![[logo] US EPA](https://webarchive.library.unt.edu/eot2008/20090508232432im_/http://www.epa.gov/epafiles/images/logo_epaseal.gif){kind=logo}
Imaging Spectroscopy for Determining Rangeland Stressors to Western Watersheds
RESEARCH SUMMARY
Investigators: David J. Williams and William G. Kepner
U.S. Environmental Protection Agency, Office of Research and Development, Reston, Virginia 20192.
The Environmental Protection Agency is developing rangeland ecological indicators in twelve western states using advanced remote sensing techniques. Fine spectral resolution (hyperspectral) sensors, or imaging spectrometers, can detect the subtle spectral features that make vegetation and soil discrimination possible. This study will use hyperspectral remote sensing data, such as NASA’s Airborne Visible-Infra-Red Imaging Spectrometer (AVIRIS), a system capable of 5 to 20 meter spatial resolution. Airborne and satellite remote sensing will provide vegetation mapping at the species level, soil types and characteristics, and landscape information such as erosional features. Vegetation community structure, spatial distribution, and health can then be determined and combined with climatic data to classify rangeland condition and identify disturbed regions.
Accurate determination of rangeland vegetation and soils is required to establish reliable landscape indicators. Rangelands in the West encompass a range of ecological conditions or states from healthy to at risk to degraded. This gradient of conditions can be quantitatively determined and used to develop landscape indicators. Vegetation communities differ over the gradient of rangeland conditions. Soil attributes such as organic matter content, salinity, moisture, mineralogy, and physical condition influence and are influenced by vegetation cover. The water quality of the watershed is directly impacted by these rangeland variables. Imaging spectroscopy can detect these variables and allows for landscape scale assessment and monitoring of stressors to water resources in the West.
Potential research with the Bureau of Land Management, US Department of Agriculture, and U.S. Geological Survey will correlate remote sensing data with ground measurements. The long-term goal of this work is to develop a methodology using current technologies for use with the forthcoming hyperspectral satellite platforms scheduled for operational service within the next 2 to 3 years.
General approach:
The complex nature of rangeland ecological research requires an interdisciplinary approach. Multiple agencies involved in research and management of rangelands such as USGS, BLM, and USDA will be asked to enlist in this effort. Accessing the considerable knowledge of these groups will help ensure a better understanding of the nature of the ecological interactions occurring in these ecosystems and identify gradients in ecological condition. Cooperation with pertinent field research stations and scientific personnel will be important to this study. The research will consist of field data collection, laboratory analysis of collected samples, and remote sensing data collection of hyperspectral and polarimetric radar imagery. Ancillary data such as climatic information, animal stocking rates, and spatial data (DEM’s, DLG’s) will be merged with the remote sensing data products. Modeling of data will be done in a GIS framework in order to develop the landscape indicators required for rangeland assessments and monitoring.
This research will be conducted in a two-phase approach. A pilot study will be done on several areas that have recent and historical imagery. Sites may be part of or adjacent to field research stations or part of a research study such as the US EPA’s Environmental Monitoring and Assessment Program (EMAP). The mission of EMAP is to provide documentation on the current condition of the Nation’s ecological resources, why that condition exists, and predict future ecological conditions (EPA, 1990). This will allow for the collection of good representative data and will provide assistance in the data analysis effort. Sites to be included in the study will encompass a range of ecological conditions or rangeland states from healthy to at risk to degraded. Landscape indicators will then be developed across this gradient of rangeland conditions using remote sensing data and existing data and knowledge that will be obtained from other rangeland management agencies.
The second phase will be an extrapolation of the pilot study to a larger area of the western United States. Subsets of the imagery collected during normal operations by government and commercial data vendors will be added to the study. This imagery will be analyzed according to the methods developed in the pilot study to evaluate the effectiveness of the model over a wide range of ecological conditions. It is hoped that near the end of this phase there will be an operational space borne hyperspectral satellite. This will provide the final segment of this study’s objectives, to use advanced remote sensing systems for large scale assessments of this nation’s rangelands.
Data Analysis and Interpretation:
Imaging spectroscopy data will be analyzed using techniques outlined by Clark (1999), Boardman (1995). Surface reflectance calibration of the imaging spectrometer data will be accomplished using the steps outlined in Clark et al., (1999). Briefly, a radiative transfer algorithm is applied to the data to remove atmospheric absorptions and Rayleigh and aerosol scattering from the data. Next, ground spectral measurements using a portable field spectrometer of areas imaged by the sensor is used to correct the overflight data. Highly reflective and spectrally uniform areas occurring within the flightlines of the sensor are used as ground calibration sites. The spectra of these sites are collected by the field spectrometer and are compared to the imagery and corrections to the image data are made through the use of multipliers to achieve accurate reflectance signatures of materials in the image.
Materials of interest such as soils, minerals, and vegetation will be identified and mapped by comparing the imaging spectrometer reflectance data to in situ ground measurements, standards contained in spectral libraries, and field samples analyzed using laboratory spectrometers. Multiple endmember analysis techniques will be applied to the imagery to detect and map whole pixel and sub-pixel occurrences of target materials (Okin et al., 1998). The steps in the image processing and analysis are as follows (Kruse, 1999): 1) imaging spectrometer data reduction using a minimum noise fraction (MNF) transformation; 2) spatial data reduction using a pixel purity index (PPI); 3) endmember extraction using n-dimensional visualization; 4) spectral identification of target materials by comparison of the image derived reflectance data to in-situ measurements and known standards; 5) mapping of materials using the mixture tuned matched filtering (MTMF) method (Boardman, 1998; Kruze, 1996).
Software tools such as ENVI, Grams32, Tetracorder, SpecPR will be used to process imagery and map target materials. Landscape metrics will be developed using GIS tools such as ESRI ARC/INFO and ArcView Spatial Analyst, and Erdas Imagine. Modeling of data to develop landscape indicators will be accomplished using S-Plus and Systat statistical software packages.
Radar imagery will be processed and analyzed using an approach developed by van Zyl (1989). Radar backscatter modeling will be accomplished using the methods outlined in Taylor et al. (1996), and Kierein-Young and Kruse (1992). Radar fusion with hyperspectral imagery will use the method described in Kierein-Young (1997).
A standardized, effective, and accurate method for determining the state of this nation’s rangelands is needed. This research will provide rangeland managers with the tools they need to assess and monitor these resources. The long-term goal of this work is to develop a methodology using current technologies for use with the forthcoming hyperspectral satellites due in the next 2 to 3 years.
This research will enhance other ESD research efforts that deal with the development of landscape indicators and those that deal with ecosystem assessments in the western EMAP regions. Specifically, the results of this work are directly related to GPRA goal 8.1.1, Development of Landscape Indicators for Use in Regional Risk Assessments and its following subtasks: H - EMAP Western Landscape Pilot and J - Quantification of Landscape Indicators/Watershed Conditions Relationships in a Semiarid Watershed.
- Research Plan. A fully developed and peer-reviewed research plan will
be developed and in place by 10/1/99.
- Field Data Over the next two years, selected sites will be visited
and spectral data Collection will be collected with EPIC’s ASD
FR Spectrometer. Ongoing
- Remote Sensing Working with NASA and commercial vendors, hyperspectral and radar Data Collection imagery will be acquired over many of the study areas. Ongoing
References:
Boardman, J.W., Kruse, F. A., and Green, R. O., 1995, Mapping target signatures via partial unmixing of AVIRIS data: in Summaries, Fifth JPL Airborne Earth Science Workshop, JPL Publication 95-1, v. 1., p. 23 - 26. U.S. Gov. Print. Office, Washington, DC.
Boardman, Joseph, 1998, Leveraging the high dimensionality of AVIRIS data for improved subpixel target unmixing and rejection of false positives: mixure tuned matched filtering, in Summaries of the Seventh Annual JPL Airborne Geoscience Workshop, Pasadena, CA.
Clark, R N. 1999. Spectroscopy of rocks and minerals, and principles of spectroscopy. In: A. N. Rencz (ed), Remote Sensing for the Earth Sciences. American Society for Photogrammetry and Remote Sensing, John Wiley and Sons Inc. New York, New York, USA.
Clark, R.N., G.A. Swayze, T.V. King, K.E. Livo, R.F. Kokaly, J.B. Dalton, J.S. Vance, B.W. Rockwell, and R.R. McDougal. 1999. Surface reflectance calibration of terrestrial imaging spectroscopy data: a tutorial using AVIRIS. US Geological Survey Open File Report (in press).
Congalton, R.G. and K. Green. 1998. Assessing the accuracy of remotely sensed data. CRC Press, Lewis Publishers, Boca Raton, Florida, USA.
EPA. 1990. An overview of the environmental monitoring and assessment program. EMAP Monitor. EPA-600/M-90/022. pp 1-3.
Kierein-Young, K. S., 1997. Integration of optical and radar data to characterize mineralogy and morphology of surfaces in Death Valley, California: International Journal of Remote Sensing, v. 18, no. 7, p. 1517 - 1541.
Kierein - Young, K. S. and Kruse, F. A., 1992. Extraction of Quantitative Surface Characteristics from AIRSAR data for Death Valley, California: Proceedings of the Fourth AIRSAR Workshop, JPL, pp. 46-48. U.S. Gov. Print. Office, Washington, DC.
Klute. A. 1986. Methods of Soil Analysis, Part 1- Physical and Mineralogical Methods. Soil Science Society of America Book, No 5. American Society of Agronomy, Madison, Wisconsin, USA.
Kruse, F. A., 1996, Geologic Mapping Using Combined Analysis of Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) and SIR-C/X-SAR Data: in Proceedings, International Symposium on Optical Science, Engineering, and Instrumentation, SPIE Proceedings, August 1996, Denver , v. 2819, p. 24 - 35.
Kruse, F. A., 1999. Mapping Hot Spring Deposits with AVIRIS at Steamboat Springs, Nevada: in Proceedings of the 8th JPL Airborne Earth Science Workshop: Jet Propulsion Laboratory Publication (in press).
Okin, G.S., W.J. Okin, D.A. Roberts, B. Murray. 1998. Multiple endmember spectral mixture analysis: application to an arid/semi-arid landscape. Proceedings of the 8th Annual JPL Airborne Geoscience Workshop, Pasadena, California, USA. Taylor, G.R., A.H. Mah, F.A. Kruse, K.S. Kierein-Young, R.D. Hewson, and B.A. Bennett. 1996. The extraction of soil dielectric properties from AIRSAR data. Int. J. Remote Sens. 17(3):501-512.
Van Zyl, J.J., R. Carande, Y. Lou, T. Miller, and K. Wheeler. 1992. The NASA/JPL three-frequency polarimetric AIRSAR system. Proceedings of IGARSS 1992.