Quantifying Diurnal and Spatial Variations in CO2 Concentrations and Partial Columns using High-Resolution Global Model Simulations

Abstract:

Model-data fusion approaches, such as global inverse modeling for surface flux estimation, have traditionally been performed at spatial resolutions of several tens to a few hundreds of kilometers. Use of such coarse scales presents a fundamental limitation in reconciling the modeled field with both the atmospheric observations and the distribution of surface emissions and uptake. Emissions typically occur on small scales, including point sources (e.g. power plants, forest fires) or with inhomegeneous structure. Biological uptake can have spatial variations related to complex, diverse vegetation, etc. Atmospheric observations of CO₂ are either surface based, providing information at a single point, or space based with a finite-sized footprint. For instance, GOSAT and OCO-2 have footprint sizes of around 10km and proposed active sensors (such as ASCENDS) will likely have even finer footprints. One important aspect of reconciling models to measurements is the representativeness of the observation for the model field, and this depends on the generally unknown spatio-temporal variations of the CO₂ field around the measurement location and time. This work presents an assessment of the global spatio-temporal variations of the CO₂ field using the “7km GEOS-5 Nature Run” (7km-G5NR), which includes CO₂ emissions and uptake mapped to the finest possible resolution. Results are shown for surface CO₂ concentrations, total-column CO₂, and separate upper and lower tropospheric columns. Spatial variability is shown to be largest in regions with strong point sources and at night in regions with complex terrain, especially where biological processes dominate the local CO₂ fluxes, where the day-night differences are also most marked. The spatio-temporal variations are strongest for surface concentrations and for lower tropospheric CO₂. While these results are largely anticipated, these high resolution simulations provide quantitative estimates of the global nature of spatio-temporal CO₂ variability. Implications for characterizing representativeness of passive CO₂ observations will be discussed. Differences between daytime and nighttime structures will be considered in light of active CO₂ sensors. Finally, some possible limitations of the model will be highlighted, using some global 3-km simulations.