Data Assimilation of Forest Ecology from Modis Parallax
Adam
Wolf, Carnegie Institution, Dept of Global Ecology, adamwolf@stanford.edu
(Presenting)
Joseph
A
Berry, Carnegie Institution, Dept of Global Ecology, joeberry@stanford.edu
Greg
P
Asner, Carnegie Institution, Dept of Global Ecology, gpa@stanford.edu
Monitoring the state and change of land cover globally is important for predicting the carbon cycle and climate, short term weather prediction, and identifying conservation priorities. Global observations of NDVI give insight into green leaf area and thus photosynthesis, but are uninformative of forest biomass or respiration. Fine-scale observations using lidar, radar, or imaging spectrometry give information on forest structure and composition, but are not available on the large spatial and long temporal scales that are relevant to global ecology.
This poster summarizes our work on the assimilation of forest structure - the number, shape, and spatial arrangement of trees - into land surface models from daily moderate resolution satellite observations. Paired differences of sequential MODIS observations from different view angles are shown to be largely sensitive to forest structure and not from leaf or stem area. These forest structural attributes are highly correlated with forest age, biomass, and other attributes which define the position of a forest within its succession trajectory. A separate study shows that most land surface models are unable to accomodate information on forest structure because they largely ignore forest population biology in defining growth, allocation, and mortality. These carbon fluxes which determine the carbon balance of a forest are likewise highly correlated with forest structure - the size and number of trees. Finally, a ray-tracing approach is presented for improved depiction of geometric optics of forest canopies that emerges as a consequence of competition for space. Together, these model refinements permit a data assimilation system that translates remote sensing observations into estimates of forest ecological attributes in improved land surface models.
Presentation Type: Poster
Poster Session: Ecosystems Science
NASA TE Funded Awards Represented:
NONE: Related Activity or Previously Funded TE Award