Forest Cover and Height in Topographically Complex Landscapes from MISR Assessed with High Quality Reference Data

Mark James Chopping, Montclair State University, chopping@pegasus.montclair.edu (Presenter)
Malcolm North, USFS Pacific Southwest Research Station, mnorth@ucdavis.edu
Jiquan Chen, University of Toledo, jiquan.chen@utoledo.edu
Crystal Schaaf, Environmental Earth and Ocean Sciences, schaaf@bu.edu
John V Martonchik, NASA JPL, john.v.martonchik@jpl.nasa.gov
Michael Bull, NASA JPL, michael.a.bull@jpl.nasa.gov
Bryan Blair, NASA GSFC, james.b.blair@nasa.gov
Michelle Hofton, University of Maryland, Geography Dept., mhofton@geog.umd.edu

Reflectance data from the NASA Multiangle Imaging SpectroRadiometer (MISR) can be used to invert geometric-optical (GO) models to provide spatially contiguous maps of forest mean top-of-canopy height and fractional cover. Earlier work showed that when this approach was applied over the entire southwestern US, anomalous results were sometimes obtained with cover overestimated and height underestimated. These poor retrievals may be owing to model inadequacy, poor estimates of the background bidirectional reflectance distribution function (BRDF), and/or problems with the inversion protocol. In this study, we use high quality reference data to investigate these issues and perform retrievals of forest crown cover and canopy height for a large study area in the Sierra Nevada National Forest in California, an area of considerable topographic and understory complexity. Forest crown cover and mean canopy height values were obtained through adjustment of a modified version of a simple GO model against red band surface reflectance estimates from MISR mapped onto a 250 m grid, with the soil-understory background contribution predicted a priori for each location via regression relationships with the isotropic, geometric, and volume scattering kernel weights of a BRDF model and MISR nadir camera green and near-infrared BRFs. The GO model used in this study differs from the previous version in that shaded components are no longer discarded, the Ross volume scattering function is replaced by a simple leaf reflectance; and the 3-parameter RossThick-LiSparse BRDF model replaces the 4-parameter modified Walthall model. The cover retrievals were assessed using estimates derived from analysis of QuickBird 0.6 m spatial resolution panchromatic imagery using the CANAPI algorithm, while the height retrievals were assessed using canopy height metrics derived from the NASA Laser Vegetation Imaging Sensor (LVIS). These data, together with the predicted background BRDFs, were also used to drive the model in forward mode for the MISR viewing and illumination angles. Comparison of the model-predicted and MISR BRF patterns for 1048 pixels provided a distribution in which 99.9% have RMSE < 0.02 and 98.4% have R^2 > 0.8. The GO model inversion results provided R2 values of 0.55 and 0.50 for cover and height, respectively, with RMSE values of 0.11 and 6.8 m, for the 1048 MISR grid cells that have LVIS RH100 (range: 7.8 - 59.3 m) and CANAPI fractional cover (range: 0.02 - 0.47) estimates. This study shows that MISR can be used for mapping forest canopy cover and height over areas with severe and complex topography and highly variable backgrounds.

Presentation: 2011_Poster_Chopping_143_272.pdf (2867k)

Presentation Type:  Poster

Session:  Other   (Tue 11:30 AM)

Associated Project(s): 

• Chopping, Mark: A Decade of Changes in Aboveground Live Standing Dry Biomass, Canopy Cover, Height, and Understory Density in the Southwestern United States from EOS MISR and MODIS ...details
• Chopping, Mark: A new approach for mapping woody plants in the southwestern United States using NASA Earth Observing System Data ...details

Poster Location ID: 143