Kellndorfer, Josef: Earth Big Data, LLC (Project Lead)
Goetz, Scott: Northern Arizona University (Co-Investigator)
Walker, Wayne: Woodwell Climate Research Center (Co-Investigator)
Dubayah, Ralph: University of Maryland (Participant)
Gonzalez, Sergio: Forestal Arauco S.A. (Participant)
Rombach, Markus: Digimapas Chile Aerofotogrametria Ltda. (Participant)
Project Funding:
2008 - 2010
NRA: 2007 NASA: Terrestrial Ecology
Funded by NASA
Abstract:
The overall
goal of the proposed research is to investigate technological options
and data fusion algorithms for ecosystem structure measurements from
the proposed NASA DESDynI mission. The experiments described in this
proposal are possible given the timely confluence of highly relevant
remote sensing and ground reference data sets. Through collaboration
with the leading geospatial information company in Chile, Digimapas
Chile, we will have access to 75,000 km2 of 1-meter resolution
full-waveform small footprint lidar (SFPL) data and 0.5 m resolution
digital orthophoto imagery covering the commercial forests of Arauco,
one of the largest cellulose producers in Latin America. Arauco is a
collaborator on this proposal and will provide access to relevant
timber survey data, which is regularly acquired across their land
holdings. The SFPL acquisitions commenced in October of 2006 and are
scheduled to be completed in mid-2008. The area covered with the SFPL
data has also been mapped by the ALOS/PALSAR at several resolutions and
acquisition modes. In addition, multi-spectral optical imagery from
ALOS/AVNIR-2 and CBERS-2, as well as lidar data from ICESat/GLAS are
available for use. All remote sensing data will have been acquired
during a very narrow time frame spanning less than two years.
Given the size
of the study area (75,000 km2), the availability of very high
resolution lidar and optical imagery, and the dense network of field
reference data, very realistic simulations of spaceborne lidar sampling
design covering multi-scene radar and optical imagery can be conducted.
In turn, these simulations provide an ideal framework within which to
test various approaches to DESDynI lidar/radar data fusion leading to
canopy height and aboveground biomass retrievals. Data fusion will be
accomplished using randomForest, a state-of-the art statistical
modeling approach based on ensemble machine learning techniques.
Four specific objectives are proposed to achieve this overall goal:
(1) Evaluate multi-sensor data-fusion strategies for canopy height (CH) and aboveground biomass (AB) retrieval.
(2) Study the effect of lidar sampling density on the accuracy of CH and AB predictions.
(3) Study the effect of lidar footprint size on the accuracy of CH and AB predictions.
(4) Study the effect of radar resolution on the accuracy of CH and AB predictions.
One of several
key DESDynI mission science questions involves the performance of
off-nadir multi-beam lidar data for CH and AB retrieval. As part of the
proposed research, approximately 300 km2 of Arauco forest holdings will
be remapped with Digimapas Chile’s SFPL at off-nadir angles of up to
10o. This acquisition will facilitate an investigation into how the
off-nadir pointing affects the lidar waveforms and hence the accuracy
of CH and AB predictions.
The proposed
work addresses the following research needs identified in the DESDynI
Workshop Report (2007): (1) studies to develop and evaluate algorithms
and analysis strategies that address the merger of lidar and radar
measurements (data fusion), and (2) studies to quantify the effects of
sampling design and measurement accuracy, frequency, and resolution on
the ability to improve our quantitative knowledge of global carbon
dynamics and ecosystem structure and function. This focused study was
designed as a two-year project so as to make results available as early
as possible for the purposes of informing the development of the
DESDynI mission concept.
2013 NASA Terrestrial Ecology Science Team Meeting Poster(s)
- Update of the National Biomass and Carbon Dataset 2000 using ALOS PALSAR L-band
-- (Josef Kellndorfer, Oliver Cartus, Wayne Walker)
[abstract]
More details may be found in the following project profile(s):