Remote Sensing
At its simplest definition, remote sensing is obtaining information about an object by a device that is not in contact with the object. In ecology remote sensing usually involves sensors on satellite platforms or airplanes. Most de­vices have a series of sensors that record the intensity of electromagnetic radiation in particular segments of the spectrum for each point, or pixel, in an image. These sensors are designed to collect data in the visible wave­length as well as in other portions of the electromagnetic spectrum (such as the infrared region) that are needed to examine specific aspects of the phys­ical world.
In addition to collecting data from a large part of the electromagnetic spectrum, remote sensing systems collect data over large areas. For instance, the U.S. Landsat satellites record continuous data over an area 71.4 square miles (185 square kilometers) wide. Since some satellites have been in orbit since the mid-1970s scientists can effectively “collect data” from this time period. Therefore, remote sensing offers scientists a wide spectral, spatial, and temporal data range.
For remote sensing to be of use to ecologists the spectral data must be related to some ground-based measurement such as land cover type or veg­etation characteristics (biomass or net primary production, evapotranspi- ration rates, water stress, vegetation structure). Most work in ecology is done at the scale of a small plot, or piece of a field or forest. It can be dif­ficult to extrapolate these small-scale measurements to larger, heteroge­neous areas. Because sensors record continuous data over large areas, remote sensing can be used to “scale-up” plot-based measurements to examine land­scape or even regional patterns. For example, ecologists have used remote sensing data to determine the rate at which rainforest in Brazil is being con­verted to agricultural land. In North America, scientists using satellite data have determined that one of the most endangered ecosystems, the tallgrass prairie, is being replaced by woody vegetation at an alarming rate.
Another set of questions that can be addressed with remote sensing data involves landscape heterogeneity. In these analyses, any of a number of spa­tial statistics can be applied to the original spectral data. Also, the original bands can be recombined to create indices. The most common of these is the Normalized Difference Vegetation Index, a ratio of red to near infrared bands, which has been useful in quantifying vegetation in numerous loca­tions around the world.
Spectral data can be analyzed directly (total infrared reflected) or a clas­sification can be performed on the data. With this method, the spectral data are analyzed and each pixel is assigned to a land cover type: forest, grass­land, or urban. For instance, forests reflect less infrared than grasslands. These land cover data can then be incorporated into a Geographical Infor­mation System (GIS) for further analysis. A GIS is a computer-based sys­tem that can deal with virtually any type of information that can be referenced by geographical location.
Once the land cover types are identified and GIS coverage is generated, additional data such as soil type, elevation, and land use history can be en­tered into the GIS. Ecologists can then ask questions about landscape-level patterns such as the average patch size of a certain land cover type or its dispersion across the landscape. This information can then be related to some ecological process such as the movement or dispersal of animals.
References
Lillesand, Thomas M., and Ralph W. Kiefer. Remote Sensing and Image Interpretation, 4th ed. New York: John Wiley & Sons, 1999.
Schott, John R. Remote Sensing: The Image Chain Approach. Oxford: Oxford Univer­sity Press, 1996.
Schowengerdt, Robert A. Remote Sensing: Models and Methods for Image Processing. San Diego, CA: Academic Press, 1997.