Modeling natural wetlands: A new global framework built on wetland observations

Thursday, 17 December 2015
Poster Hall (Moscone South)
Elaine Matthews, NASA, Goddard Institute for Space Studies, New York, NY, United States; NASA, Ames Research Center, Moffett Field, CA, United States, Joy Romanski, Columbia University/NASA Goddard Institute for Space Studies, New York, NY, United States and David Olefeldt, University of Alberta, Edmonton, AB, Canada
Natural wetlands are the world's largest methane (CH4) source, and their distribution and CH4 fluxes are sensitive to interannual and longer-term climate variations. Wetland distributions used in wetland-CH4 models diverge widely, and these geographic differences contribute substantially to large variations in magnitude, seasonality and distribution of modeled methane fluxes. Modeling wetland type and distribution—closely tied to simulating CH4 emissions—is a high priority, particularly for studies of wetlands and CH4 dynamics under past and future climates.

Methane-wetland models either prescribe or simulate methane-producing areas (aka wetlands) and both approaches result in predictable over- and under-estimates. 1) Monthly satellite-derived inundation data include flooded areas that are not wetlands (e.g., lakes, reservoirs, and rivers), and do not identify non-flooded wetlands. 2) Models simulating methane-producing areas overwhelmingly rely on modeled soil moisture, systematically over-estimating total global area, with regional over- and under-estimates, while schemes to model soil-moisture typically cannot account for positive water tables (i.e., flooding). Interestingly, while these distinct hydrological approaches to identify wetlands are complementary, merging them does not provide critical data needed to model wetlands for methane studies.

We present a new integrated framework for modeling wetlands, and ultimately their methane emissions, that exploits the extensive body of data and information on wetlands. The foundation of the approach is an existing global gridded data set comprising all and only wetlands, including vegetation information. This data set is augmented with data inter alia on climate, inundation dynamics, soil type and soil carbon, permafrost, active-layer depth, growth form, and species composition. We investigate this enhanced wetland data set to identify which variables best explain occurrence and characteristics of observed wetland ecosystems. The novelty of the new approach is that it starts from what we know about wetlands, builds ecosystem-specific models from these observations, and avoids known biases in current hydrology-based approaches to wetland definition in methane models.