Boundary Values for Regional to Continental Scale Greenhouse Gas Flux Estimation

Abstract:

Errors in prescribed boundary values can bias estimates of surface fluxes in data-assimilation and inverse modeling studies of regional greenhouse gas budgets. Sensitivity to boundary value errors is particularly important for CO₂, since strong seasonally opposing fluxes result in comparatively small net annual uptake.

We have developed empirical boundary value products for North America for CO₂, CH₄, N₂O and other long-lived gases using data from aircraft profiles and marine boundary layer sites in NOAA’s Global Greenhouse Gas Reference Network. The influence of each aircraft sample is mapped forward and backward from the measurement location using trajectories generated with NOAA’s HYSPLIT model driven by meteorological fields from the North American Regional Reanalysis system. These data and influence functions are used to create free-tropospheric reference surfaces that describe monthly scale variability as a function of longitude, latitude, altitude, and time. Data from remote marine boundary layer sites are used to generate Atlantic and Pacific marine boundary layer reference surfaces that vary with latitude, altitude and time. Taken together, the free-troposphere and marine boundary layer reference surfaces provide 4-dimensional boundary values for the continent. This product has been significantly improved compared to earlier versions used in several published studies.

We have also developed a related framework for simultaneous optimization of boundary values and surface fluxes in the NOAA CarbonTracker-Lagrange regional inverse modeling system, which uses surface and boundary value footprints from the WRF-STILT model. In this case, we adjust a prior estimate for the boundary values such as can be obtained from the global Eulerian CarbonTracker modeling system or another global model. Vertically resolved data from aircraft and/or from a combination of surface and column measurements are needed to reliably separate surface and boundary influences.

We will describe both methods for boundary value estimation and our efforts to investigate errors using synthetic data. We will also consider the extent to which currently available data enable separate estimation of boundary and surface fluxes and what additional measurements would be most helpful.