9.3 Infonnation Integration Foundations
Sections 9.4 through 9.6 and elsewhere III this
book.
Using time and technical tools for exploration,
from GIS to decision support systems, still does little more than juxtapose or overlay information. Developing narrative histories, or pathway diagrams
for the origins of possible impacts, provides opportunities for connections to be discovered and for
possible implications to be explored. Few have explained this better than McHarg (1969) and Holling
(1978). There are many forms of playing and experimenting, whether through qualitative scenarios
or quantitative computer simulations, and maybe
even some form of delphi analysis is critical to exploring the implications of the information. One
rule is to use several. Another is that in the context
of ecological assessment and ecosystem management it may be best to start with systems models,
GIS, and simulation (cf. van der Weide, 1993; Patten, 1994; Russell, 1996), even if later efforts pursue more qualitative approaches.
Whatever the means, integration must yield a
truly integrated product, a picture, or maybe even
a story, depending on the problem and the type of
information available. Simply compiling information, as in many an environmental impact statement
or resource survey, is not enough. True integration
means trading off quantitative and qualitative, certain and uncertain, information. We start by building a systems frame; solidifying it with quantitative, descriptive material from the natural or social
sciences; and then relating and integrating qualitative and anecdotal data to flesh out the picture of
possibilities, trends, dynamics, and implications of
change. Olson (1987) offers a wonderful example
of the use of largely qualitative information on ecological events to develop information for environmental management.
9.3.2 Integration across Space and Time
Integration across space and time entails both conceptual and technical challenges. Conceptual challenges include identifying the appropriate scale for
integration, integrating information from different
scales, and identifying and understanding processes
at different scales. Technically, the critical challenges relate to standardizing and displaying data
and maintaining metadata about different information.
Very often the purpose in integrating data across
space and time is to identify patterns and correlations between events or structures or processes that
can be related to domain-specific knowledge to
yield management-relevant understanding (cf.
123
Fresco and Kroonenberg, 1992). Examples include
forest, vegetation, and biodiversity patterns or land
use and water qUality. Identifying patterns helps to
develop a sense of ecosystem history, larger-scale
dynamics such as path dependency, and constraints
on future possibilities due to past changes.
This requires attention to problems of at least
three different kinds. First is the need to identify
and separate structures and processes at different
scales and the ways they interact to produce macroscopic behaviors and patterns (e.g., Clark, 1987;
Kirkby et aI., 1996). The second is attending to issues of resolution and scale and the different patterns and emphases that may develop from attention at different scales (Costanza and Maxwell,
1994; Slocombe and Sharpe, 1997). And the third,
related to the last, is choosing the appropriate scale
for analysis and integration (Hoekstra et aI., 1991;
Dovers, 1995; Rowe, 1996).
Ideally, both spatial and temporal data will be
integrated in the same system and analysis. In the
case of a study of land use and water quality, for
example, overall data on land-use change and point
data on temporal change in water quality are linked
through their location in the spatial database system. Fostering integration of spatial and temporal
dimensions is critical to real understanding of
ecosystem dynamics and the effects of activities
and change. We seek, for example, to use hierarchy theory to identify multiple, significant scales
or levels at which variables from one domain or
subsystem appear in others (O'Neill, 1988; Fox,
1992; also see Chapter 2).
The practical challenge with spatial and temporal data is to get them into a common format and
system so that they can be integrated through analysis and visualization. This is often primarily a technical question of compatibility. It also requires attention to resolution issues and comparability of
data (metadata issues again). Ideally, all data will
be spatially explicit, with standardized unit boundaries and resolution. Failing this, it is important to
maintain awareness of differences in resolution and
to use finer data to illustrate detail within the larger
grain of the overall data system. In a spatial sense,
this might mean preparing large-scale maps of sites
of special interest within a landscape well understood at a smaller scale and, in a temporal sense,
offering both long-term, low-resolution and detailed, short-term time-series data for systems that
exhibit rapid short-term change within a longer
cyclic pattern. Understanding the limitations of our
data and the ways data were acquired and developed is critical. Some patterns will not be visible
at large-scale resolution; others will not be visible
Sections 9.4 through 9.6 and elsewhere III this
book.
Using time and technical tools for exploration,
from GIS to decision support systems, still does little more than juxtapose or overlay information. Developing narrative histories, or pathway diagrams
for the origins of possible impacts, provides opportunities for connections to be discovered and for
possible implications to be explored. Few have explained this better than McHarg (1969) and Holling
(1978). There are many forms of playing and experimenting, whether through qualitative scenarios
or quantitative computer simulations, and maybe
even some form of delphi analysis is critical to exploring the implications of the information. One
rule is to use several. Another is that in the context
of ecological assessment and ecosystem management it may be best to start with systems models,
GIS, and simulation (cf. van der Weide, 1993; Patten, 1994; Russell, 1996), even if later efforts pursue more qualitative approaches.
Whatever the means, integration must yield a
truly integrated product, a picture, or maybe even
a story, depending on the problem and the type of
information available. Simply compiling information, as in many an environmental impact statement
or resource survey, is not enough. True integration
means trading off quantitative and qualitative, certain and uncertain, information. We start by building a systems frame; solidifying it with quantitative, descriptive material from the natural or social
sciences; and then relating and integrating qualitative and anecdotal data to flesh out the picture of
possibilities, trends, dynamics, and implications of
change. Olson (1987) offers a wonderful example
of the use of largely qualitative information on ecological events to develop information for environmental management.
9.3.2 Integration across Space and Time
Integration across space and time entails both conceptual and technical challenges. Conceptual challenges include identifying the appropriate scale for
integration, integrating information from different
scales, and identifying and understanding processes
at different scales. Technically, the critical challenges relate to standardizing and displaying data
and maintaining metadata about different information.
Very often the purpose in integrating data across
space and time is to identify patterns and correlations between events or structures or processes that
can be related to domain-specific knowledge to
yield management-relevant understanding (cf.
123
Fresco and Kroonenberg, 1992). Examples include
forest, vegetation, and biodiversity patterns or land
use and water qUality. Identifying patterns helps to
develop a sense of ecosystem history, larger-scale
dynamics such as path dependency, and constraints
on future possibilities due to past changes.
This requires attention to problems of at least
three different kinds. First is the need to identify
and separate structures and processes at different
scales and the ways they interact to produce macroscopic behaviors and patterns (e.g., Clark, 1987;
Kirkby et aI., 1996). The second is attending to issues of resolution and scale and the different patterns and emphases that may develop from attention at different scales (Costanza and Maxwell,
1994; Slocombe and Sharpe, 1997). And the third,
related to the last, is choosing the appropriate scale
for analysis and integration (Hoekstra et aI., 1991;
Dovers, 1995; Rowe, 1996).
Ideally, both spatial and temporal data will be
integrated in the same system and analysis. In the
case of a study of land use and water quality, for
example, overall data on land-use change and point
data on temporal change in water quality are linked
through their location in the spatial database system. Fostering integration of spatial and temporal
dimensions is critical to real understanding of
ecosystem dynamics and the effects of activities
and change. We seek, for example, to use hierarchy theory to identify multiple, significant scales
or levels at which variables from one domain or
subsystem appear in others (O'Neill, 1988; Fox,
1992; also see Chapter 2).
The practical challenge with spatial and temporal data is to get them into a common format and
system so that they can be integrated through analysis and visualization. This is often primarily a technical question of compatibility. It also requires attention to resolution issues and comparability of
data (metadata issues again). Ideally, all data will
be spatially explicit, with standardized unit boundaries and resolution. Failing this, it is important to
maintain awareness of differences in resolution and
to use finer data to illustrate detail within the larger
grain of the overall data system. In a spatial sense,
this might mean preparing large-scale maps of sites
of special interest within a landscape well understood at a smaller scale and, in a temporal sense,
offering both long-term, low-resolution and detailed, short-term time-series data for systems that
exhibit rapid short-term change within a longer
cyclic pattern. Understanding the limitations of our
data and the ways data were acquired and developed is critical. Some patterns will not be visible
at large-scale resolution; others will not be visible