At USEPA, a model being used daily to determine the impact of various
factors on water quality is Total Maximum Daily Loads (TMDLs). This
model process is being used to help resource managers determine appropriate resource management decisions:“How much of a pollutant like nitrogen can enter a specific watershed on an annual basis?” The TMDL model
then helps guide regulators to develop permit limits for future nitrogen
point sources to the watershed. The TMDL model is not necessarily a prescriptive-type model because management decisions are not built into the
model. It is an informational model that is used to prioritize and clarify the
water-quality impacts in a watershed. In this particular case, some existing
water-quality data (such as STORET) are appropriate to use based on the
developed questions. Thus, the reliability of field data is a key issue for the
environmental modeler because the statistical significance of the sampling
point(s) may never be known to the decision maker.
9.5 Scale and Variability
Natural systems contain hierarchies of scale in time and space. Events that
control ecosystems often occur as pulse events from the next higher system
(Odum 1988). These control pulses organize the ecosystem in a way that
maximizes energy flow and builds structure that can integrate these pulses
over time and space (Holling 1992). Models that incorporate these features
of differing scales in time and space and account for natural hierarchical
control processes are best suited for use in developing resource management decisions. This type of model may be difficult to conceptualize and
parameterize. The data needed to support these models may also be difficult to acquire.
Often, real-world data measurements that do not conform to the normal
sampling distribution may be difficult to validate in a standard statistical
analysis. These data represent events or processes that exist on a different
time or spatial scale than the rest of the data. Linear (first-order) sampling
schemes tend to disregard or throw out data that is far from the median or
outside the normal distribution. Outlier data may not fit the standard
statistical model, but their importance is hard to neglect. Oftentimes, the
events that do not fit within the normal 95% confidence interval are the
kinds of events that may control the rest of the system. Disturbance regimes
often fit into a hierarchical time and spatial scale that may be difficult to
measure within the time frame and spatial scale of field sampling. When
they do happen, they are sometimes overlooked or discarded because their
effect is far outside the 95% confidence interval of the sampling regime.
Understanding the relationship of “outliers” or other nonnormal data may
be very important in building a model to support resource decisions in an
environment controlled by disturbance.
174
John R. Richardson and Cory W. Berish
factors on water quality is Total Maximum Daily Loads (TMDLs). This
model process is being used to help resource managers determine appropriate resource management decisions:“How much of a pollutant like nitrogen can enter a specific watershed on an annual basis?” The TMDL model
then helps guide regulators to develop permit limits for future nitrogen
point sources to the watershed. The TMDL model is not necessarily a prescriptive-type model because management decisions are not built into the
model. It is an informational model that is used to prioritize and clarify the
water-quality impacts in a watershed. In this particular case, some existing
water-quality data (such as STORET) are appropriate to use based on the
developed questions. Thus, the reliability of field data is a key issue for the
environmental modeler because the statistical significance of the sampling
point(s) may never be known to the decision maker.
9.5 Scale and Variability
Natural systems contain hierarchies of scale in time and space. Events that
control ecosystems often occur as pulse events from the next higher system
(Odum 1988). These control pulses organize the ecosystem in a way that
maximizes energy flow and builds structure that can integrate these pulses
over time and space (Holling 1992). Models that incorporate these features
of differing scales in time and space and account for natural hierarchical
control processes are best suited for use in developing resource management decisions. This type of model may be difficult to conceptualize and
parameterize. The data needed to support these models may also be difficult to acquire.
Often, real-world data measurements that do not conform to the normal
sampling distribution may be difficult to validate in a standard statistical
analysis. These data represent events or processes that exist on a different
time or spatial scale than the rest of the data. Linear (first-order) sampling
schemes tend to disregard or throw out data that is far from the median or
outside the normal distribution. Outlier data may not fit the standard
statistical model, but their importance is hard to neglect. Oftentimes, the
events that do not fit within the normal 95% confidence interval are the
kinds of events that may control the rest of the system. Disturbance regimes
often fit into a hierarchical time and spatial scale that may be difficult to
measure within the time frame and spatial scale of field sampling. When
they do happen, they are sometimes overlooked or discarded because their
effect is far outside the 95% confidence interval of the sampling regime.
Understanding the relationship of “outliers” or other nonnormal data may
be very important in building a model to support resource decisions in an
environment controlled by disturbance.
174
John R. Richardson and Cory W. Berish