Research

Introduction

Research at the Viper Lab is based primarily on a strong physics-based analysis of remote sensing data. Major research tools include spectral mixture analysis for surface decomposition, decision trees for land-cover mapping and radiative transfer for atmospheric correction and retrieval of atmospheric constituents such as water vapor and methane. A common theme has been the development of spectral libraries for many surfaces, ranging from natural vegetation to urban areas and the development of tools for analyzing these spectra. Additional research specialties include remote sensing of wildfire fuels and fire danger, land-use/land-cover change analysis in the Amazon, mapping vegetation species and plant functional types using imaging spectrometry and, recently, methane mapping using AVIRIS. A primary funding source is NASA, although support has also been provided by the NSF, NGA, and Naval Post Graduate School among other funding sources.

Terrestrial Ecology

Spatial, Spectral and Temporal Requirements for Improved Hyperspectral Mapping of Plant Functional Type, Plant Species, Canopy Biophysics, and Canopy Biochemistry

  • Collaborators: Philip E. Dennison (University of Utah), Susan Ustin (UC Davis), Ray Kokaly (USGS) for gulf work

    • Funding: NASA Terrestrial Ecology program

  • Imaging spectrometry has considerable potential for mapping plant functional types and plant species. However, the ability to discriminate plants will vary depending upon the wavelengths sampled, when spectra are collected within a season and the spatial resolution of the sensor. This information is critical to the success of future imaging spectrometry missions such as HyspIRI.

  • In this project we have developed a standardized methodology for developing statistically robust spectral libraries and identifying subsets of spectra that best discriminate plants at the level of species and plant functional types from imaging spectrometry data. This approach is being applied to a diversity of ecosystems across the United States, including, most recently oil-impacted wetlands in the Gulf of Mexico. Across all ecosystems we are evaluating spatial and spectral requirements for species-discrimination. Locally, in Santa Barbara, we are evaluating temporal requirements using AVIRIS data acquired multiple times over the years and field spectra.

Fire

Remote Sensing of Pre-fire Fuels, Fire Danger and Post-fire Vegetation Response

  • Collaborators: Philip E. Dennison (University of Utah), Ted Eckmann (Bowling Green University), Phillip Schneider (JPL), Bodo Bookhagen (UCSB), Jean Carlson (UCSB Physics).

    • Funding: NASA Terrestrial Ecology, NASA Regional Earth Science Applications Center (RESAC) program, NASA Earth System Science Fellowships (three), Joint Fire Science Program, National Science Foundation.

  • Remote sensing has considerable potential for aiding in the study of wildfire. In this area, we have focused on several elements, including mapping pre-fire fuel properties, current fire danger, active fire properties and post-fire response. We have also developed our own fire spread model, HFIRE.

    • In this project, pre-fire fuel properties were mapped using AVIRIS, MODIS and Landsat including live fuel moisture, fuel type and fuel condition (ratio of green to senesced plant materials). Fire danger was assessed primarily by developing a VARI-based version of the Fire Potential Index (FPI) using MODIS. Active fire temperature and area were mapped using multiple endmember spectral mixture analysis applied to AVIRIS, ASTER and MODIS data. Post-fire response is currently on-going using AVIRIS time series to evaluate the impacts of the GAP, TEA and JESUSITA fires. One key question will be to evaluate how live fuel moisture may have impacted fire spread, which will be evaluated using HFIRE.

Endmember fraction images (RGB=GV, NPV, and Soil) for Pre- and Post- Jesusita Fire. The large expanses of blue in the lower image represent formerly chaparral areas with near complete combustion.

Methane

Remote-Sensing Methane Emissions: Field Validation with Seepage from Marine, Urban, and Submerged-City Sources

Roberts, D.A., Bradley, E., Cheung, R., Leifer, I., Dennison, P.E., & Margolis, J. (2010). Mapping methane emissions from a marine geological seep source using imaging spectrometry. Remote Sensing of Environment, 114, 592-606.

    • Collaborators: Ira Leifer (UCSB), Philip Dennison (University of Utah), Jack Margolis.

    • Funding: NASA NACP

    • The relative source contributions of the important greenhouse gas methane, CH4, have high uncertainty and there is a need for expedient and spatially complete local-scale characterization for climate research and inventories.

    • In this project, we utilized the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) to map methane emissions from concentrated sources: the Coal Oil Point (COP) marine seep fields, offshore from Santa Barbara, California and the La Brea Tar Pits in Los Angeles, CA. The COP seep field is a heterogeneous, sonar quantifiable, strong methane source (0.015 Tg yr-1 from ~3 km2), while the La Brea region is known for natural seepage of oil and methane.

Land use and land cover change in Rondonia, Brazil

This project is part of a collaborative research effort between UCSB, Salisbury University (Maryland) and North Carolina State University. Under the direction of principal investigators Jill Caviglia-Harris,

Daniel Harris (Salisbury), Erin Sills (NCSU) and Dar Roberts (UCSB), we have been active in the production and management of a gapless annual Landsat imagery archive (1984-2009) covering the majority of

the state of Rondonia. The state, along with neighboring Acre and

Mato Grosso, constitutes part of the southern "arc of deforestation"

and has undergone extensive and rapid transformation since the

1970's. The project, "Living with Deforestation: Analyzing Transformations in Welfare and Land Use on an Old Amazonian Frontier" is focused on the use of longitudinal household panel survey data which have been collected in 1996, 2000, 2005 and 2009. Our principal contribution is furnishing the researchers with spatially

explicit data on land-cover and agriculture for analysis of land-use change drivers and the long-term impacts of deforestation on household welfare.

Landsat-derived classification of the state of Rondonia, Brazil, where dark green is forest, pink is pasture, red is rock, blue is water, and light green is secondary forest. Click on the image to see the time-series.

Education

IDEAS (Innovative Datasets for Environmental Analysis by Students)

Roberts, D.A., Bradley, E., Roth, K.,Eckmann, T., & Still, C. (Accepted pending rev.). Linking physical geography education and research through the development of an environmental sensing network. Journal of Geoscience Education

  • www.geog.ucsb.edu/ideas

    • Collaborators: Dr. Chris Still (UCSB), Dr. Oliver Chadwick (UCSB), Dr. Ted Eckmann (Bowling Green University), Dr. Doug Fischer, SBARC, MCA, Dr. Robert West (ELAC)

    • Funding: NSF CCLI-I program

    • The IDEAS project includes four automated meteorological stations deployed in Santa Barbara County to capture the diverse environmental conditions and plant ecosystems of the region (coastal plain, grassland, oak woodland). The website provides access to real-time and archived data and webcam images, along with many supporting materials.Several undergraduate classes at UCSB have used this website, and we invite you to use it in your classes or research as well.

    • IDEAS also led to a follow on project: PanOpt, which is a website for visualization and exploration of remote sensing time-series. PanOpt integrates webcam, GOES, and MODIS satellite imagery.