malone, J.B., Abdel-Rahman, M.S., El Bahy, M.M., Huh, O.K., Shafik, M., Bavia, M.E., 1997. Parasitol. Today 13, 112 – 119. McGinness, D.F, Tarpley, J.D., 1985. Vegetation cover mapping from NOAA/AVHRR. Adv. Space Res. 5, 359 – 369. Norwine, J., Greegor, D.H., 1983. Vegetation classification based on advanced very high resolution radiometer (AVHRR) satellite imagery. Remote Sens. Environ. 13, 69 – 87. Rogers, D.L., 1990. Pers. Comm.
.
47
Tucker, C.J., 1979. Red and photographic infrared linear combinations for monitoring vegetation. Remote Sens. Environ. 14, 219 – 233. Tucker, C.J., Gatling, J.A., Schneider, S.R., 1984. Monitoring vegetation in the Nile Delta with NOAA-6 and NOAA-7 AVHRR imagery. Photogram. Engng Remote Sens. 50, 53 – 61. Yates, H.W., Tarpley, J.D., Schneider, S.R., McGinnis, D.F., Scofield, R.A., 1984. The role of meteorological satellites in agricultural remote sensing. Remote Sens. Environ. 14, 219 – 233.