![[logo] US EPA](https://webarchive.library.unt.edu/eot2008/20081107112906im_/http://www.epa.gov/epafiles/images/logo_epaseal.gif){kind=logo}
2000 Progress Report: Assessment and Analysis of Ecosystem Stressors Across Scales Using Remotely Sensed Imagery Reducing Uncertainty in Managing the Colorado Plateau Ecosystem
EPA Grant Number: R825152Title: Assessment and Analysis of Ecosystem Stressors Across Scales Using Remotely Sensed Imagery Reducing Uncertainty in Managing the Colorado Plateau Ecosystem
Investigators: Weigel, Stephanie J.
Institution: Colorado State University , University of Wisconsin - Madison
EPA Project Officer: Levinson, Barbara
Project Period: October 1, 1996 through September 30, 1999
Project Period Covered by this Report: October 1, 1999 through September 30, 2000
Project Amount: $251,237
RFA: Ecological Assessment (1996)
Research Category: Ecological Indicators/Assessment/Restoration
Description:
Objective:The project investigated issues of scale in reducing uncertainty in ecosystem management for the Colorado Plateau ecosystem, by examining potential characteristic scales at which environmental stressors and their effects may be manifested on ecosystem landscapes, as detected by remotely sensed imagery. The project developed techniques for using multiscale, remotely sensed data in the characterization and analysis of landscapes at the ecosystem level. Thus, environmental stressors are characterized across both temporal (1970s?1990s) and spatial (60 m?1 km pixel resolution) scales. Knowledge of characteristic scales provides managers and researchers with guidelines for selecting scales at which to capture or aggregate data, as well as information on the scales of processes and factors that have the potential to threaten ecosystem integrity.
Progress Summary:Year 1 of the project focused on development of the tools for the scale analysis methods evaluated: fractal analysis, multiscale variance, variogram analysis and local variance analysis. These were developed using both the ICAMS software (Qui et al. 1999) and the modeling capabilities of the ERDAS Imagine image processing software. Also included in the Year One research was initial evaluation and development of the image resampling algorithm for rescaling. A rescaling algorithm was chosen and developed based on a dampened sine wave interpolation function. Dr. Kenneth McGwire was a project consultant for the development and implementation code of the algorithm. Years Two and Three focused on image mosaic creation and scale analysis of mosaics and subsets. Mosaics were created for the 70s, 80s and 90s from the Landsat MSS NALC data sets. Several issues arose with the 1970s data set, involving missing data resulting in a non-continuous surface. Efforts to amend or supplement this data through the NALC program were unsuccessful, for that reason the subsequent ecosystem-wide (mosaic) analyses were performed only on the 1980s and 1990s data sets. A series of image subsets were developed to represent some of the variety of landscapes in the ecosystem (e.g., agricultural, sparsely vegetated desert, vegetated riparian), and analyses were performed on all available years of these data sets. Scale analysis using the four scale analysis algorithms was performed on the mosaic images (1980s and 1990s) and subset images (1970s where available, 1980s, and 1990s).
Results of the scale analysis as performed on both the mosaic images and the subset images was somewhat mixed. There was less between-technique commonality in results than was expected. The local variance and fractal techniques appeared to provide the most insight and information as to issues of characteristic scales and landscape structures. Global variance analyses were less definitive, and variogram analyses using the methods chosen here were difficult to execute and interpret. Recommendations (summarized in last year's lengthy annual report) are for the use of local variance and fractal analysis as the preferred scale analysis methodologies.
Change detection analyses were completed on the image subsets, to assess the influence of scale on the ability to detect change, using established methodologies and available software programs (here ERDAS Imagine). Methodologies for change detection were assessed from several in the published literature with potential to accurate description of change and appropriate to the landscape. The methods assessed were: NDVI image differencing (discussed in Lyon et al. 1998); single band image (MSS Band 2) differencing based on At Satellite Planetary Reflectance (ASPR) normalization (see Chavez and MacKinnon 1994); and selective Principal Components Analysis (PCA) using a single band from each image (Chavez and MacKinnon 1994, Mas 1999). After initial assessment, the selective PCA method was chosen for work assessing change at the different resolution cell sizes.
The PCA change detection analysis was performed across the rescaled subset images. Accuracy assessment was performed using National Aerial Photography Program (NAPP) aerial photography as supplemental ground truth. Overall, the effects of scale on accuracy indicated that change detection was assessed with similar accuracies at the majority of the scales utilized in this research, except for the coarsest pixel resolution (1920 m x 1920 m). In eight of the ten subsets there was a noticeable decrease in accuracy at this coarse resolution. Since change occurring on the landscape is likely to be occurring at a resolution that is higher than can be captured by this pixel scale, loss of accuracy is likely a reflection of inadequate representation at this scale.
References:
Chavez PS, MacKinnon DJ. Automatic detection of vegetation changes in the southwestern United States using remotely sensed images. Photogrammetric Engineering and Remote Sensing 1994;60:571-583.
Lyon JG, Yuan D, Lunetta R, Elvidge C. A change detection experiment using vegetation indices. Photogrammetric Engineering and Remote Sensing 1998;64:143-150.
Mas JF. Monitoring land-cover changes: a comparison of change detection techniques. International Journal of Remote Sensing 1999;20:139-152.
Qui H-L, Lam NS-N, Quattrochi D, Gamon JA. Fractal characterization of hyperspectral imagery. Photogrammetric Engineering and Remote Sensing 1999;65:63-71.
Future Activities:This is the last annual report for the project. The final report will be prepared and submitted to EPA, and an article is in preparation for submission to a professional remote sensing journal.
Journal Articles:No journal articles submitted with this report: View all 6 publications for this project
Supplemental Keywords:ecosystem, scaling, integrated assessment, scale effects analysis, Landsat, NALC. , Ecosystem Protection/Environmental Exposure & Risk, Geographic Area, Scientific Discipline, RFA, Ecosystem/Assessment/Indicators, exploratory research environmental biology, Ecology, Ecological Risk Assessment, Ecological Indicators, Ecological Effects - Human Health, Chemical Mixtures - Environmental Exposure & Risk, Ecological Effects - Environmental Exposure & Risk, Ecosystem Protection, Ecology and Ecosystems, Environmental Monitoring, State, multi-scale biophysical models, Colorado Plateau ecosystem, remote sensing, landscape characterization, environmental stressor, EOS, scaling, analytical algorithm, variance analysis, ecological exposure, ecosystem assessment, environmental stress, fractal analysis, ecological assessment, assessment methods, ecological impacts, multiple stressors
Relevant Websites:
The project is included in the Environmental Health Advanced Systems Laboratory
(EHASL) Web page at http://ehasl.cvmbs.colostate.edu
. The page is currently under
revision, but the link to the project can be found at http://ehasl.cvmbs.colostate.edu/remote
. The PI no longer works at EHASL,
so future updates to the project page would occur at another site.
Progress and Final Reports:
1998 Progress Report
1999 Progress Report
Original Abstract
Final Report