least static object and relational database elements will be
compatible in space and time.
Latitude, longitude, and elevation coordinates for each
raster element are simultaneously recorded through
triangulation from a constellation of satellites at the time
of data collection. As the technology gets incorporated
into future aerial and satellite sensors, the technique
should become pervasive in field biology from pinpointing
specimen collection sites to recording physical
environmental attributes. At present, however, almost
none of the millions of site and specimen records have
GPS coordinates. While these older data would be useful
in model building, they may be beyond use in GIS terms
because they would impart spatial inaccuracies into models
that would disrupt self-adaptive processes. The dilemma
of rectifying or replacing these early records over the next
few decades, in time to be useful for biodiversity and
environmental modeling, places further importance on
remotely acquired data sets, and argues in favor of
automating traditional field biology techniques (Morain,
1993).
Remote Atmospheric Measurements
Remote sensing technology has developed rapidly since
the mid 1960s (Morain and Budge, 1996). Some
atmospheric data sets that began as sensor experiments
have progressed into operational programs managed
national and international agencies (e.g. National Oceanic
and Atmospheric Administration [NOAA], and
METEOSAT). Experimental global data sets from the
Very High Resolution Radiometer (VHRR), for example,
were initiated on NOAA/POES-1 in 1970 and progressed
to an advanced sensor (AVHRR) on NOAA-6 in 1978.
Despite its name, AVHRR is neither advanced nor high
resolution. It collects five channels of visible, near
infrared, and thermal infrared data primarily for cloud
cover and cloud formation information. In its local area
coverage (LAC) mode, its spatial resolution is = 1.1
kilometer. Calibrated retrospective data from this sensor
are now available from 1985-1991 (NOAA/NESDIS,
1992).
Retrospective AVHRR measurements represent one of
many available time-series data sets. Other satellite data
sets include solar irradiance, aerosol content, trace gas
species and concentrations, water vapor, and land use.
For broad area climate change research, these data are
conveniently, and perhaps only, acquired by satellite-
based sensors, or by other remote platforms. To
prototype a process that self-adapts to the stability resident
in order, on the one hand, but which can also respond to
the chaos of entropy on the other, would be a quantum
step for global change research programs. There is
arguably no higher reward for collecting data from aerial
and satellite platforms than one which links the emergent
processes of nature with resulting patterns in the
landscape, and which accomplishes this without
introducing human bias.
The Earth Radiation Budget Instrument (ERBI) was also
inaugurated in 1978 on NASA's Nimbus-7, but has since
flown on a dedicated platform (ERBS) in 1984 and as part
of the payload on NOAA/POES 9-10. Its purpose is to
record radiation budget, aerosol, and ozone data for global
climate change research. It measures monthly and
seasonal variations in radiation balance at regional scales
on a = 50 kilometer grid.
Many other atmospheric measurements are being collected
508
from satellites. These are mainly water vapor, trace gas
and aerosol concentrations, ultraviolet radiation, radiation
budget, total irradiance, wind, and temperature profiling.
Limb and vertical sounding measurements, and
measurements with specific tropospheric, mesospheric, or
stratospheric depth sensitivities are among the many data
collection strategies. For a variety of logistical and
technical reasons (but mainly because of budget
constraints), few of these data, even for the troposphere,
have been assembled for experimentation in ecosystem
models. The point being made here, however, is that
remote sensing systems of the future will be collecting
data sets that can be used to characterize biological
systems in addition to their primary use in developing
meteorological predictions and global circulation models.
Most of the current ecosystem experiments and landscape
studies still rely on in-situ, near-ground measurements
because remotely acquired data seem too inaccessible to
incorporate into research designs. Sensor systems of the
future should provide a continuous stream of spatially and
temporally contiguous data sets to augment these
discontinuous in-situ measurements, which should in turn
hasten the development of rule-based models.
Remote Terrain Measurements
AVHRR data are also a striking example of secondary
uses for satellite observations. They are often analyszed
for their spectral content in terrestrial studies of vegetation
index patterns; and to analyze principal components of
long time-series data sets to map and monitor severe
landscape changes like fire and deforestation (Eastman and
Fulk, 1993). There is much more that might be done with
these data, if the pixels were spatially correctable with
greater accuracy using GPS coordinates. Future AVHRR
sensors will have this capability, allowing the data to be
analyzed pixel-by-pixel for subtle and gradual landscape
changes, and as continuous data that might drive one or
more modeling rules.
Beside AVHRR, there are numerous sensors already
collecting terrain data in a variety of spectral wavelengths
with ground sampling distances ranging between one and
thirty meters. They operate from aircraft and space
altitudes employing visible, infrared, and microwave
frequencies. Most are opto-mechanical or electro-optical
scanners, radiometers, radars, or lidars, or digital
cameras. Measurement strategies and recording
techniques have undergone ten to twenty years of basic
and applied research, and more recently, operational use.
Some of them offer relatively high pointing accuracy
and/or pixel-to-pixel registration, but most suffer from
imprecise pixel locating ability. As with atmospheric
sensors, the next generation of sensors acquiring terrain
data will be equipped with simultaneous GPS recording
that will allow accurate spectral and temporal data
merges. The technology is trending toward
multiresolution capabilities in the spatial, spectral, and
temporal domains.
Next-generation sensor systems are being designed to
enhance their utility for ecosystem modeling. In addition
to better geolocational capabilities, hyperspectral sensors
are being developed to record data in hundreds of bands
and with bandwidths <5 nanometers. Calibrated data
cubes generated from such sensors should enable much
finer analyses of biological signatures, and depending
upon platform altitude should have ground sampling
distances on the order of meters to kilometers. In the best
circumstances, individual species and microhabitats will
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996
