Mapping Canopy Structure & C Pools in the SW US with Multiangle EOS Data

Background

Mapping Canopy Structure & Biomass in the SW US with Multiangle EOS Data

NASA Award NNX08AE71G: A New Approach for Mapping Woody Plants in the Southwestern United States Using NASA Earth Observing System Data was selected for funding by NASA in late 2007, with the project starting on January 2, 2008.
The goal is to apply multi-angle remote sensing methods to map woody plant crown cover, mean canopy height, aboveground woody biomass, and understory density over large areas in the southwestern United States, for landscapes in which either shrubs or trees are abundant. A large part of the effort will be devoted to improving retrievals of woody plant canopy parameters via GO model inversion against MISR and MODIS data -- i.e., by accessing the structural signal embedded in these multiangle data -- and to obtaining adequate calibration and reference data.
The work will involve expansion of a data base of validation data derived from high resolution imagery from the IKONOS and QuickBird satellites, as well as data from other sources (FS-FIA, SWReGAP, NLCD, VCF). It will also use canopy metrics from from NASA's LVIS full waveform lidar instrument, USDA, ARS Jornada Experimental Range UAV hyperspatial imagery, and field data.
Go to Project website
Summary: We propose to use data from the NASA Earth Observing System (EOS) Multi-angle Imaging SpectroRadiometer (MISR) and Moderate Resolution Imaging Spectroradiometer (MODIS) with reflectance modeling techniques to map woody plant cover, mean canopy height, and understory foliage density at large scales in the southwestern US. Unlike current maps, our products will include shrub and tree crown cover, mean canopy height, and understory density. This is possible because the method we have developed accesses the structural information captured in EOS data. We will validate these new products intensively at four established, data-rich sites; and extensively throughout the region using data from both in-house and external sources. Shrubs are an invasive species whose proliferation has important socio-economic implications through the loss of herbaceous productivity: shrub encroachment may affect up to 84% of current and former U.S. grasslands (Gori and Enquist 2003); however the rate and extent of woody encroachment remain unknown (Pacala et al. 2001). The forests of the southwestern US have experienced increased fire frequencies, extents, and durations in recent years with shifts to earlier snowmelt and longer, hotter, and drier summers (Running 2006); knowledge of forest structure is thus critical. In our recent work we have shown that geometric-optical models can provide good estimates of woody canopy metrics even where the soil-understory background is heterogeneous. These techniques will be implemented over large areas including desert grassland, woodland and forested areas. Successful tests have covered extensive areas in S.E. Arizona and S. New Mexico: there is good agreement with the spatial distribution of tree cover from the MODIS Vegetation Continuous Fields % Tree Cover product; with USDA Forest Service Forest Inventory Analysis crown cover and canopy height maps; and with shrub fractional cover estimates derived from very high resolution (1 m) images using image segmentation techniques.

CinSWUS: Mapping C Pools in the SW US with Multiangle EOS Data

December 2003: proposal to map C in desert grasslands of the SW U.S. using EOS multiangle data selected as part of NASA's Earth Science Enterprise program. See the NASA Land Cover Land Use Change Program web site (see here). The project officially began on June 1, 2004, with funding for three years (recently extended to May 2008).
Primary instrument: the Multiangle Imaging SpectroRadiometer (MISR) ...combining these explicitly multiangular data with those from MODIS, an across-track scanning spectroradiometer.
Principal Investigator: Mark J. Chopping, Co-Investigators: Lihong Su (Earth and Environmental Studies, Montclair State University, Upper Montclair, NJ), Albert Rango, Debra P.C. Peters (USDA/ARS Jornada Experimental Range, Las Cruces, NM), John V. Martonchik (NASA/JPL, Pasadena, CA). Collaborator (from April 17, 2006): Dr. Andrea Laliberte, Rangeland Remote Sensing Scientist USDA-ARS, Jornada Experimental Range. Dr. Lihong Su was appointed as Research Associate at Montclair State University and started with the project on July 12, 2004. He was assigned the role of Co-Investigator. The first project meeting was held on October 28, 2004, at the Jornada Experimental Range, Las Cruces, NM. Since then we have presented research results at NASA Land Cover Land Use Change (3) and MISR Science Team (3) meetings; various conferences (AGU, ISPMSRS'05, IGARSS'06, AAG'05, CEOS Global Vegetation Workshop '06, NACP January '07, IGARSS'07, Veg3D & Biomass '08); and submitted a number of manuscripts for publication. The project has also employed two graduate students and one undergraduate in work to collate a validation database using high resolution imagery (primarily Ikonos 1 m panchromatic images).
Abstract (PDF). Summary: In tandem with the ongoing world-wide increase in the abundance of woody plants within former grasslands, desert grasslands throughout the southwestern U.S. have experienced a dramatic increase in the abundance of shrubs since the end of the 19th century. The region thus provides an adequate subject for the development of remote sensing and modeling methods which will be useful at global scales and in other regions. We propose to estimate carbon pools in this region using remotely-sensed inputs from the NASA Earth Observing System satellites Terra and Aqua. We propose to refine mapping of vegetation through improved plant community type differentiation, estimating contributions from soil, shrub and grass components and provision of structural measures. This addresses three aspects of the problem of estimating C pools: large differences in total C storage in different plant communities; the need to reliably estimate the areal proportions of soil, grass and shrubs; and the need to obtain measures of canopy structure over large areas. We will accomplish this using multi-angle spectral reflectance data and derived parameters and metrics from the MISR and MODIS instruments. The approach is based on the premise that incorporating multi-angle measures in plant community type mapping results in substantial improvements in classification; that canopy heterogeneity is reflected in these data; and that useful canopy structure measures are available in the angular domain.