|
Detection of Changes of Tropical Forest Biomass from Fusion of Multi-temporal Radar and Lidar Data
Sassan
Saatchi, CALTECH/JPL, sassan.saatchi@jpl.nasa.gov
(Presenter)
Uncertainties associated with the magnitude and distribution of the residual component of the terrestrial carbon sink is large (15-30% of global emissions from fossil fuels). It is assumed that this sink is largely sequestered in aboveground biomass and is associated with the post-disturbance recovery of forests modulated with climate change (warming) and potential fertilization by rising atmospheric CO2. To resolve this uncertainty in the global carbon cycle, the detection of changes of aboveground biomass on the landscape scale (1-10 km) is considered one the primary products of existing and future remote sensing techniques. In this paper, we examine how existing data and methods can potential provide this product. We use multi-temporal data from a combination of L-band radar (AIRSAR and ALOS) and small footprint lidar to estimate and to model changes of forest biomass over the tropical wet forest of La Selva in Costa Rica. We present the study in three steps: 1. Radar change detection using ALOS data (2007-2010) and comparison with AIRSAR L-band data (2004). 2. Changes observed from Lidar data (1996-2009), 3. Modeling changes over the landscape using the fusion of lidar and radar data. The results show that L-band radar is capable to detect changes of secondary forest biomass up to 100 tons/ha qualitatively (positive or negative change). Lidar data can provide changes quantitatively, however with large uncertainty (up to 50%). However, the fusion method using a Baysian approach improves the biomass change detection significantly at resolutions greater than 250 m (6.25 ha). Presentation Type: Poster Session: Other (Tue 11:30 AM) Associated Project(s):
Poster Location ID: 279
|