Speaking the Language of Remote Sensing
By Doug Muchoney and Bob Unnasch
While satellite images of the earth have been available for over 25 years, only recently has remote sensing become a household term among managers. Recent advances in technology and ownership have made remote sensing a tool not only sharp enough for natural resource monitoring but also affordable for natural resource budgets.
The cost of Landsat Thematic Mapper data, for example, has decreased precipitously from $4000 to just $450 per scene. Further, the data are now being collected more frequently and scenes can be downloaded from the Internet, opening up opportunities for using remote sensing as an integral part of biological monitoring projects.
Measuring the change in biological features of the environment over time is called “change detection.” This term embraces methods such as simple visual comparisons of two aerial photographs of the same forest and complex analysis of long-term time series of satellite images. Change detection techniques can help answer monitoring questions of both a physical and physiological nature. For example, physical changes such as habitat conversion, encroachment of incompatible land use, or habitat fragmentation can be observed using aerial photographs. Physiological changes such as vegetation damage due to insects or changes in moisture levels, however, can be measured using infrared aerial photography or digital satellite data.
Monitoring with remote sensing does not have to be a technical endeavor. How technical you want to get and how much analysis you do depend on personal skill levels. The primary roles of managers are 1) to understand the technology to the extent that they can interact with remote sensing experts and 2) to interpret and understand photos, images, and analysis results. While remote sensing experts know the technology, managers know the ecosystem and their information needs.
To help you consider the appropriate remote sensing imagery, platforms, acquisition strategies, and analysis techniques, we introduce some key techniques and their applications in the context of fundamental monitoring questions.
What Type of Information Do you Need?
The type of information that you want and the geographic area in which you are working will determine the type of imagery you need. Choose a sensor that operates with the appropriate spectral resolution, that is, the number and size of the windows in the electromagnetic spectrum that a sensor records. For example, to monitor vegetation density or biomass, you will need a sensor that measures reflected energy, such as the near-infrared wavelengths. Thermal infrared wavelengths often are used to map fire because they measure radiant heat rather than reflected energy. Microwave radar is usually an active sensor that sends out pulses. The return signals highlight land surface structure and vegetation types. Microwave radiation is particularly useful because it canpierce cloud cover, whereas most optical sensors can not. Because they penetrate water, blue channels are used for coastal and coral mapping. Blue and red channels can measure chlorophyll absorption in healthy green vegetation and are useful for discriminating vegetation types.
How Often Do You Need to Monitor?
The many satellite and airborne sensors sample at different rates. The appropriate data acquisition rate will depend on whether you need to monitor annually, seasonally, or daily. Whatever the monitoring schedule, timing of data acquisition (i.e., temporal resolution) can be critical.
To compare year-to-year changes, for example, data should be collected as close as possible to an annual anniversary date. This not only controls for variation in the sun’s illumination but also allows for comparison of conditions at the same time of the growing season. Beware, however; this kind of reliability is not easily achieved.
While satellites have a fixed acquisition schedule (e.g., every 16 days for U.S. Landsat), the utility of the data will depend on the amount of cloud cover. Cloud cover is an issue for all optical sensors, especially in the humid tropics. To prevent spurious results, each recording should be measured in as similar conditions as possible.
How Much Detail Do You Need?
The trade-off between fine and coarse spatial resolution sensors is the size of the image footprint. The larger the footprint, the less detailed is the spatial information you receive. The spatial resolution of U.S. Landsat data is 30 meters and its footprint 185×170 km. The spatial resolution of an aerial photograph can be centimeters or less. However its footprint would be about 10×10 km.
Fine spatial resolution sensors sample small areas of the earth’s surface at a time and typically do not sample frequently. Practitioners call small footprints “postage stamps.” Small footprints are appropriate either for detailed mapping of small areas or for taking samples of a larger area and extrapolating the results.
Coarse spatial resolution sensors have a larger footprint and can sample an area more frequently. These sensors provide information on phenology, such as temporal profiles of land cover and vegetation types.
How Will You Analyze the Data?
Whether data are collected in analog (e.g., photographs) or digital formats, they can be analyzed either digitally or using manual techniques. For example, U.S. Landsat data are acquired digitally, but photographic prints of the data often are interpreted using standard photo-interpretation techniques. Aerial photographs, alternatively, are taken as analog film and can be interpreted as prints or scanned and analyzed as digital images.
Considerations and Limitations
Ecological monitoring still requires hip-waders and field notebooks. Despite technological advances and highly sophisticated remote sensing hardware, field calibration is vital. Field data are used to calibrate the remote sensing estimates and to link the actual changes observed on the ground with the ability of sensors and analytical techniques to discriminate these changes.
Keep in mind that the sharpest tool may not be the one that is most needed. As technically approachable and as financially feasible as remote sensing has become, it is still the job of on-site practitioners to determine the nature of the changes that the data reveal.
NASA’s Earth Science program: A web-based remote sensing tutorial. (http://landsat.gsfc.nasa.gov/education/tutorials.html)
The International Society for Photogrammetry and Remote Sensing (ISPRS)
TerraServer: Detailed maps and digitized aerial photographs of the entire U.S. available free for download by the public.
The US Geological Survey’s EROS Data Center: The distribution point for U.S. government holdings of satellite data, aerial photography, and much more earth sciences data.
Satellite imagery from Spot Image Corporation. (www.spot.com/)