993
UNDERSTANDING THE BIOSPHERE FROM SPACE:
STRATEGIES TO EXPLOIT REMOTE SENSING DATA
Michel M. Verstraete,
Institute for Remote Sensing Applications, TP 440
CEC Joint Research Centre, 1-21020 Ispra (VA), Italy
Bernard Pinty,
Laboratoire de Météorologie Physique
URA 267/CNRS Université Blaise Pascal, F-63177 Aubière, France
and Ranga Myneni
Biospheric Sciences Branch, Mail Code 923
NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA
ABSTRACT
The quantitative interpretation of satellite observations requires the use of mathematical tools to extract the
desired information on terrestrial environments from the radiation data collected in space. A whole range of
approaches can be pursued, from the development of models capable of explaining the nature of the physical
signal being measured and of characterizing the state of the system under observation, to the empirical
correlations between the variables of interest and the space measurements. The premises and implications
of these approaches are outlined, paying special attention to the mathematical and numerical requirements.
The role and specific applications of empirical bidirectional reflectance models is also discussed, even though
these models do not contribute to our understanding of the theory of radiation transfer or to the assessment
of the variables of interest. The advantages and drawbacks of these various approaches and the research
priorities for the next few years are discussed in the context of the planned availability of new sensors.
Key words: Bidirectional reflectance, inversion procedure, modeling, remote sensing, vegetation index.
INTRODUCTION
The scientific community needs to develop the analytical tools and to acquire the relevant data sets to address
issues such as detecting climatic change or assessing its impact, documenting environmental degradation, or
providing support to natural resource exploitation. Models need to be developed to describe and test our
understanding of the relevant processes, to predict the evolution of the global system, and to define data
needs, such as sampling and accuracy requirements.
This modeling effort requires the acquisition of data over a range of spatial scales from regional to global,
with a resolution of 10 m to 100 km, depending on the application. These data should be acquired over a
period of many years with a temporal resolution from hours to a few days. Although some data could and
should be acquired in situ, remote sensing from space platforms appears to be the only economically feasible
way to repetitively gather data on a global basis with a high spatial and temporal resolution.
Satellite remote sensing techniques have raised the expectations of users of data and information. Typ
ical applications include mapping, event detection, support to agriculture, forestry and fishing, weather
forecasting, military surveillance, and the monitoring of unpredictable events, either resulting from human
activities (e.g., deforestation) or from natural disasters. Most of these applications hinge on the availability
of a body of knowledge on the fundamental processes that control or affect the environment. The knowledge
base that justifies these activities and applications is largely due to basic scientific research. The latter
requires specific data to set up the initial and boundary conditions of environmental and climatic models,
to prescribe those variables that are not predicted by these models, as well as to validate them. The re-
evaluation of remote sensing data gathered over the past decade or more also provides a unique opportunity
to assess climatic and environmental issues in a short but nevertheless significant historical perspective.
The two specific objectives of remote sensing research are to provide (i) a reliable characterisation of
the atmosphere, oceans, land, biosphere or cryosphere, and of their interactions, by estimating the values of
certain variables describing the state or properties of the system, or the rates of exchanges of mass, energy
