Time-windows-based filtering method for near-surface detection of leakage from geologic carbon sequestration sites

We use process-based modeling techniques to characterize the temporal features of natural biologically controlled surface CO{sub 2} fluxes and the relationships between the assimilation and respiration fluxes. Based on these analyses, we develop a signal-enhancing technique that combines a novel time-window splitting scheme, a simple median filtering, and an appropriate scaling method to detect potential signals of leakage of CO{sub 2} from geologic carbon sequestration sites from within datasets of net near-surface CO{sub 2} flux measurements. The technique can be directly applied to measured data and does not require subjective gap filling or data-smoothing preprocessing. Preliminary application of the new method to flux measurements from a CO{sub 2} shallow-release experiment appears promising for detecting a leakage signal relative to background variability. The leakage index of ?2 was found to span the range of biological variability for various ecosystems as determined by observing CO{sub 2} flux data at various control sites for a number of years