tion in 36 spectral bands (in contrast to 5 on the Advanced Very High Resolution Radiometer and 7 on the Landsat Thematic Mapper Plus) in three
nested spatial resolutions (250, 500, and 1,000 m, depending on the product)
and images the entire globe daily (Running et al. 2000). Large data sets with
short time steps, such as MODIS data, require more effort in terms of data
management. Storage and access must be considered before acquiring the
data.
Plentiful data can be considered a mixed blessing because they allow
for more powerful analysis but demand more planning for delivery, use,
and translation for the audience (Arnold et al. 2000). As a rule, it is wise to
keep a copy of all original data because important information may be lost
through processing and subsetting throughout the analysis.
10.3.4 Field Data Collection
If a review of existing data reveals a lack of applicable data for the model,
new data need to be collected, or the model structure needs to be modified. With the appropriate sampling methods and design, field data collection can obtain specific data for a modeling project. Although field data
collection can be more expensive and time consuming than using existing
data, being able to collect current species- or site-specific data at a relevant
spatial scale can reduce limitations of the data for use in the model. In
addition, the expense and time length of the field data collection can significantly vary, depending upon the data needed and the sampling design.
The following discussion represents some of the considerations for field
data collection. Sample design is an extensive topic in the ecological literature, so only a brief overview is presented here. Generally, these important topics should be considered with a statistician during the modeling
process.
10.3.4.1 A Priori Concepts
A priori knowledge or beliefs about the functioning of the ecological system
influence the structure of the model, the data that are perceived as necessary for the model, and the design of the data collection methods. A model
reflects the modeler’s perception of the system and the factors that are
perceived to influence ecosystem dynamics. Although these a priori ideas
may not be explicitly acknowledged, they are assumptions in the modelbuilding process, and the modeler should be aware of potential biases.
10.3.4.2 Choice of Designs
The choice of design, whether sampling design or experimental design, is
an intricate topic that involves multiple decisions and tradeoffs. The decision between conducting an experimental design or a sampling design must
take into consideration data-collection objectives, the spatial and temporal
10. Effective Ecological Modeling: Data Issues
191
nested spatial resolutions (250, 500, and 1,000 m, depending on the product)
and images the entire globe daily (Running et al. 2000). Large data sets with
short time steps, such as MODIS data, require more effort in terms of data
management. Storage and access must be considered before acquiring the
data.
Plentiful data can be considered a mixed blessing because they allow
for more powerful analysis but demand more planning for delivery, use,
and translation for the audience (Arnold et al. 2000). As a rule, it is wise to
keep a copy of all original data because important information may be lost
through processing and subsetting throughout the analysis.
10.3.4 Field Data Collection
If a review of existing data reveals a lack of applicable data for the model,
new data need to be collected, or the model structure needs to be modified. With the appropriate sampling methods and design, field data collection can obtain specific data for a modeling project. Although field data
collection can be more expensive and time consuming than using existing
data, being able to collect current species- or site-specific data at a relevant
spatial scale can reduce limitations of the data for use in the model. In
addition, the expense and time length of the field data collection can significantly vary, depending upon the data needed and the sampling design.
The following discussion represents some of the considerations for field
data collection. Sample design is an extensive topic in the ecological literature, so only a brief overview is presented here. Generally, these important topics should be considered with a statistician during the modeling
process.
10.3.4.1 A Priori Concepts
A priori knowledge or beliefs about the functioning of the ecological system
influence the structure of the model, the data that are perceived as necessary for the model, and the design of the data collection methods. A model
reflects the modeler’s perception of the system and the factors that are
perceived to influence ecosystem dynamics. Although these a priori ideas
may not be explicitly acknowledged, they are assumptions in the modelbuilding process, and the modeler should be aware of potential biases.
10.3.4.2 Choice of Designs
The choice of design, whether sampling design or experimental design, is
an intricate topic that involves multiple decisions and tradeoffs. The decision between conducting an experimental design or a sampling design must
take into consideration data-collection objectives, the spatial and temporal
10. Effective Ecological Modeling: Data Issues
191