1 / 28

Application of HUT snow emission model for practical retrieval of snow cover parameters

Application of HUT snow emission model for practical retrieval of snow cover parameters. Juha Lemmetyinen, Anna Kontu, Matias Takala, Kari Luojus, Jouni Pulliainen Arctic Research, Finnish Meteorological Institute, Finland. Microsnow2014. Modelling – practical purpose.

norm
Télécharger la présentation

Application of HUT snow emission model for practical retrieval of snow cover parameters

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Application of HUT snow emission model for practicalretrieval of snowcoverparameters Juha Lemmetyinen, Anna Kontu, Matias Takala, Kari Luojus, Jouni Pulliainen Arctic Research, Finnish Meteorological Institute, Finland Microsnow2014

  2. Modelling – practicalpurpose • Modeling of microwave propagation is needed to understand factors affecting the detected signal in Earth Observation • Models can be empirical, based on analytical modeling of the target physics, or a combination of these. • A working model makes possible the separation of the wanted parameter from the observation

  3. Complexity vs. simplicity

  4. HUT snow emission model (Pulliainen et al. 1999) • Development initiated for ESA project “Retrieval of Geophysical Parameters with Integrated Modeling of Land Surfaces and Atmosphere”; 1998 • Helsinki University of Technology (HUT) • University of Bern • Gamma Remote Sensing • Two snow emission models resulted from the project • Microwave Emission Model for Layered Snowpacks (MEMLS) • HUT snow emission model • In addition, recommendations for modeling atmosphere and vegetation influences in the microwave regime

  5. HUT snow emission model(Pulliainen et al. 1999) • Original model formulation • Considers a single homogeneous snowpack • Snow characterized by • Thickness • Density • Temperature • Grain size (‘traditional’ Dmax) • Moisture • Salinity • Snow dielectric properties from theoretical & empirical formulae • Total extinction from an empirical relation • Assumption of strong forward scattering (q=0.96) • Assuming the influence of multiple scattering • Omission of backscattering component in radiative transfer equation • 2-stream (or, 1-stream….) approximation of radiative transfer equation

  6. HUT snow emission model(Pulliainen et al. 1999) • Modification for multiple layers (Lemmetyinen et al., 2010) • Main motivation: simulation of (relatively) simple layered structures such as lake ice, depth hoar, ice lenses • Original simulation methods for individual layers retained (e.g. wave coherence effects omitted)

  7. Radiative transfertheoryScalarform losses caused by absorption radiation losses due to scattering increase of radiation energy increase of energy from radiation directed at the unit volume

  8. ; Main equations • Total extinction in snow: empirical (Hallikainen et al., 1987) • Absorption: complex permittivity calculated by theoretical and empirical relations (e.g. moisture inclusions) • Radiative transfer equation: approximation of strong forward scattering (q = 0.96) & omission of backscattering component (‘1-flux ‘ approximation) • presentation by Jinmei on Friday!

  9. Validation example: 1-layer model • NoSREx snowpits & observations 2010-2011 • Snowpits aggregated to 1- and 2-layer representative pits (depth weight averaging for grain size) • 3rd order fit applied to grain size data 1-layer

  10. 2-layer model: validation example • NoSREx snowpits & observations 2010-2011 • Snowpitsaggregated to 1- and 2-layer representativepits (depthweightaveraging for grainsize) • 3rd orderfitapplied to grainsize data 2-layer

  11. Model inversion • For complex models f, numerical methods have to be applied for inversion • E.g. minimization of sum-of-squares of model results fi(x) vs. observations y, weighted by variance i(uncertainty of model + uncertainty of observation) • Constrained minimization: one or more of model inputs xjcan be constrained by reference value xref,j and variance of reference j,ref

  12. SWE vs. Tb? 1-layer simulation Largegrainsize Medium grainsize Small grainsize

  13. ‘Engineering solution’ - FMI approach for SWE retrieval (Pulliainen 2006) (2) Snow depth data (SYNOP) (4) Kriged SD map & SD variance map (1) Satelliteobservation (5) Land cover maps (3) Grain size map (using Kriging interpolation) SWE estimate Forward model inversion

  14. Step 1: optimization of grain size

  15. Step 1: optimization of grain size

  16. Step 1: optimization of grain size Retrieval of effectivegrainsize: unconstrainedminimization model observation

  17. Step 1: optimization of grain size Mean and standarddeviation of d0,reffrom retrieval of Mcloseststations (M = 6)

  18. Step 1: optimization of grain size Krikinginterpolationapplied to obtain 2D maps of do,ref and do,ref

  19. Step 2: model inversion for SWE

  20. Step 2: model inversion for SWE Constrainedminimization: weight of radiometerretrievalthroughuncertaintyestimate (variancet). Dt (or, SWE) constrainedbyKrikinginterpolatedfield of weatherstationsnowdepth

  21. Step 2: model inversion for SWE Variance (uncertainty) of forwardmodeltobtainedbyapproximatingTB(Dt, do,ref) as Taylor series. Referencegrainsize from Krikinginterpolatedmaps Variance of final SWE estimate:

  22. Ongoing & future development at FMI • Improvement of understanding of microwave extinction properties • Arctic Snow Microstructure experiment (ASMEx); see poster by Maslanka et al.! • Passive MW forward model refinement/development • Active models (X/Ku band) • Application of forward models for hemispheric scale SWE retrieval • Use of alternate forward models (MEMLS, DMRT-ML) • Use of improved vegetation consideration (Langlois et al., 2011) • Inclusion of complex land cover (lake ice consideration) • Retrieval of ancillary parameters, e.g. effective soil properties from L-band passive MW measurements (SMOS & SMAP) • Coupling with physical snow models to obtain a priori estimates of e.g. grain size and density

  23. Observation SWE: original SWE: model-basedconsideration for lake ice emission Difference

  24. Snow structure – simulated and measured grain size = Visually determined grain size (dmax) Anna Kontu IGARSS 2014, Quebec, Canada

  25. Simulated vs. measured grain size Mass-weighted average (full vertical profile) of simulated vs. measured grain size Optical grain size (dopt) dmax Anna Kontu IGARSS 2014, Quebec, Canada

  26. Retrieved vs. measured and simulated grain size dmax 2009-2010 = retrieved grain size deff = dmaxsimulated using SNOWPACK = visual estimates of dmax 2010-2011 • Scaling factor neede to fit SNOWPACK dmax to deff • Scaling factor varies annually (0.71…0.89) 2011-2012 2012-2013 Anna Kontu IGARSS 2014, Quebec, Canada

  27. Retrieved vs. measured and simulated grain size dopt 2009-2010 = retrieved grain size deff = doptsimulated using SNOWPACK = measured dopt 2010-2011 • Scaling factors for doptclose to 1 for 2009-2010, 2012-2013 2011-2012 2012-2013 Anna Kontu IGARSS 2014, Quebec, Canada

  28. Thank you for your attention! ASMEx IOP1, January 2014

More Related