Tropospheric ozone retrieval from S5P/TROPOMI satellitedata at middle latitudes – first results

Mark Weber, Kai-Uwe Eichmann,

Swathi M. Satheesan, and John P. Burrows

Email: weber@uni-bremen.de

Neue Perspektiven der Erdbeobachtung 3. Symposium zur angewandten Satellitenerdbeobachtung, 15. bis 16. Juni 2021

„good ozone“

„bad ozone“

Weber et al.

Motivation & Goals

Motivation

Tropospheric ozone is a secondary air pollutant and important greenhouse gas (“bad ozone”)

Goal

• Extend the applicability of two existing satellite

tropospheric ozone retrievals (convective cloud

differential CCD, cloud slicing algorithm CSA) from the

tropics to middle latitudes

• Enable global observations (S5P, Sentinel-5)

• New Application for geostationary satellites like ESA

Sentinel 4, NASA Tempo, and GEMS (Korea) covering

only middle latitudes (hourly AQ observations)

• Extension of long-term observations into coming

decades (e.g. Copernicus Sentinel-4/Sentinel-5)

How is that possible

Taking advantage of the very high spatial resolution (3.5

× 5.6 km2) of the Sentinel-5P (S5P) Tropomi satellite

instrument (launched in 2017) by combining total ozone

column and cloud observations

2

Sentinel-4TEMPO

GEMS

Cloud height Ozone column

Weber et al.

Current status: Convective Cloud Differential (CCD)

Standard method

→ Determine mean ozone columns above convective clouds in the Pacific sector (~ stratospheric ozone column)

→ Subtract above cloud column ozone (ACCO) from total column amounts under clear-sky condition (all longitudes) to obtain tropospheric ozone column amounts up to cloud-top height in a grid box

→ Correct tropospheric column up to reference altitude (e.g. 270 hPa ~10.5 km)

Assumption

→ stratospheric ozone is invariant (approximately only true in the tropics)

surface

270 hPa

cloudy scene

clear-sky scene

30 DU

• Tropospheric ozone column = 340 -310 DU =30 DU (up to 300 hPa)

• Climatological correction (1DU) to extend column amount up to 270 hPa (reference altitude): TTCO = 31 DU

20°S

20°N

Eq.

S5P CHORA 3-y mean 3

Data: S5P/TROPOMI

→ Resolution (3.5 × 5.6 km2) leads to smaller grid boxes [0.5°x0.5°]

→ statistics for 1 day (instead of month)

→ more full cloud and clear-sky scenes

→ above cloud columns from nearby regions instead from the Pacific alone

→ extension into middle latitude possible

Weber et al.

Validation of CCD (CHORA) with SHADOZ ozone sondes

4

TTCO: CHORA-SONDE Median IP68/2

Difference [DU] 1.1 5

0.5°x0.25° Rel.Diff. [%] 4 25

2.5°x5° Rel.Diff. [%] 6.5 25

Good agreement in global bias and Pearson correlation (0.79) for different versions (#~500). Large dispersion partly due to pinpoint vs area averaged measurements. IP68/2 stands for half of

the 68% interpercentile (equal to 1σ).

Southern Hemisphere ADditional OZonesondeshttps://tropo.gsfc.nasa.gov/shadoz/Archive.html

Comparison with 2019 Ascension Island NASA/GSFC SHADOZ V6 ozone sonde: bias 9%. The small dots are the daily CCD values.

Time dependencies of monthly mean biases [DU] for all SHADOZ ozone sonde sites 2018-2020

number of ozone sondes

Weber et al.

Sliding sector CCD: improvements

5

Problem: Global pattern of biases [%] using the Pacific sector as cloud reference. UT/LS ozone can change around the equator and cloud height correction only depends on CSA volume mixing ratios from the Pacific sector.

Solution: Take changes of ACCO into account by adjusting the reference sector calculation for each grid box (ACCO 2D->3D). Reduction of positive bias with negative bias in summer months.

• An optimized ACCO grid box size is needed to extend the retrieval to the extratropics❑ Ascension Island is a region with only low clouds making a reduction from 50° unfeasible. ❑ Hilo is on the edge to the extratropics. First results with a reduction to ±10° longitudinal

extent are encouraging.

Weber et al.

Summary

• Existing algorithms are accurate in the tropics (CCD: 4%, Cloud slicing (not shown here): 1%) but

exhibit high dispersions (CCD: 25%, CSA: 26%)

• Variable cloud reference sector improves statistics for „problematic“ stations (Ascension,Hilo)

• Reduction of grid box size and smaller and variable ACCO sectors makes an extension to

extratropics possible

• Further optimising local cloud detection and retrieval

• Combine CCD (homogenous cloud heights) and cloud-slicing (variable cloud heights) into the DEcision Making Algorithm for Tropospheric Ozone Retrieval (DEMATOR) for global application

• Validation of DEMATOR by comparisons with ozone sondes and ozone profiles

• Writing an Algorithm Technical Base Document (ATBD) in preparation for prototyping an operational algorithm in the ground segment

6

Outlook

Weber et al.

Tropospheric ozone retrieval methods (2): Cloud slicing (CS)

standard method

→Regression of above cloud ozone columns against cloud-top-pressure results in mean ozone volume mixing ratios

→Statistics in a given grid box (monthly average value)

assumptions

→ stratospheric ozone is invariant in medium areas (approximately only true in the tropics)

S5P/TROPOMI

→ smaller grid boxes [3°x3°]

→more cloudy scenes per box

→ statistics possible for one day

→ extension into middle latitude possible

Ziemke et al. 2001

7

Weber et al.

DEMATOR algorithm

DEcision Making Algorithm for Tropospheric Ozone Retrieval (DEMATOR)

• regional total ozone and cloud statistics in each grid box

• case selection if total (stratospheric) ozone is invariant

→ variable cloud heights (case a) -> apply regional modified CS algorithm

→ stable cloud height (case b) -> apply regional CCD algorithm

• Case selection when scene is cloud-free

→ interpolate from neighboring grid boxes (case c)

8

case a

case b

case c

Weber et al.

Motivation

Role of tropospheric ozone (0-15 km):

• air pollution

→Smog (produced by nitrogen oxides & hydrocarbons)

→toxic for plants (reduced photosynthetic activity & crop yield)

→respiratory problems and heart disease (premature mortality)

• climate

→greenhouse gas

→modifies lifetimes of other green house gases (methane)

Changes in tropospheric ozone

• ~13% increase since pre-industrial times (IPCC 2013)

• Uncertainty in current trends (Gaudel et al., 2018)

• Future changes depends on future greenhouse gas scenarios (CO2 doubled=18% increase in 2100)

• Mean O3 lifetime ~23d → long-range ozone transport into remote regions

Need for continued and improved global satellite measurements of tropospheric ozone

-2.5 to +2.9 Tg/year (Gaudel et al. 2018)

http://www.wsl.ch/info/mitarbeitende/schaub/Mayer_and_Schaub_2010

9