Welch Bostick1, James W. Jones1, Wendy Graham1, and Sibiry Traore2. (1) University of Florida, 320 University Village South #3, Dept. of Ag. and Biological Eng., Gainesville, FL 32603, United States of America, (2) ICRISAT, ICRISAT Samanko, P.O. Box 320, Bamako, Mali
Concern
about the effect of anthropogenic greenhouse gas (GHG) emissions on climate has
led to interest in sequestration of carbon in terrestrial sinks, including
soils. Industrialized nations with emission reduction commitments under the
Kyoto Protocol (Annex I countries) can use soil organic carbon (SOC)
sequestration to reduce their GHG emissions. However, a major problem with this
mechanism is the high uncertainty of SOC estimates over the large,
heterogeneous areas needed to sequester carbon in quantities that are
significant relative to CO2 emissions. A methodology that aims to
improve estimates of SOC changes over large areas was developed and tested. The
methodology uses Monte Carlo simulation to estimate SOC change over time at
multiple locations and assimilates SOC measurements with these estimates using
the Ensemble Kalman Filter algorithm (EnKF). The methodology was tested for
estimating SOC changes in the farming community of Oumaroubougou, Mali
(Latitude: 12.19 °N, Longitude: 5.14 °W). The performance of the methodology
was evaluated over a range of specifications for the uncertainty of model
parameters and the initial spatial correlation of SOC and model parameters
between plots in the experiment. In all cases, EnKF estimates of changes in SOC
had lower variances than measurement-based estimates. Specification of the
initial spatial correlation of SOC and the model rate parameter, versus
assuming no spatial correlation, gave more power to the EnKF when measurements
were assimilated. Although further research and testing at larger scales is
needed, the methodology appears promising for giving improved estimates of SOC
changes with decreased uncertainty.