Sea Ice, Climate Change and Remote Sensing

Transcription

1 Sea Ice, Climate Change and Remote Sensing Prof. David Barber Canada Research Chair in Arctic System Science Director, Centre for Earth Observation Science University of Manitoba Winnipeg, MB. Canada ESA Summer School, August, 2006

2 Lecture outline 1) Arctic Climate Change and Remote Sensing 2) Geophysics, dielectrics and thermodynamics 3) Scattering and emission modeling

3 Outline of this talk The Electro-thermophysical concept The complex dielectric constant Scattering and emission models A few examples Conclusions

4 Three key features of the Arctic: 1) it is cold 2) it is dark 3) it is cloudy 4) it is changing

5 The Electro-thermophysical concept

6 Atmospheric Transmission and EMI

7 Review of Sea Ice Microwave Scattering Theory Microwave scattering from Sea Ice is controlled by three factors: 1) The Complex Dielectric Constant 2) The inhomogeneities of Scattering Inclusions 3) The Frequency, Polarization and Sensor Geometry of the SAR

8 Forward Approach Ice Type Ice Thickness Ice Salinity Ice Temperature Snow Depth Freeze onset date. Complex Dielectric!*=!"+j!# Radiative transfer model Multifrequency & polarized EM Signatures Inverse Approach The electromagnetic properties of sea ice IEEE TGARS, ONR ARI special issue. 36(5):

9 Temperature is the control Showtwave Flux (K ) Snow 0 Depth (cm) 30 Sea Ice 60 Average T Winter ablation 1 Diurnal! ablation 2 ablation 3 ablation 4 ablation Temperature ( C) Barber et al. 1998

10

11 An electro-thermophysical model of snow covered sea ice Q* Q* Kd Ku Ld Lu QH QE Z=0x Z=1x Z=2x K*o K*1x=Q*1x Qso Qs1x Salinity (ppt) snow QM ice Z=Nx Z=B K*B=Q*is QsB Qio Mass Density (kg m-3) Liquid (% by vol.)

12 Snow Sea Ice

13 Snow Snow Density ("s; Kg m -3 ) Snow Temperature (Ts; C) Air-Snow 18.0 Geophysics Thermodynamics Energy Flux Mass Flux Gas Flux? Snow Depth (!s; cm) # Vb "s 3.0 Snow-Ice 0.0 Grain Size Snow Grain Size (mm -2 ) Sea Ice Snow Brine Volume (Vb; %/100)

14 Snow Radiative Transfer Spectral Diffuse Attenuation Coefficient K d ($) Sea Ice

15 The Temporal Evolution of!º -5! (ERS-1) Multiyear First-Year -20 Pendular Funicular Ponding Drainage Freeze-up Winter Early Melt Melt Onset Advanced Melt

16 The Temporal Evolution of Tb

17 Coupled sea Ice thermophysical and dielectrics model. The complex dielectric constant is defined as:!% =!" + j!#!" is the permittivity!# is the loss

18 Frequency/Polarization and Sensor Geometry Frequency Polarization # of Looks Incidence Angle! h and L Snow Surface Snow Volume Ice Surface Ice Volume R i, R w, $ s, W v, S s, % s,! h and L R i, R a, R b, $ i, I s,! total =! ss + " as (#) *! sv (#') + " s (#') *! is + " si (#") *! iv (#")

19 The dielectric constant

20 The Complex Dielectric Constant The complex dielectric constant consists of a complex number!% =!" + j!#!" is the permittivity!# is the loss

21 The Complex Dielectric Constant # w " = # w$ + # w0 %# w$ # " 1+(2&f' w ) 2 w = 2$f% (# &# ) w w0 w' 1+(2$f% w ) 2 " w0 (T ) = # T $10 #4 T $10 #5 T 3 2"# w (T ) = $10 %10 % $10 %12 T $10 %14 T 2 % 5.096$10 %6 T 3 where w0 is static dielectric constant of pure water, w! is high-frequency (or optical) limit of w ( w! = 4.9), w is relaxation time of pure water (s); f is electromagnetic frequency (Hz). The Debye Model - Water

22 The Complex Dielectric Constant # b " = # w$ + # b0 %# w$ # " 1+(2&f' b ) 2 b = 2$f% b(# b0 &# w' ) + ( b 1+(2$f% b ) 2 2$# 0 f " b0 = (939.66#19.068T ) (10.737# T ) " # = ( T 2 ) (15.68+T 2 ) 2"# = $10 %2 T $10 %3 T $10 %5 T 3 The Debye Model - Brine

23 Dielectric Mixture Models $ " " ds = 1+ 3" ds v i #1 ' i & ) %" i + 2" ds ( Dry snow " ws = " ds + m v" ws 3 (" #" ) [" +(" #" )N w ds ws w ws i ]#1 3 $ i =1 Wet snow " si = " i + 3" si f a (1#" i ) 2" si +1 + " si f b (" b #" i ) 3 $ 1 1 ' & + ) %" si (1# N i )+ N i " b " si (1+ N 1 )+" b (1# N 1 )( Sea Ice Inclusion dielectric in an air background

24 Dielectric Mixture Models 5 Ice Surface Permittivity ("') Ta = -5.0 C Ta = C Ta = C Snow depth (Ds) in cm. Si= 10ppt Di = 100cm Fi = 5GHz!s = 350kg Change in!' and!' at the sea ice surface as a function of air temperature and snow thickness.

25 Dielectric Mixture Models #' mix !s=300 kg m-3 " = 5.3 GHz -4 C -12 C -20 C!s=300 kg m-3 " = 5.3 GHz -12 C -4 C -20 C #" mix Salinity Salinity Modeled permittivity () and loss () as a function of snow salinity and snow temperatures.

26 Dielectric Mixture Models 3.5 Frequency = 5.3 GHz gm cm-3!" gm cm gm cm gm cm gm cm Water Volume (%/100) Bulk dielectric permittivity (!') of snow as a function of water volume and snow density

27 Link between dielectrics and scattering/emission Snow Frazil layer! 1 ",! 1 ""! 2 ",! 2 "" S 1, T Phy1 S 2, T Phy2 Columnar layer! 3 ",! 3 "" S 3, T Phy3 T(z) ' total = ' ss + ( as ()) * ' sv ()') + ( s ()') * ' is + ( si ()") * ' iv ()") Layer properties defined by the Fresnel reflection (&) coefficient

28 Scattering & Emission Models

29 Classes of Models Multi-layer reflection Reflection-volume signature Volume scattering Integrated surfacevolume signature model Rough surface scattering Multilayer Fresnel formula No scattering effects Applicable only to Young saline ice Many layer SFT Scattering effects to some extent Applicable to Young saline ice, frozen melt ponds (up to 40 GHz) bubbly ice (up to 20 GHz) Rayleigh scattering DMRT DMT No reflection effects Poor agreements (at H-pol) over young ice, frozen melt ponds Bubbly ice up to 19 GHz Numerical method Empirical method Good potential Backscattering problem Snow application DMT-integration model No significant improvement from DMT This indicated no significant rough surface scattering effects PS GO SP Integration DMRT; Dense Medium Radiative Transfer; DMT: Dense Medium Theory, PS: physical optics under the scalar approximation, GO: geometric optics approximation, SP: small perturbation method (Winebrenner et al., 1992; Nassar et al., 2000; Wiesmann and Matzler, 1999)

30 Emission/Scattering Theory Backscattering o o ss 2! ( k o) =! + T ( ko)! + T ( ko)! snow snow Snow/sea ice surface volume interface o sv 2 o s Emission T ) T a ( k0,0) =! a ( k0 ' si 2 1 k0) = 1" R "![ % ah( k0, k) + % av( k, k)] sin$ d$ d# 4& Interface Volume reflection Scattering loss a ( 0 1 Rs =![ $ ah( ko, k) + $ av ( ko, k)] sin# d# d" 4% Snow Frazil layer! 1 ",! 1 ""! 2 ",! 2 "" S 1, T Phy1 S 2, T Phy2 Columnar layer! 3 ",! 3 "" S 3, T Phy3 T(z)

31 Frequency Polarization # of Looks Incidence Angle ' h and L Snow Surface Snow Volume Ice Surface Ice Volume R i, R w, * s, W v, S s, + s, ' h and L R i, R a, R b, * i, I s, ' total = ' ss + ( as ()) * ' sv ()') + ( s ()') * ' is + ( si ()") * ' iv ()") Forward scattering model defined for Microwave Scattering

32 physical optics/geometric optics Surface Scattering RMS height Correlation length " < # 32cos$

33 Volume scattering Number density Volume fraction Scattering physics Air-Snow !s 12.0 Ts Vb 3.0 Grain Size Snow-Ice

34 Surface Scattering! " (#) = 2 $ %% 2 cos 2 # exp (-4 K o 2! 2 cos 2 #) s Where:! & n=1 (4K o 2! 2 cos 2 #) n n! 2 (K o n / l) /2 (4K o sin # + n / l ) " x cos # - " x cos #' 2 1! HH = "2 x cos # + " 1 x cos #'! 1 = 2 1, for material #1 (air) "' + "" Forward scattering model defined for Microwave Scattering

35 Volume Scattering o! s v (") =! v cos " ' (1-2K e 1 (exp (K e d sec(" ' ))) 2 ) Where:! v = N i! bi + N w! bw! b = 64 # 5 r 6 K 2 " o 4 Forward scattering model defined for Microwave Scattering

36 Features in the mm range affect scattering Snow Density ("s; Kg m -3 ) Snow Temperature (Ts; C) Air-Snow 18.0 Snow Depth (!s; cm) # "s 6.0 Vb 3.0 Snow-Ice 0.0 Grain Size Snow Grain Size (mm -2 ) Snow Brine Volume (Vb; %/100) Ehn et al.

37 Air-Snow Ts!s Snow-Ice 0.0 Grain Size Vb

38 Many layer SFT model (scattering and emission) Description -assumes the snow/sea ice is a piecewise-continuous random medium and accounts for the interference between waves reflected and transmitted coherently by the various planar layers - accounts for the mean propagation and first-order multiple scattering effects by using bilocal and distorted born approximations (Winebrenner et al., 1992)

39 Many layer SFT model (scattering and emission) Input parameters General: frequency (GHz), angles For ice: temperature, salinity, density, ice grain size (mm), air bubble size, brine aspect ratio and tilted angle. For snow: temperature, density, snow wetness (fractional volume)*, snow grain size (mm). *liquid water distributed between grains as well as around grains (Stogryn, 1985). Output parameters Microwave brightness temperature (emissivity) and backscattering (sigma) for V and H polarizations. (Winebrenner et al., 1992)

40 Extending signatures temporally - Ice emissivity simulated by the many layer strong fluctuation theory model. The brine skim/wet slush was set to be 5 mm, and ice salinity based on field data Hwang and Barber, JGR, in press

41 Some examples

42 Snow Thickness Scattering Response May 6 (Clear) May 9 (Cloudy) Barber and Thomas, 1998! (db) - May 9 (Cloudy) N= 900 x ! (db) - May 6 (Clear)

43 Snow water equivalent (SWE) Air-Snow Ts!s Snow-Ice 0.0 Grain Size Vb Sea ice

44 SWE and Scattering Observed Response to Snow Thickness Barber et al. 1998

45 Yackel and Barber, 2000

46 SWE and Radiometry (20 sites) Snow thickness Depth (cm) SWE (mm) SWE Site Number 0 Barber et al

47 Explained variance versus ƒ,p, and ) Percent Explained Variance V 30 19V 50 19H 35 19H 55 37V 40 37V 60 37H 45 85V 30 85V 50 Frequency (19, 37, 85) Polarization (V, H) Incidence (30-60 in 5º increments) Barber et al

48 Explained variance versus ƒ,p, and ) Tb and Ts 80 Percent Explained Variance V 30 19V 50 19H 35 19H 55 37V 40 37V 60 37H 45 85V 30 85V 50 Frequency (19, 37, 85) Polarization (V, H) Incidence (30-60 in 5º increments) Barber et al

49 Late season sea ice!% =!" + j!# Melt Pond Dielectrics!% = j36.51 for pure water at 0 C, 5.3 GHz Snow Patch Dielectrics!% = j0.11 for wet snow at -1 C, 0.3 gm.m -3, 0.1 Wv

50 Melt Ponds on Landfast First Year Sea Ice.

51 ) Small -' ) Large -' ) ) Ice Patches Light Melt Pond Dark Melt Pond Albedo Ice Patches Light Melt Pond Dark Melt Pond Albedo 0.2 8,, = = =0.28

52 The Temporal Evolution of Sigma Naught Several variables must be taken into consideration: The effect of incidence angle The effect of wind The contribution to backscatter (!º) by volume and surface scattering as dictated by the dielectrics of the system

53 The Effect of Incidence Angle During periods of cold temperatures the dielectrics of the system are considered static and changes to backscatter are a function of incidence angle and surface roughness The Incident Angle Calibration Model (IACM) standardized!º to the near range of the RADARSAT- 1 swath (!º) The IACM explained in excess of 99% of the variability in backscatter which resulted from changes in incidence angle

54 The Effect of Wind Under calm conditions, volume scattering within bare ice (! i ) results in backscatter which is greater than backscatter caused by surface scattering from melt ponds (! m ) or! i >! m. This allows for an estimation of melt ponds from SAR Over melt ponds, there is an amplification of!º as a function of wind speed provided wind direction is orthogonal to the SAR pulse Between 1.5ms ms -1! i =! m. Above 2.5ms -1! i <! m.

55 Pond fraction (PF), wind speed (W), wind direction in degrees (D) and weather are all indicated for each image. The images have been calibrated to ASF gamma values and areas of low backscatter appear dark. All images are courtesy of the Canadian Space Agency ( CSA, 2002)

56 Surface Albedo l A e v a w t r( o ho ) Sd! e db e t a r g e t n I ! = * " o R 2 = ! = * " o R 2 = Transect 8 Wind Speed = 3.2 m/s 0.52 July 3 Desc; 13: 27 UTC Standard Beam Strong Wind Transect 8 Wind Speed = 5.3 m/s July 3 Asc; 23:24 UTC Standard Beam 5 Moderate Wind! = * " o R 2 = Transect 10 Wind Speed = 1.5 m/s July 6 Asc; 23:36 UTC Light Wind Standard Beam RADARSAT-1 Scattering Coefficient (" o ) Extracting Summer Ice Information from SAR

57 The Temporal Evolution of!º -5! (ERS-1) Multiyear First-Year -20 Pendular Funicular Ponding Drainage Freeze-up Winter Early Melt Melt Onset Advanced Melt

58 The Temporal Evolution of Tb

59 Conclusions Electro-thermophysical model (heuristic to physical) Emission/scattering models Geophysical vs Thermodynamic state (processes) Initialization and steering of models, data assimilation Scale related science (micro to hemispheric) Merger of environmental science and technologies

60 A metaphor for science and technology

61 Acknowledgements: John Yackel John Hanesiak Tim Papkyriakou Ryan Galley Alex Langlois Phil Hwang John Iacozza CJ Mundy Rob Kirk Sheldon Drobot Theresa Fisico Theresa Nichols Chistina Blouw Isabelle Harouche Andrew Thomas Klaus Hochheim Jens Ehn Mats Granskog Wanli Woo Randy Scharien Katherine Wilson Wayne Chan Jennifer Lukovich The real forces behind this work

62 For more information

63 Addendum