The wet tropospheric correction issue for the WATER HM mission

Water HM meeting 29.10.07 - 1 -

E. Obligis (CLS) and L. Eymard (LOCEAN)

The wet tropospheric correction issuefor the WATER HM mission


• The presence of water vapor in the troposphere induces an additional delay in the altimeter range measurement that has to be corrected for.

• This term is between 0 and 50 cm, and characterized by a high variability in space and time

• A microwave radiometer is usually added to altimetry missions (ERS1, Topex-Poseidon, ERS2, Jason1, Envisat, Jason2, AltiKa, Sentinel topography mission…)

1. Do we need a radiometer on the WATER HM mission ?

2. What type of radiometer ?

3. The mixed « land-sea » pixel issue


Figure 1 : Moyenne des différences entre le radiomètre de Topex (TMR) et le CEPMMT en moyenne sur 2003 dans des boîtes de 2°, après correction du biais

Mean differences between Topex/TMR and ECMWF dh for year 2003

-3cm +3cm

I. Do we need a microwave radiometer ?

Global quality of the meteorological models is still insufficient

Radiometer wetter

Radiometer dryer


Variance gain when using TMR dh instead of ECMWF dh

-3cm +3cm



Scharroo et al (2004)



The spatial and temporal resolutions are insufficient The actual grid is half a degree but the resolved scales are not better than 1 degree. Operationnal outputs every 6 hours (problem of localisation and temporal evolution of

the meteorological structures)

The model stability is not assured Radiometers do not provide a constant quality product (all experienced drifts and jumps,

more or less accurately corrected for) On the contrary, meteorological models assimilate different satellite and in-situ

measurements so they are less liable to drift with time Nevertheless they regularly change due to new assimilation schemes or new

assimilated data).


ECMWF-TMR wet tropo

ECMWF change 10/01/2002


We need a Microwave Radiometer onboard the mission


Assumption that is not impacted by the wet tropo (same along r1 and r2) But absolute dh is necessary to determine h We need dh into the swath

II. What type of radiometer ?



Enjolras et al (2007)


A scanning or a fixed radiometer ?

Scanning radiometer not compliant with the plateform stability requirements driven by the altimeter


MeghaTropiques mission



Standard or high frequency radiometer ?HF: Much better resolution Easier accomodation But clouds opacity and saturation for very wet atmospheres

Différences between AMSU-A and AMSU-B wet tropo estimations


II. What type of radiometer ?Standard or high frequency radiometer ?

HF:

Much better resolution Easier accomodation But clouds opacity and saturation for very wet atmospheres Maybe not mature enough

BF: Worse resolution But consolidated performances (retrieval) New generation of antennas (15km of resolution at 800km 24km at 1300km)



Nad

ir tr

ack

Interferometric swath80km

Interferometric swath80km30km

Across track

Alo

ng tr

ack

PropositionOne optimized reflectorAnd several horns to cover the swath with two frequencies (Ka and K for each view)


II. What type of radiometer ?In case of no radiometer to characterize the wet tropo into the swath,an optimal combination of the available satellite water vapor products is possible

94,9% of the Jason1 pixels can benfit from an estimation at less than 2hours/0.25°Difference JMR-ECMWF over 1 cycle :1.3 cm rms

Difference JMR-MERGED over 1 cycle :1.05 cm rms

AMSU-TMI-SSMI-AMSR

3H/0.25°

2H/0.25°


III. The mixed pixels issue

Over open ocean:

Accuracy: 1 cm

Horizontal resolution: 50 km

Mixed zones: the surrounding land surfaces contaminate the signal and make the humidity retrieval method unsuitable.


Topex/TMR illustration

TOPEX/Poseidon coverage:blue tracks are further than 50 km to the coasts (Mercier, 2005)

And not contaminated


We developed a dedicated simulator

to better understand and quantify the difficulty…


Evaluation of some current methods

Error maps

1. without correction

2. using ECMWF path delay near coasts

3. propagating the last uncontaminated path delay

1

3

+2 cm

2

-2 cm +2 cm

-2 cm +2 cm

-2 cm +2 cm


corr ( p , f ) = [TBland( f ) – TBsea ( f )] × p ( f )

Desportes et al, IEEE TGRS, vol. 45, n°7, pp 2139-2149, 2007.

p : Gaussian smoothing of a land/sea 0.01° mask that takes into account the antenna pattern for each channel

RMS error on contaminated PD

RMS error on propagated PD

RMS error on corrected PD

16 March

12.4 cm 5.2 cm 2.3 cm

15 April

10.9 cm 4.6 cm 2.6 cm

A correction method based on the proportion of land in the pixel


The one-dimensional variational method (1D-Var)

))H(()())H((2

1)()(

2

1)J( 11 XYFEXYXXBXXX 00bb TT

X : Control vector containingparameters to be adjusted

Xb : Background vector

H(X) : Simulated TBs

B, E, F : error covariance matrices

Y0 Measured TBs

H : radiative transfer model

The quality and stability of meteorological models will certainly increase in the next few years so we shall

probably combine radiometer and model estimations

RTTOV

Radiometer

ECMWF


CONCLUSIONS

• Meteorological models are not yet accurate enough (resolution, accuracy, stability) to provide the wet tropo with the required accuracy (1cm for current missions)

• Wet tropo is needed in the entire interferometric swath. The use of a nadir estimation implies an error higher than 2 cm rms depending on the position into the swath

• The nadir estimation could be provided by a bi-frequency radiometer – one around the 22.235 GHz water vapor absorption line

– one around 37 GHz for cloud liquid water content

– Surface roughness may be provided by the altimeter backs. coefficient

• A pushbroom radiometer with a big reflector and several horns to cover the swath could be a solution

• On going activities to improve the wet tropo in case of mixed pixels…

Documents

The wet tropospheric correction issue for the WATER HM mission