Close Window

CD-32 Abstract

Carbon Balance in Amazon Forests from Site to Region: Integrating Remote Sensing from Satellites and Aircraft with Ground-based Tower and Biometric Data

Humberto Rocha — USP (Universidade de Sao Paulo) (SA-PI)
Scott Reid Saleska — University of Arizona (US-PI)

Objective: to link remote-sensing indices to flux tower and forest plot datasets to derive empirical predictions and test models of carbon dynamics at large scales. We focus on:

     1. indices of seasonal variation in driving variables (e.g. sunlight and precipitation), and in the response of key processes (e.g. carbon fluxes and foliar activity);

     2. indices of forest structure and disturbance history such as canopy height, canopy gap-size distribution, and stocks of coarse wood debris; and,

     3. a hypothesized relation between seasonality and structure.

We focus on seasonality because ecosystem response to seasonal forcing provides a window onto mechanisms that also control long-term responses, but recent work (Saleska et al., 2003; Goulden et al., 2004) shows that such mechanisms are still not well understood. We focus on forest structure (a correlate of disturbance history) because recent work (Moorcroft et al., 2001; Saleska et al., 2003) shows that although carbon balance is acutely sensitive to disturbance history, the distribution of forest disturbance states across larger scales is not well known.



Methods: (A) Satellite Remote Sensing: We will use 5+ years of data from MODIS, focusing on the Enhanced Vegetation Index, EVI (canopy greenness) and on Land Surface Water Index, LSWI (canopy water) to reveal seasonal patterns in foliar activity. (B) Integration of remote sensing with ground studies at eddy flux towers, including: (1) Flux data: We will use eddy flux data from 12 Amazonian towers which sample primary forest, transitional forest, and converted lands to calibrate MODIS indices to predict basin-wide GPP carbon fluxes; and (2) forest plots: We will integrate forest plot data (biomass, growth rates) with forest structure from coincident ground-based LIDAR surveys. (C) Modeling. We will use LIDAR data to constrain Ecosystem Demography (ED) model simulations of Amazonian carbon balance to test against flux observations.



Significance: This submission proposes an innovative plan to link local to regional scale measurements in forests and converted lands to rigorously test remote-sensing based predictions of Amazonian carbon dynamics, an LBA priority topic. More generally, this work advances NASA’s national objective to Study the Earth system from space [by testing space-derived predictions of earth system processes].

Close Window