CIPEC - Remote Sensing

The global environmental change research community faces a significant "information gap" problem. In most areas in the world, little information has been collected on change in land cover, and specifically data on change in forest resources. In cases where forest measuration has been applied (e.g., forest plot sampling), it has usually been conducted for only one point in time. Remote sensing is particularly important in CIPEC's research agenda because it helps fill this information gap.

To date, Landsat Multispectral Scanner (MSS) and Thematic Mapper (TM) images have been the primary data source utilized in CIPEC studies. The Landsat image library is particularly important to the environmental change research agenda and to CIPEC because it provides a common time series of land cover condition, starting in 1972 and continually acquiring new images, for nearly the entire Earth's terrestrial surface. Landsat images therefore provide an opportunity to assess land cover change any chosen site in the Western Hemisphere. However, one of the great challenges facing the Human Dimensions of Environmental Change research program is how to relate human incentives, behavior and action at particular localities to land cover change at broader (e.g., regional) geographic regions. The fine spatial resolution of Landsat images (e.g., 30 meters by 30 meters for Landsat TM) along with its broad geographic extent provides an opportunity in making this linkage.

There are, however, technical challenges that have hindered social scientists from fully capitalizing on remote sensing for land cover change studies. Remote sensing techniques, specifically as one moves toward comparative studies, is technically difficult.

There are several image processing methods that are often ignored in land cover change studies but are critical to ensure correct conclusions are made about human dimensions of global environmental change. When the satellite takes an image, the resultant data recorded represents light reflectance at the satellite. This image data is often referred to as "digital numbers" or DNs. Raw DN images contain noise that cannot be attributed to human actions. This noise interferes with the conclusions we make about human induced changes in comparisons across time and geographic space. This noise includes sensor, illumination, and atmospheric effects at the time of image acquisition. Techniques to remove the noise in Landsat data, specifically radiometric calibration, atmospheric correction and radiometric rectification, have been developed by remote sensing scholars. These techniques convert the raw DN data to light reflectance at the Earth's surface. These techniques have existed for some time, but are extremely technical and, until now, have not been readily accessible for use by non-remote sensing specialists. CIPEC has spent a considerable amount of resources working through various procedures for the different Landsat satellite platforms, and we now have partially automated procedures to help social scientists convert their Landsat data in DNs to surface reflectance.

Why include such a technical discussion here? These calibration procedures are extremely important for the global change research community for two reasons. First, by converting the images from DNs to surface reflectance, the scholar is making his or her image directly comparable to other images. This allows CIPEC to broaden a study's spatial extent by building image mosaics. It also allows for direct comparisons of two similar but geographically distant sites. Because we have these processes documented and standardized, CIPEC can now compare change in ecologically similar forest ecosystems across many of our study locations in different regions of the Western Hemisphere. Second, by converting image DNs to surface reflectance, we now have image data that represents a physical measure of the Earth's surface: light reflectance. Light reflectance is as much as scientific measure as temperature or weight. But we have yet to capitalize on using it as an important measure of land cover change.

The challenge the global change research community now faces is understanding what information content light reflectance provides as a measure of land cover change and how it relates to informing us about other important global change processes. CIPEC researchers are working to identify relationships between light reflectance and other global change parameters such as carbon content in tree biomass. If such relationships can be identified, it raises the possibility that surface reflectance can be used as a proxy for inventorying and mapping global change parameters such as carbon sequestered.

Remote sensing analysis relates to the human dimensions of global change in that it captures the outcomes of human actions across the landscape. It therefore supplies geographic and temporal information for land use and land cover change analysis (see CIPEC's Land Use and Land Cover discussion). One primary analytic technique we often employ is "traditional image analysis" where image pixels are assigned to land cover classes and classification maps are created. Analysis can then look at how different land classes have increased or decreased over time. This type of land cover change analysis is most readily applied in land cover settings where the "grain" of individual change phenomena is spatially larger and more discrete, such as in landscapes where forests are being converted more permanently into agriculture or development. In some circumstances, the disturbances manifested by human actions may be quite small in spatial extent and may occur in more continuous landscapes. For example, group selection harvesting in a completely continuous deciduous forest landscape may not easily be detected using a traditional pixel classification method. In this type of setting, CIPEC has employed a linear mixing technique (sometimes called spectral mixture or sub-pixel analysis) to identify, for each pixel, percentages of important land cover types such as soil, water, and various general classes of forest vegetation. This provides a more sensitive technique to the study of less distinct, but important nevertheless, land cover change.

The goal of remote sensing analysis is to provide information about change in natural resources across a site or a region. Integration of these products with other GIS products (see CIPEC's GIS discussion), allow us to tease out the influence of "natural effects" with that from human decision-making and action. Additional remote sensing-GIS linkages allow CIPEC researchers to investigate the spatial and temporal dimensions of institutions (see the Institutional discussion ) and to studies of population dynamics over temporal and geographic space (see the Demography discussion).

408 North Indiana Avenue, Bloomington, IN 47408-3799
Phone: (812) 855-2230
TDD: (812) 855-7654
Fax: (812) 855-2634

Last Updated: May 11, 2005
Comments: cipec@indiana.edu
Copyright 2005, The Trustees of Indiana University.

Remote Sensing and the Human Dimension of Global Change