Spatial variations of L-band emissivity in Antarctica, first results from the SMOS mission. G. Picard(1), Y. Kerr(2), G. Macelloni(3), N. Champollion(1), M. Fily (1), Kerr(2) F. Cabot(2), P. Richaume(2), M. Brogioni(3) UJF-Grenoble 1 / CNRS, LGGE UMR 5183, Grenoble, F-38041, France CESBIO (CNES,CNRS,IRD,UPS) 18 avenue Edouard Belin 31401 Toulouse, France IFAC-CNR, via Madonna del Piano 10 – 50019 Sesto Fiorentino, ItalyIGARSS Summer 2011 Vancouver, Canada
ContextGeneral context:- SMOS L-band (1.4 GHz) microwave radiometer acquires radically new data that may be ofinterest for the cryosphere in general and the Antarctic in particular.Objective of our work:- Explore the information content of SMOSdata in the continental Antarctic and proposeapplications of interest in climate andglaciology sciences. SMOS track in Antarctica (12 Jan 2010)
Context SMOS main characteristics: L-band (1.4GHz), full polarizations, variable incidence angles, ~35km resolution. What was expected before SMOS launch: 1 - In dry snow, scattering by snow grain is weak at low frequencies → the emissivity at L-band should be high, and close to 1 for incidence angles close to Brewster angle (50-55o) and for V- polarization. In such a case, TB = Tsnow, snow temperature might Tsnow be retrieved everywhere in Antarctica ! 2 - Ice absorption at L-band is very weak. → Penetration depth in dry snow is expected to be several hundreds of meters, TB should be nearly constant over time. → The Antarctic plateau could be a good external calibration target. Objective of this talk: - Test 1 and 2
Outline 1 – Processing of SMOS data in Antarctica 2 – Temporal variations of TB 3 – Spatial variations of TB 4 – Radiative transfer modeling at L-band. 5 – Concluding remarks
SMOS data processing L1C data reprocessed 2010 from Brockmann Consult The result of these processing steps is a cube of TB (x,y, t, θ, p) Read and XY2HV rotation subroutines from CESBIO - Area selection space time Incidence polarisation - Flag selection angle - Daily average - Projection to the “standard” stereographic polar projection at 25 km resolution
SMOS data processing Angular diagram of TB with all the data in 2010 at Dome C (75oS, 123oE) Physical annual-mean snow temperature TB (x,y, t, θ, V) TB TB (x,y, t, θ, H) Incidence angle Dome C TB is indeed close to the snow physical temperature near the Brewster angle (50-55o) at V-polarisation → how temporal and spatial variations look like at this viewing configuration ?
Temporal variations of TB TB (x,y, t, θ=55o, p=V-pol) Daily-mean T on the Larsen C ice shelf B Daily-mean T at Dome C (-75oS, 123oE) in the Peninsula B L-band C-band C-band L-band L-band brightness temperature is fairly Brightness temperature at any frequency constant in the dry zone. We can work with increases sharply when the snowpack averaged TB. becomes wet.
Spatial variations of L-band TB TB (x,y, <t>, θ=55o, p=V-pol) Two very different zones: the wet zone (low emissivity) and the dry zone (high emissivity)...
Spatial variations of L-band TB Number of days with melt during the austral TB (x,y, <t>, θ=55o, p=V-pol) summer 2009/2010 (derived from SSM/I). Why the emissivity is low in the wet zone ?
Spatial variations of L-band TB In the wet zone, during the summer, the liquid water is responsible for the peaks noticed in TB time-series at every frequency Wet snow (snow + max 8% of liquid water) causes very strong absorption. According to Kirchoff law, emissivity is close to 1 The physical temperature of wet snow is 273K by definition Tsnow TB ~ 273 K
Spatial variations of L-band TB Melt-refreeze cycles during the summer period form coarse grain- or icy- layers. During the winter, the brightness temperature is low because: Icy or coarse grain layer causes very strong scattering. Emanating microwaves are reflected backward (=downward). → Emissivity is very low Tsnow TB < 200 K
Spatial variations of L-band TB Focus on the dry zone: TB (x,y, <t>, θ=55o, p=V-pol) ERA Interim annual-mean air temperature The scales are slightly different TB at V-polarisation and Brewster angle is close to the physical temperature...
Spatial variations of L-band TB It means, the emissivity at V-polarisation and Brewster angle is close to 1, but how close ? e=0.97 e=1 e=0.95 Each dot corresponds to a pixel in the dry zone, SMOS TB versus ERA temperature.If ERAInterim is accurate, the emissivity is in the range [0.95, 0.97] . But ERAInterim is not perfect...and known to be warm-biased in Antarctica by a few Kelvin. Emissivity may be slightly higher.
Spatial variations of L-band TBE.g. at Dome C where accurate snow temperature is measured routinely by LGGE:TB(SMOS) = 213 KTair(ERA) = 224 K → e=0.951Tsnow = 218 K → e=0.977To exploit SMOS brightness temperature at Brewster angle and V-polarization, we need torefine our understanding of the emissivity at L-band.One solution is to use radiative transfer modeling...
Angular diagram Ingredients: DMRT-ML is the snow passive microwave radiative transfer model developed at LGGE + Density, grain size and temperature profiles measured at Dome C down to 10m and extrapolated down to 100m (snow/ice transition). → Preliminary results of predicted brightness temperature at L-band: TB (x,y, t, θ, p) Results: V-pol - TB is over-estimated at both polarizations and any incidence angle TB - The difference between H and V at high incidence angles is H-pol underestimated. - Our interpretation is that the measured density profile is too smooth... Incidence angle
Angular diagramHigh contrast of density (= refractive index) between layers causes increased differencebetween H and V polarisations at high incidence angles.To test this assumption, a new simulation with noise added to the density profile: V-pol H-polThis result should be considered as preliminary. Newdensity profiles should be collected to confirm thisassumption.
Concluding remarks First year of SMOS data shows: - The brightness temperature is fairly stable relative to the noise in the dry zone. The Antarctic plateau can be used as a calibration target at 50-55o incidence angle and V- polarisation only. - In other configurations, changes of the surface state affect the signal like at the higher frequencies (work in progress...) - In the wet zone, the signal is dominated by the emissivity variations caused by ice layers. No expected application in this zone. - In the dry zone, the signal is close to the snow temperature. Retrieval of climatological temperature from SMOS data should be achievable if the departure of the emissivity from unity is corrected. - Our modeling results at Dome C suggest the density profile is a very important characteristic to understand H-polarized brightness temperature. Applications?
Thank you for your attentionRemains of a wind-crust layer, 5m deep (=50 years old) at Dome C