1

The Changing Atmosphere Infra-Red Tomography explorer (CAIRT):�An ESA proposed mission to observe stratosphere-troposphere exchange (among others)

Quentin Errera, Gérard Ancellet, Bernd Funke, Sophie Godin-Beeckman, Michael Höpfner, Marc Op de beeck, Gabriele Poli, Peter Preusse, Piera Raspollini, Jörn Ungermann, Björn-Martin Sinnhuber, Sarah Vervalcke

​

Quadrienial Ozone Symposium 2024

Observing System Simulation Experiment (OSSE) �for CAIRT ozone in the UTLS

2

Simulating CAIRT observations using model data and CAIRT error specifications

1. CAMS model used as “true” ozone field (here shown during a tropopause fold above North America on Dec 15, 2021)

Observing System Simulation Experiment (OSSE) �for CAIRT ozone in the UTLS

3

Simulating CAIRT observations using model data and CAIRT error specifications

1. CAMS model used as “true” ozone field
2. Computing CAIRT synthetic data: CAMS interpolated at CAIRT simulated geolocation, add instrumental and smoothing error according to CAIRT estimated performances, discard points obstructed by clouds using CAMS cloud fraction (and not ERA5 as mentioned in the online talk)

​

​

Observing System Simulation Experiment for CAIRT ozone in the UTLS

4

Simulating CAIRT observations using model data and CAIRT error specifications

1. CAMS model used as “true” ozone field
2. Computing CAIRT synthetic data: CAMS interpolated at CAIRT simulated geolocation, add instrumental and smoothing error according to CAIRT estimated performances, discard points obstructed by clouds using CAMS cloud fraction
3. Simulate MLS to compare with state-of-the-art limb O₃ profiles

​

​

Results: Dec-Jan zonal mean O₃ in stratosphere and troposphere

5

• In the stratosphere, CAIRT and MLS allow correcting BASCOE low bias
• In the troposphere, CAIRT partly corrects BASCOE low bias and to a larger extent than MLS...

No data assimilation here, only pure model

Results: TP-fold above North America on Dec 12, 2021 at 15 UT

6

... and in particular for this specific tropopause fold event.

Map at 7 km

Lat-Alt at 110°W

More...

​

• ... about the methodology, additional results, affiliations ⇒

​

​

• ... about CAIRT ⇒ www.cairt.eu

​

​

• If none of these works ⇒ quentin.errera@aeronomie.be

7

Authors and affiliations

Quentin Errera¹, Gérard Ancellet², Bernd Funke⁴, Sophie Godin-Beeckman², Michael Höpfner⁵, Marc Op de beeck¹, Gabriele Poli⁶, Peter Preusse⁷, Piera Raspollini⁶, Jörn Ungermann⁷, Björn-Martin Sinnhuber⁵, Sarah Vervalcke¹

​

¹ Royal Belgian Institute of Space Aeronomy, Belgium

² Centre National de la Recherche Scientifique, Sorbonne Université, France

³ European Centre for Medium range Weather Forecast

⁴ Instituto de Astrofísica de Andalucía, CSIC, Spain

⁵ Karlsruhe Institute of Technology, KIT, Karlsruhe, Germany

⁶ Institute of Applied Physics ‘N. Carrara’, Italian National Research Council, Italy

⁷ Forschungszentrum Jülich, Jülich, Germany

​

8

Abstract

The Changing-Atmosphere Infra-Red Tomography Explorer (CAIRT) is a candidate for ESA’s Earth Explorer 11. This mission has been proposed in order to achieve a step change in our understanding of the coupling of atmospheric circulation, composition and regional climate. The CAIRT concept proposes to perform limb infra-red tomography of the atmosphere from the troposphere to the lower thermosphere (about 5 to 115 km altitude) with a 400 km swath and having high spatial and spectral resolution to provide a three-dimensional picture of atmospheric structure at unprecedented scales.

This contribution will investigate the capability of CAIRT to analyse stratosphere to troposphere exchange using an Observing System Simulation Experiment (OSSE). In this effort, a reference atmosphere – the nature run in the OSSE terminology – is built based on the Copernicus Atmosphere Monitoring Services (CAMS) control run in 2021 (i.e. with a horizontal resolution ~40 km and a vertical resolution ~500 m in the tropopause region). The nature run is used to generate CAIRT ozone profiles, along with a CAIRT orbit simulator and a simulator to generate CAIRT ozone retrieval responses and error covariance matrix. Simulated CAIRT ozone profiles are then assimilated by the Belgian Assimilation System for Chemical ObsErvations (BASCOE) to provide ozone analyses – the assimilation run. In order to measure the added value of CAIRT data in the assimilation run, a BASCOE control run without CAIRT assimilation, is also done. We have also simulated and assimilated MLS in order to measure the added value of CAIRT against MLS. Assessment of the CAIRT ozone profiles will be based on the comparison of the nature run against the three other experiments in general as well as during tropopause fold events.

​

9

Background

• Stratospheric ozone is an important source of tropospheric ozone where ozone has its largest radiative impact (Forster and Shine, 1997)

​

• As stratospheric circulation is projected to intensify over the coming century, this could lead to an increase in the ozone flux from the stratosphere to the troposphere (Neu et al., 2014, Abalos et al., 2020)

​

​

10

From Abalos et al. (2020), fig. 1b

Motivations of this study

• Evaluate CAIRT capabilities to observe ozone in the UTLS

​

• Evaluate the added value of CAIRT spatial resolution against state-of-the-art limb satellite instruments (e.g. MLS)

​

• This is done using an observing system simulation experiment (OSSE)
• How would CAIRT constrain modelled ozone in the UTLS using data assimilation?

11

OSSE concept (e.g. Errera et al., 2021, AMT)

1. Need for a reference ozone atmosphere to simulate observations, i.e.:
• Simulated ozone fields (=> the nature run – NR)
• An orbit and geolocation simulator of CAIRT data
• A CAIRT L2 error estimator
2. Need of a data assimilation (DA) system to assimilate the simulated observations (=> the assimilation run(s) – ARs)
• It must be based on a different model than the one used for NR
• A control run (CR, no DA) is also needed
3. Comparing NR & AR and NR & CR will quantify the constrain of the new instrument

​

​

​

12

The Nature Run

• Ozone reference atmospheric fields from CAMS control run 3 hourly output N256 (~40km) L137 => The Nature Run
• Period: Winter 2021 during the high season of STT

​

13

NR O3 (ppmv) at 200 hPa on Dec 15, 2021 at 15 UT during a deep tropopause fold

NR O3 (ppbv) around 112°W on Dec 15, 2021 at 15 UT during a deep tropopause fold

Simulating CAIRT O₃ profiles

• CAIRT polar orbit simulated using ESA CFO package
• CAIRT retrieval grid calculated using ALTIUS Spheriod codes

​

14

←

CAIRT L1 sampling

→

CAIRT L2 O₃ sampling

Simulating CAIRT O₃ profiles

• Profiles are flagged out at and below altitude where CAMS cloud fraction>0.2

​

15

Flagged profiles (in yellow) due to clouds at 15 km on Jan 2, 2022

Flagged profiles (in yellow) due to clouds at 10 km on Jan 2, 2022

Simulating CAIRT O₃ profiles

• CAMS 3 hourly snapshots interpolated at CAIRT retrieval grid => CAMS@CAIRT (left column)
• CAMS@CAIRT profiles are perturbed according to CAIRT expected L2 ozone profile performances considering AK, noise and systematic error. Part of profiles below clouds (CAMS cloud fraction>0.2) are discarded (right column, gray area denote cloud contamination)
• The two lowest levels (at 5 and 6 km) are also discarded because of large systematic error

​

16

Stratosphere

One orbit of CAMS@CAIRT (left) and CAIRT synthetic profiles (right) for one track of CAIRT

Troposphere

CAIRT superobservations

• Compared to MLS, CAIRT will have more than 12 times profiles
• In BASCOE-EnKF implementation, there is a matrix inversion which size depends on the number of observations. With CAIRT, this matrix is too large => SIGSEGFAULT
• To solve this issue, superobservations have

been implemented: At each time step for each model

grid point, set the superobservation profile and its

error as:

​

​

where N is the number of observations inside a grid

point, y_i and 𝛆_i are the value and error of the

observations

17

CAIRT synthetic observations at 7 km. The grid represents the BASCOE spatial resolution used here. Superobservations will be at the crosses (dots) when accounting (not accounting) for clouds

Simulating MLS O₃ profiles

MLS O₃ profiles are simulated using the following equation

​

where:

• y_r is MLS simulated profile
• y_t is CAMS@MLS (i.e. CAMS model has been interpolated at MLS geolocation)
• y_a is MLS apriori profile and A is the MLS AK matrix (I is identity matrix)
• MLS apriori (v5 used here) are provided in MLS data files
• MLS AK matrices are provided at several latitudes and have been interpolated at the measurement latitude (no-longitude dependence, v5 used here)
• 𝛆 is a random Gaussian perturbation profile which amplitude is based on MLS error profile
• Here, MLS error profile available is MLS data file has been used (v5)
• Alternatively, MLS precision and accuracy profiles given in MLS data and quality document could have been used (not yet tested)

​

18

Assimilation and Control Runs

• Using BASCOE-EnKF system driven by ERA5 using COPCAT linearized ozone chemistry (Errera et al., 2021, AMT)

​

19

CAMS L137

BASCOE L51

BASCOE L66 (not used here)

Results

20

Bias

Std Dev

Correlation

MLS_AR-NR

CAIRT_AR-NR

CR-NR

Mean difference between NR and {CR,CAIRT_AR,MLS_AR} for Dec2021-Jan2022

In the stratosphere, CAIRT and MLS correct the deficiencies seen in CR

In the troposphere, CAIRT corrects a large part of the deficiencies seen in CR while MLS correction is limited below the tropopause

Results

21

MLS_AR-NR

CAIRT_AR-NR

CR-NR

Bias

Std Dev

Correlation

• Mean difference between NR and {CR,CAIRT_AR,MLS_AR} for Dec2021-Jan2022 at 7 km
• CAIRT provide significant information even down in the free troposphere

Showcase of a tropopause fold above Europe (on Dec. 3, 2021 at 6UT)

22

Even with the relatively low resolution of BASCOE, CAIRT data allow improving BASCOE during this event, larger than MLS

Conclusions

• This study shows that instruments like CAIRT could provide unprecedented tropospheric ozone information
• Future plan:
• How could CAIRT resolve tropopause fold and O₃ flux across the tropopause?
• Try to understand why does CAIRT perform better than MLS? Is it due to its larger sampling or its lower uncertainty?
• Doing the same study with CO and/or H₂O

​

• Any comments? Please send me an email at quentin.errera@aeronomie.be

​

23