Evidence of a stratospheric methane bias in the IFS against MIPAS data

In this report, we compare simulated methane fields from the Integrated Forecasting System (IFS) with retrieved stratospheric methane profiles retrieved from the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS). We assume that the used MIPAS retrieved methane profiles are the best estimate of the true atmospheric methane state in the stratosphere and we use them as the reference.

Simulated methane fields are provided by IFS in forecast mode with two different chemical mechanisms. Any differences between simulated methane and MIPAS retrievals could be attributed to following sources of error: (i) IFS transport, (ii) simplified methane chemistry, (iii) initial condition of the simulation, (iv) surface fluxes and (v) representativity error in the comparison. We believe that the representativity error (v) is of second order compared to the other errors. We also show that the initial condition (iii) is not the main driver of the difference.

The first main feature of the simulated methane from all experiments is an underestimation in the upper troposphere – lower stratosphere (UTLS) region where the tropopause acts like a barrier for the stratosphere – troposphere exchange. This underestimation of methane concentration in the model could be associated with a transport error (i) across the tropopause barrier or an underestimation in the tropospheric concentration linked with errors in surface fluxes (iv). This second hypothesis is nevertheless unlikely since the feature appears in simulations with different surface fluxes.

The second feature is a large underestimation around 10 hPa at high latitudes, stronger in the southern hemisphere during the summer season when the chemistry is the most active. This underestimation could be linked with (ii) errors in the methane chemical mechanisms. This is also the location of the surf zone where the planetary waves break in the winter, a phenomenon that could be at the origin of a descent of this methane-poor air-mass in the lower stratosphere after the summer season. This amplifies the methane underestimation in the UTLS.

The methane underestimation in the UTLS is opposite to the known overestimation of humidity, another tracer in the IFS. This humidity overestimation around 200 hPa, mainly in the summer hemisphere, produces a cold stratospheric bias persistent in the model. The stratospheric task force, an ECMWF cross-section initiative, aims to tackle this issue.

We demonstrate in this document that MIPAS methane retrievals therefore provide another insight of the model biases in various regions of the stratosphere. We claim that the model biases are likely linked to errors in the IFS transport (i) and to a less extend and more locally to an simplified methane chemistry in the IFS (ii). Methane simulations from IFS could be used to test various model configurations that impact the stratospheric transport and help improving the IFS stratospheric forecast in the future.