Retrieval of Methane Distributions from IASI A.Waterfall, R. Siddans, B. Kerridge, G. Miles, B. Latter Rutherford Appleton Laboratory Acknowledgements:

Retrieval of Methane Distributions from IASI

A. Waterfall, R. Siddans, B. Kerridge, G. Miles, B. Latter

Rutherford Appleton Laboratory

Acknowledgements: NCEO Atmospheric Composition Theme

Why measure methane with IASI?

Methane:• Important for its role in atmospheric chemistry and as a greenhouse gas• Concentrations are increasing with time (but not consistently)• Uncertainties in the global methane budget

Satellite measurements available (tropospheric methane):– mid-IR (IASI, TES, AIRS) – near-IR (GOSAT, SCIAMACHY)

Advantages/disadvantages of IASI– Global day/night coverage– Different sensitivity – Long time series of planned instruments (monitoring of trends)

IASI CH4 retrieval scheme

Retrieval Technique: Optimal estimation

Radiative Transfer model: RTTOV (with customised coefficients)

Spectral range: 1240-1290cm-1 (cf. Razavi et al, ACP 2009)

Retrieval species: CH4, N2O, H2O (log vmr)HDO scaling factor, surface temperature

Measurement noise: Derived from spectral fits (dependent on scene radiance)

Apriori + covariance matrix Fixed apriori profile (no latitude dependence)Covariance: ~ 10% error in troposphere, increased in stratosphere, with 6km correlation length

Background profiles Temperature, and apriori surface temperature, H2O from ECMWF

Simulated averaging kernels

• Retrieve methane on a fixed pressure grid (~0,6,12,16,20... km)• Principal sensitivity is in mid to upper troposphere • Limited sensitivity at the surface (dependent on air-surface temperature contrast)

Averaging kernels:

x = vmr

1km retrieval grid, 6km correlation length

Vertical sensitivity

Averaging kernels:

x = ln(vmr)

Example real averaging kernels

dx

xdˆ

Latitude

Expected methane precision (fractional error)

• Apriori error values: 10% in troposphere, higher in stratosphere• Significant improvement over apriori in mid and upper tropospheric

layers

Retrieval error Retrieved/apriori error

Fra

cti

on

al

err

or

Latitude Latitude

Version 1 of methane data• 4 continuous months of data (August –

November 2009), April 2009, August 2008

• Profile retrieval• => Column averaged mixing ratios

XC

H4 /

ppm

v

Day

Night • ‘least cloudy’ out of every 4 pixels

• Nb. retrieval very sensitive to cloud

23rd August 2009

Xch4 averaging kernel

© 2010 RAL Space

September 2009

Day

Night

Monthly mean column averaged mixing ratios• gridded 1x1 degree bins

Nb. Apriori data is a constant profile => N-S gradient comes completely from IASI

Xch4 (ppmv)

Possible reasons for day/night difference: • Difference in sensitivity• Problem with apriori

• CH4 apriori has a low bias, inconsistent with apriori error

• Different cloud sensitivity

DAY

NIGHT

NCEO Model + satellite comparisons: Aug 2009

:…..

IASI (night) GEOSCHEM (U. Edinburgh)

TOMCAT (U. Leeds)

GOSAT (U. Leicester)

Aug 2009, bias: -0.0296

Xch4 (ppmv)

April 2009 August 2009 November 2009

IASI (night time only data)

GOSAT

Monthly mean data, 1x1°bins

GOSAT data produced by R. Parker, U. Leicester, see poster by Byckling et al.

Xch4 (ppmv)

Monthly mean data on retrieval levelsAugust 2009

ppmv

178 hPa 422 hPa Column averaged mixing ratio

DAY

NIGHT

ppmv

Plots from MACC website.

MACC Reanalysis data

300 hPa

500 hPa

IASI night: 422 hPa

August 2009

Are these distributions reasonable?

ppmv

Cloud effects on the retrieval data

• Real retrievals show a tail of high methane retrievals believed to be due to cloud

• Simulations based on AVHRR-3 data indicate that the apparent methane is expected to increase with increasing cloud top height and optical depth

• Multiple scattering simulations based on more complicated (realistic) cloud distribution also show higher methane values due to increasing cloud amount

• Will be hard to filter out as even small amounts of cloud can cause slightly increased radiances.

• Retrieval of cloud top height/cloud fraction may improve retrievals in many cases.

ORAC

IASI cloud sensitivity - AVHRR/3

LUT of retrieval simulations of error due to cloud on IASI CH4 column averaged vmr

Cloud optical depth

Clo

ud A

ltitu

de /

km

Estimated Relative error on CH4 column

Cloud height / km

CH

4 en

hanc

emen

t

Summary and Future Plans

• A scheme for retrieving methane profiles from IASI has been developed at RAL. • Version 1 global distributions of methane are now available for:

August-November 2009, April 2009, August 2008• Agreement with many features seen in models / GOSAT• Current version has some issues related to cloud

contamination and certain land types• Simulations show cloud can introduce a high bias to the IASI data• Further simulations are planned to assess the impact of other

important error sources• Future versions will include improved handling of clouds

and improved apriori data