Comparison between the assimilation of IASI Level 2 ozone retrievals and Level 1 radiances in a chemical transport model

Emanuele Emili,Brice Barret,Daniel Cariolle,Eric Le Flochmoën

doi:10.5194/amt-12-3963-2019

Abstract

Abstract. The prior information used for Level 2 (L2) retrievals in the thermal infrared can influence the quality of the retrievals themselves and, therefore, their further assimilation in atmospheric composition models. In this study we evaluate the differences between assimilating L2 ozone profiles and Level 1 (L1) radiances from the Infrared Atmospheric Sounding Interferometer (IASI). We minimized potential differences between the two approaches by employing the same radiative transfer code (Radiative Transfer for TOVS, RTTOV) and a very similar setup for both the L2 retrievals (1D-Var) and the L1 assimilation (3D-Var). We computed hourly 3D-Var analyses assimilating L1 and L2 data in the chemical transport model MOCAGE and compared the resulting O3 fields among each other and against ozonesondes. We also evaluated the joint assimilation of limb measurements from the Microwave Limb Sounder (MLS) in combination with IASI to assess the impact of stratospheric O3 on tropospheric analyses. Results indicate that significant differences can arise between L2 and L1 assimilation, especially in regions where the L2 prior information is strongly biased (at low latitudes in this study). In these regions the L1 assimilation provides a better variability of the free-troposphere ozone column. L1 and L2 assimilation instead give very similar results at high latitudes, especially when MLS measurements are used to constrain the stratospheric O3 column. A critical analysis of the potential benefits and drawbacks of L1 assimilation is given in the conclusions. We also list remaining issues that are common to both the L1 and L2 approaches and that deserve further research.

Highlights

The global monitoring of atmospheric composition relies on a large number of dedicated satellite missions and on the sustained improvement of numerical forecast models
Results in the Southern Hemisphere (SH) (30–90◦ S) are again similar: a lower root mean square error (RMSE) than for the control simulation is found for both L1 radiances (L1a) and L2a in the upper and lower stratosphere
In this study we addressed the following question: what are the differences between the direct assimilation of Infrared Atmospheric Sounding Interferometer (IASI) radiances (Level 1) and the assimilation of Level 2 products for O3 analyses and reanalyses? We used an experimental setup in which differences between the L2 retrieval and the L1 assimilation have been minimized as much as possible, for example by using the same radiative transfer model (RTM) (RTTOV) and control vector (O3 and surface skin temperature (SST)) in both approaches

Summary

Introduction

The global monitoring of atmospheric composition relies on a large number of dedicated satellite missions and on the sustained improvement of numerical forecast models. Research and operational centres provide both satellite-based reanalyses and forecasts of atmospheric composition for a large number of applications, spanning from stratospheric ozone monitoring (van der A et al, 2010) to climate change (Flemming et al, 2017) and air quality (Zhang et al, 2012; Marécal et al, 2015). Satellite sensors measure the spectral signature of gases and aerosols on the radiation field that traverse the atmosphere. The accuracy of the solution depends in general on the intensity of the spectral signature of the retrieved compound, the source of radiation (e.g. the Earth or the Sun), the observation geometry and the accuracy of the radiative transfer model (RTM). The last-mentioned factor means correctly accounting for all the atmospheric constituents or surface properties that affect the radiation field but are not retrieved themselves (auxiliary RTM inputs)

Objectives

Methods

Results

Conclusion