Land managers are charged with making decisions that affect multiple ecosystem services, including carbon sequestration, nutrient cycling, and water production. Measuring any of these on the ground can be challenging, and remote sensing measures few ecosystem properties directly. Rather, remote sensing measures electromagnetic energy, which often does relate to these processes but through indirect relationships that contribute to “noise” and make the “signal” harder to detect. The signal/noise ratio is an important concept not just to engineers who develop remote sensing systems, but to ecologists who collect and use the data. The main attraction of remotely sensed images is that they provide synoptic (i.e., wall-to-wall) coverage across space and (if repeated) over time, such that spatiotemporal patterns in ecosystem properties can be inferred. Although, remote sensing data need not be synoptic to be useful; examples of sampling systems will be shown. Underlying any potential application is a simple yet persistent question: what can we realistically infer from remotely sensed data? Examples of successful applications will be described, including forest carbon sequestration, vegetative fuel consumed by fire, plant community composition, and wildlife habitat. The utility of the suite of examples to be presented ranges from obvious to potentially misleading if misused.
Results/Conclusions
A challenge for communicating improved understanding of remote sensing applications not just to the general public, but to specialized and skilled scientists and managers, is getting past the “pretty pictures” that remote sensing often provides. Some ecosystem properties, such as vegetation cover, are relatively easy to estimate, while others are more difficult to estimate, such as how dry (and thus flammable) vegetative fuels are during the fire season. However, some common elements that are critical to the successful application of remote sensing for characterizing ecological conditions are 1) the need for calibrated measurements; 2) the need for a meaningful relationship between the remotely sensed data and the ground measures of interest; and 3) the need for the ground and remote measures to capture the range of variation. Once these conditions are met, then the ecosystem processes may be inferred over space and time for the greater benefit of ecosystem scientists and managers. Often the primary data product is a map, which although never free from error, provides a useful representation of the ecological attribute of interest. Because maps are visual, they impart greater understanding and insight upon the user making decisions, and are therefore recommended instead of figures or tables.