Earth Institute Research Projects

Use of GRACE to estimate terrestrial carbon fluxes

Lead PI: Dr. Pushker A. Kharecha

Unit Affiliation: Center for Climate Systems Research (CCSR)

September 2010 - August 2013
South America ; Amazonia ; Brazil ; New York
Project Type: Research

DESCRIPTION: The ultimate goal of this research was to analyze GRACE data and try to detect a signal from biomass changes in the Amazon. In areas of high deforestation, it was expected that the gravity signal from biomass loss should be detectable using hi-resolution GRACE data developed at NASA-GSFC, given that the signal would occur over a sufficiently large area and would be secular (non-cyclical). Although it was also expected that such a signal would be relatively small and therefore difficult to distinguish from substantially larger ones (notably hydrology-related mass changes), it was nonetheless deemed worthy of investigation given that, among other things, it could result in a novel observational dataset which could provide very useful constraints on a variety of topics in carbon cycle and ecosystem science. A subset of a globally gridded dataset representing a ~7-year period (2003-2010) for the Brazilian Amazon forest was used for the analysis.

OUTCOMES: Contrary to expectations, the trend in gravity signal is upward, i.e. these particular regions of the Amazon forest have evidently gained mass over the observation period. Furthermore, the magnitude of this trend is ~15-30 times larger than the expected trend from biomass loss. This is most likely due to changes in hydrology in these areas (based on previous studies showing large variability in Amazon water storage over similar time periods). Removing the hydrology signal thus requires either accuracy to within a few percent in hydrology models and/or field observations, or a longer GRACE time series combined with reasonably accurate models/observations. As neither of these is likely anytime soon, unfortunately, it is therefore apparent that the gravity signal from biomass loss in these data is, at the moment, not distinguishable from more dominant signals. In light of this null result, continued funding for this grant beyond the first year was no longer requested.