Post on 27-Nov-2021
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
SMOSOcean Salinity Validation Overview
1
J. Font, J. Boutin, N. Reul, P. Spurgeon, J. Ballabrera, A. Chuprin, C. Gabarró, J. Gourrion, C. Hénocq, S. Lavender, N. Martin, J. Martínez, M. McCulloch, I. Meirold-Mautner, C. Mugerin, F. Petitcolin, M. Portabella,
R. Sabia, M. Talone, J. Tenerelli, A. Turiel, J.L. Vergely, P. Waldteufel, X. Yin, S. Zine, S. Delwart (The SMOS L2 OS Team)
ICM-CSIC/SMOS-BEC Barcelona, LOCEAN Paris, IFREMER Brest,ARGANS Plymouth, ACRI-ST Sophie-Antipolis, ESA-ESTEC Noordwijk
jfont@icm.csic.es
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
SMOS SSS mission requirements
• The multiangular measurements of any point on the Earth’s surface provided by the SMOS interferometric radiometer MIRAS at each satellite overpass are aimed at:
– Determining sea surface salinity with an accuracy of the order of 0.1 practical salinity units, 100 – 200 km spatial resolution and 10 – 30 days temporal resolution
2
Overall SMOS scientific goalTo provide global coverage of Sea Surface Salinity fields, with repetition rate and accuracy adequate for oceanographic, climatological and hydrological studies and increase the present knowledge on: • Large-scale ocean circulation• Water cycle exchange rates quantitative estimation• Occurrence of natural catastrophic events• Management of water resources• Role of the ocean in the climate system
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010 3
Sensitivity of Tb to SSS is small especially at low temperatures (few K range, 100 K for SM)
Retrieving salinity with SMOS is a challenge that requires very good performance of the instrument and a very demanding data processing: models and algorithms being implemented and now improved
Polarisation and viewing geometry provide a higher range of Tb values
SSS and Tb values
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
SMOS Sea Surface Salinity retrieval
• Retrieval from brightness temperature through a convergence loop
• Comparing model (guessed SSS) with measurements at all incidence angles
• Sea surface emissivity model (top 1 cm) including roughness effects
• Other contributions from external sources and atmospheric effects
• Need for auxiliary data to describe environmental conditions (ECMWF)
4
Tb,p = Tb,p flat (θ, SST, SSS) + Tb,p rough (θ, Ф, wind waves, swell, other wave characteristics, foamcoverage, foam emissivity, rain)
Ocean
Atmosphere
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010 5
• L2 accuracy estimated during algorithm validation- 1-2 psu range - function of distance to track- depending on environmental variables (uncertainty larger in cold waters)
SMOS SSS retrieval expectations
• Mission requirements to be reachable through spatio-temporal averaging (level 3)
• CATDS http://www.cesbio.ups-tlse.fr/fr/smos/smos_catds.html
• CP34 http://www.cp34-smos.icm.csic.es/.
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
The SMOS L2 OS processor
6
Retrieved SSS along a SMOS ascending orbit in the Pacific Ocean.Unfiltered (left) and filtered (right, removing flagged data) values
Developed by the SMOS L2 OS team and implemented by ACRI-ST, Fr and Argans Ltd., UK
Last version 317, delivered late November 2010Includes first modifications using models fit to SMOS data
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
L2OS processor performance
Good to detect strong SSS gradients: the Amazon plume
SMOS descending orbits, March 2010
7
by N. Reul, J. Tenerelli, IFREMER & CLS, Brest
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
5 day average SSS, 30 August – 3 September 2010
SSS maps built from L2OS processor
by P. Spurgeon & A. Chuprin, Argans Ltd., Plymouth
Looks good, but not meeting requirements yet
SSS climatology
8
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
Global SMOS L2 error: 1.5 psu
SMOS L2 SSS too saltyin the mean by ~0.45 psu
Global Std error ~1.5 psu
Histogram of the différencesbetween SSS in Situ and SSS SMOS L2
Top: Comparaisons between SSS in Situ and SMOS L2Bottom: median differences per bins of ±0.5 psu of in situ SSS
Reul et al., IFREMER
Comparison to ARGO SSS
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
• Bias in the comparison of measured and modeled Tb
– Spatial pattern persistent along and in different orbits– Similar using different ocean emissivity models: mainly related to
instrument and image reconstruction imperfections
– Removal techniques being tested: Ocean Target Transformation (OTT, mean residual bias over homogeneous ocean areas) now implemented in L2OP
– Analysing how often and by what method to build OTT10
Main problems in SSS retrieval
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010 11
• Impact of land on Tb bias patterns
Main problems in SSS retrieval
by J. Tenerelli & N. Reul
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
• Contamination from radio frequency interferences (RFI)
Main problems in SSS retrieval
12
by R. Oliva, ESA
Point sources or large areas affected by extended sources
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
L1 TB Temporal Drift for each pass typeEvaluated with the vicarious metrics:
Spatially Averaged ΔTb=<Tb meas –Tbmod>
The potential combination of :
Instrumental drifts (NIR AB, thermal correlations between Tbs & TP7, …),
seasonal geophysical variability in Tbs contaminations (sunglint, direct sun,galaxy, sea ice extent, faraday..) ,
forward model & auxilliary data errors(roughness, galactic glints,..)
results in clearly observable temporal drifts in the « Data‐Model ΔTbs » averages as reported since April. Since mid‐august:
‐the increasing Descending pass X pol data started to diverge from their Ascending X‐pol conterparts which decreasedcontinuously.
‐while for TY a similar decreasing trend isobserved for both passes =>Drifts are Polarization & pass dependent
Reul et al, IFREMER.
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
• Asymmetry ascending-descending passes
- Different land contamination impact- Different sun position with respect to spacecraft- Different galactic noise scattering
Main problems in SSS retrieval
by J. Boutin, LOCEAN, Paris
SMOS SSS ascending August 2010 SMOS SSS descending August 2010
14
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
• Roughness correction models to be improved– Three options implemented in SMOS L2OS processor– Models fail at high winds
Main problems in SSS retrieval
Weekly global salinity map, July 2010: weighted averaging + discarding flagged data, plus difference with climatology
by M. Talone & R. Sabia, SMOS-BEC, Barcelona
15
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
New empirical roughness models
SMOS SSS anomalies wrt climatology, using one of the models (Model 3) implemented in L2OS (top) and a new simplified roughness model built by fitting to data (bottom)
by J. Gourrion, SMOS-BEC, Barcelona
16
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
SSS SMOS – ARGO collocations show relatively good performance of theoretical roughness correction models (defined previous launch) in 3-12 m/s wind speed range; need further correction at high wind speed
Using ARGO for model improvement
by J. Boutin, X. Yin, LOCEAN, Paris
S. Pacific: N=134243-12m/s=0.33+/-1.34
ITCZ: N=153503-12m/s=-0.14+/-0.67
Wind speed m/s
SS
Ssm
os–
SS
Sar
go(p
su)
0
12
Existing models are going to be improved in next version of the L2 OS processor by refining theoretical models and/or adjusting empirical parameters to fit the data
17
S. Indian: N=29813-12m/s=0.17+/-0.80
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
Differences with ARGO buoys observations (1060 profiles)One week of data, March 2010Simplified SSS linear retrieval, X and Y pol separately
0.31.8
0.11.5
Preliminary SMOS SSS validation
by J. Gourrion, SMOS-BEC, Barcelona18
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
SMOS SSS validation
• Still solving issues at L1 and L2• Receivers drift (modeling physical temperature)• Land-sea contamination• RFI detection and mitigation• Galactic noise scattering,• Faraday correction, …• Roughness and foam effect correction
• Comparing with in situ data is now helping to improve forward models
• Diagnostic sites for validation• Further improvements expected in L2 processor
• New version every 6 months• Reprocessing every 12 months
19
SMOS Validation and Retrieval Team Workshop, ESRIN, November 2010
Diagnostic sites
• Selected areas of special interest for product performance analysis
20
(now Pb of RFI)
(now Pb of coast vicinity)
COST Action network
Coordinate pan-European teams to define common protocols to produce high-level salinity maps and related products, and broaden expertise in their use for operational applications.
Benefits
• Unify disperse knowledge and efforts
• Define standardized data processing protocols
• Address problems with high societal/industrial impacts
SMOS-MODE SMOS Mission Oceanographic Data Exploitation
Schedule
• Sep.09: Preliminary proposal
• Nov.09: Selected for full-proposal
• Mar.10: Selected for hearing session
• Jun.10: Approved!
• Jan.11: Kick-off meeting
Features
• Workgroups meetings
• Workshops
• STSM- Short Term Scientific Missions• Training school• Keynote speakers
R. TonboeS. Scory
C. GommengingerA. Turiel A. Ostrovskii A. Martins J. Johannessen
A. Drago P.M. Poulain B. Ward G. Korres
N. ReulM. HallikainenG. Zodiatis
D. Stammer
Interested groups:contact your national delegates or join as new member country
21
SMOS SAG, ESAC, 15-16 March 201022
Thank you for your attention!