This report presents the second 9-km global nature run of the CO₂ Human Emission (CHE) project – hereafter referred to as Tier 2 nature run -- with improved transport and emissions. The main purpose of this simulation is to provide a reference global simulation to be performed at the higher operational resolution (9km) as part of the CHE library of simulations. This library of simulations will provide data for observing system simulation experiments (OSSEs) to the CHE consortium and wider scientific community. The configuration of the Tier 2 nature run is based on the Copernicus Atmosphere Monitoring Service (CAMS) CO₂ forecast using the Integrated Forecasting System (IFS) at the European Centre for Medium-range Weather Forecasts (ECMWF), but with improved transport and emissions. The main differences with the Tier 1 CHE nature run are (i) the meteorological analysis, (ii) the model transport is based on the latest version of the NWP model at ECMWF, (iii) the anthropogenic emissions which have upgraded to the latest available EDGARv4.3.2FT2015, (iv) the ocean fluxes with the SOCAT-based ocean fluxes from Jena-Carboscope and (v) a revised bias correction of Net Ecosystem Exchange (NEE).

The meteorological aspects of the nature run have not been evaluated in this report because they are consistent with the ECMWF analysis and short-range forecasts which have been extensively investigated and evaluated in various ECMWF Technical Memoranda.

This report illustrates the capability of the nature run to represent the variability of CO₂ at different scales from seasonal and the inter-hemispheric gradient to regional/local variability of synoptic weather systems and plumes from emission hotspots and the diurnal cycle. Comparison with in situ and total column data shows a realistic variability of CO₂. The systematic errors are in the range of 1 to 2ppm for the total column on monthly timescales and less than 1ppm on global scales at baseline sites. These systematic errors are associated in large part to the prescribed and modelled surface fluxes which are not constrained by observations. The seasonal cycle, synoptic and diurnal cycle are all within the range of observed variability recorded by surface and total column observations. A preliminary evaluation of column-averaged CH₄ and CO also show a realistic representation of variability at synoptic and diurnal time-sales. In addition to the upgrades in the 9-km resolution nature run, the Tier 2 simulations have been performed at lower resolution using an ensemble approach to include information on uncertainties in the fluxes and transport.