Hot Spots and Hot Moments of Wetland Methane Emissions in the San Francisco Bay Sacramento San Joaquin River Delta: Detection and Prediction with a Data-Driven Model

Restored and managed wetlands in the San Francisco Bay Sacramento San Joaquin River Delta (Bay-Delta) exhibit strong biogeochemical heterogeneity, which results in elevated methane flux at certain times and in certain locations - termed 'hot spots and hot moments' (HSHM). These instances and locations can experience biogeochemical rates so high they can disproportionally contribute to annual flux rates. While instances of heightened emissions are captured by eddy covariance estimates of annual methane budgets in Bay-Delta wetlands, it is unclear 1) what triggers these hot moments 2) whether or not specific locations within each wetland are more prone to being hot spots of methane emissions, and 3) if these hot spots contribute outsize emissions to annual methane budgets. Identifying HSHM of methane emissions, further resolving Bay-Delta methane balances, and identifying potential greenhouse gas tradeoffs associated with wetland restoration is of particular interest in the Bay-Delta, where there are plans to convert 200,000 acres of farmland into managed wetlands or carbon farms. To address this gap, we synthesized methane flux data monitored by eddy covariance systems and associated flux footprints from seven wetland sites (35 site-years of data) in the Bay-Delta to identify instances and locations that experience elevated methane flux statistically different from baseline methane flux (i.e., HSHM of methane emissions). Preliminary results indicate HSHM exert an outsize effect on annual methane flux rates relative to their occurrence at each study site. To model and determine the best predictors of HSHM of methane emissions, we trained a random forest regression using methane flux, biometeorological, climatic, and remote sensing data to reproduce and predict HSHM of methane emissions on an hourly interval. We thank FLUXNET-CH4 and AmeriFlux contributors Dr. Patty Oikawa, Dr. Dennis Baldocchi, and Dr. Brian Bergamaschi for providing the data used in these analyses.