acquisition of both ground and remotely sensed data. For example, for satellite
imagery, the plot should be of sufficient size to constitute an adequate sample of a
satellite image pixel, whereas for airborne laser data which are acquired by an airplane
in strips, efficiency dictates that the ground plots be located along straight, systematic
lines to facilitate flight paths. The point is that sampling designs must be constructed in
advance to accommodate data from multiple, diverse, independent sources (ground
crews, satellites, airplanes).
Some other examples of applications where careful thought has been put into the
location of the monitoring stations includes monitoring water quality (Dobbie et al, 2008),
air pollution (Zidek et al, 2000), assessment of ecological resources via the US
Environmental Protection Agency’s Environmental Monitoring and Assessment Program
(EMAP) (Stevens and Olsen, 2004), and the monitoring of US forests by the USFS FIA
program as previously discussed. More recently the sampling designs for sensor
networks have been of interest. There are both engineering and statistical sampling
design issues to be considered with these problems (Porter et al, 2005; Borgman et al.
2007).
Invasive species – One example, which is also relevant to sustainability issues,
is the use of sampling designs for the monitoring of invasive species. For example,
understanding and estimating the rate of long distance dispersal is critical for monitoring
and controlling the spread of invasive species. Adaptive spatial sampling designs have
been used in this context (Piellat et al. 2006). In this case, a sequential sampling design
was used where sampling locations were added sequentially by modeling the dispersal
pattern of seed based on data observed. Potential information about the dispersal
parameters at each unsampled location was considered and the new location that
provided the largest information gain was selected. Similarly, effective estimation of the
probability of establishment of the initial population (often referred to as an “Allee effect”)
is essential for controlling the spread of invasive species. Effective temporal sampling
designs that provide information on the initial growth phase of the population are critical
for the estimation of the Allee effect precisely (Dennis, 1989), and since the estimation of
this effect is critically dependent on the size and rate of change of the initial, invading
population, sampling strategies must be developed to target this critical time period.
NEON – Most national-scale observing systems are not built around a cause-
and-effect model but instead seek to efficiently monitor a small number of driving
variables (e.g., environmental, climate) or response variables (e.g., forest productivity,
animal abundance). That is, the sampling design is often not be guided by conceptual
models or scientific questions. The design of monitoring and measurement systems for
understanding, quantifying, and forecasting quantities related to sustainability issues
should, however, be motivated by underlying issues that can be quantified in terms of
conceptual or quantitative models (see “design criteria”). The National Ecological
Observatory Network (NEON) is as an example of a measurement system whose design
was strongly guided by conceptual models within a complex systems perspective. For
example, one focus of NEON is to provide the information necessary to quantify and
forecast changes in biodiversity. In particular, NEON will obtain and maintain a wide
diversity of datasets on both driving and response variables. The selection of these
datasets and the sampling intensity and frequency were identified a priori to capture
changes in time and space and to integrate information across scales. Measurement
systems such as NEON will be invaluable for revealing underlying processes and
barry
(Barry)
#1