consider how the data should be collected, with consideration of statistical design and
other data collection issues. Finally, we discuss several overarching issues associated
with application of models for synthesizing such data and for forecasting future needs
and states.
2.1 What to measure/monitor?
A key issue in sustainability science is to construct a framework to collect data
that will help scientists monitor progress toward sustainability. We consider examples
from three categories of variables: biotic, abiotic, and human-element datasets, and we
explore the challenges with collecting and converting the raw data into useful data
products.
2.1.1 Biotic Datasets
An example of biotic data and datasets is the Forest Inventory and Analysis (FIA)
program of the U.S. Forest Service. The USFS conducts a national forest inventory of
the USA (McRoberts et al. 2005). The program’s mission is to assess the current state
and health of the Nation’s forest resources and the change in those resources over time.
On plots distributed across the country at a sampling intensity of one plot per 2400 ha,
field crews observe or measure human variables such as land ownership, abiotic
variables such as soils and topography, and a suite of biotic variables that includes tree
species, diameter, height, health, and live/dead status and plot variables such as
proportion forest, regeneration, and understory vegetation. The data cover the entire
country, are acquired in a nationally consistent manner, and are maintained in a publicly
accessible database that is updated annually. Collection of these data is at least partially
motivated by the necessity of reporting to an international sustainability convention using
a set of agreed-upon criteria and indicators. A good reference for this criteria is at:
(http://www.rinya.maff.go.jp/mpci/criteria_e.html). While such datasets are very useful
for addressing basic science questions, they may also provide an important time-series
for evaluating progress towards sustaining, for example, our forest ecosystems.
2.1.2 Abiotic Datasets
Natural datasets that are not biological in origin include climatic variables,
measures of air and water pollution, and oceanographic measurements. Although the
original intention of obtaining such data may not have been motivated by sustainability
issues, such data are often critical to understanding changes in biotic variables that may
be the target of sustainability research and monitoring. Many abiotic datasets are
collected by government agencies with public funding, but standards vary for the
archiving and accessibility of the datasets. For effective use in long-term monitoring,
many of the datasets require gridding. For example, climatic datasets are often reported
as averages over five degree latitude and longitude grid cells, which are then further
aggregated into hemispheric or global averages. However, the aggregation procedure
raises many issues connected with efficient interpolation methods, the effect of changes
over time of the observing network, and quality control. For example, raw temperature
data may show a sudden changepoint due to moving the observing station, or a gradual
but localized trend due to the urban heat island effect, but these sources of variability do
not reflect long-term variability in the true climate system, and therefore the temperature
data may need to be corrected in the calculation of climatic datasets.