Uploaded on


More in: Business , Technology
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Be the first to comment
    Be the first to like this
No Downloads


Total Views
On Slideshare
From Embeds
Number of Embeds



Embeds 0

No embeds

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

    No notes for slide
  • Wavenumber Spectra of Pacific Winds Measured by the SeasatScatterometer (article)AuthorM.H. Freilich and D.B. CheltonJournalJournal of Physical OceanographyYear1986Volume16Pages741-757KeywordsspectraAbstractVector winds measured by the Seasat-A Satellite Scatterometer (SASS) are analyzed to determine the spatial structure of oceanic surface winds over wavelengths from 200 to 2200 km. The analysis is performed in four latitudinal bands in the Pacific Ocean. Sampling characteristics of SASS preclude the possibility of determining full two-dimensional spectra; the analysis is therefore limited to one-dimensional (along the satellite ground track) spectra of vector wind components and kinetic energy.The salient features of the results are summarized as follows. (i) For each of the four geographic regions, the spectra of meridional and zonal wind components and of kinetic energy are consistent with a power-law dependence on wavenumber for midlatitude regions in both the Northern and Southern hemispheres the wave-number dependence of kinetic energy is k−2.2, while for tropical regions in both hemisphere it is k−1.9. (ii) For each individual region, the spectral dependence on wavenumber is nearly the same for both velocity components and for kinetic energy. (iii) Comparisons of zonal and meridional component spectra indicate that midlatitude winds may be isotropic, while tropical winds may be significantly anisotropic. (iv) The coherence between wind components is small everywhere.


  • 1. Retrieving Ocean Surface Wind Speeds from the Nonspinning QuikSCAT Scatterometer
    Bryan W. Stiles, R. Scott Dunbar, and Alexandra H. Chau
    Simon Yueh (presenting for the authors)
    Jet Propulsion Laboratory,
    California Institute of Technology
  • 2. Overview
    Current Status of QuikSCAT
    QuikSCAT stopped spinning in November 2009.
    Current data is all single azimuth data
    Large number of looks; reduced noise
    No directional discrimination
    Narrow swath, global coverage once per month.
    Single look wind retrieval method
    Determine wind speed from backscatter by assuming ECMWF wind direction is correct.
    Spectral and Noise characteristics of single look winds
    Comparison of spinning and non-spinning QuikSCAT wind spectra
    Investigation of residual differences from ECMWF winds
    Map of Differences from ECMWF
    Rain effects
    Overall westward bias in scatterometer winds
    Ocean current effects
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 3. Current Status of QuikSCAT
    QuikSCAT stopped spinning on November 23, 2009
    Since then we have obtained single azimuth data from a variety of incidence angles and polarizations.
    Data will be used to
    Develop geophysical model functions for OceanSAT-2
    Calibrate cryosphere products for OceanSAT-2
    Retrieve accurate wind speed profiles on a narrow (30 km) swath with global coverage once per month.
    This is what we are discussing today.
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 4. Current Status of QuikSCAT
    The two primary purposes of this talk are to
    Compare the spectra of the nonspinning wind data with that of the spinning wind data to quantify the resolution and noise characteristics of each data set.
    Compare the relatively noise-free non-spinning winds to ECWMF winds, SSM/I rain rate, and OSCAR current information in order to better quantify the differences between scatterometer and NWP winds and to demonstrate the utility of the data set.
    We have concentrated on the June-July 2011 data, because it has the same incidence angle and polarization as the nominal QuikSCAT outer beam and thus we can use the same GMF to retrieve winds and more readily compare the spinning and nonspinning data sets.
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 5. Single Look Wind Retrieval Method
    Step 1: Average 50 consecutive footprint (egg) measurements to produce a ~ 30 km by 30 km backscatter measurement.
    Slice processing is not done because it would require extensive recalibration and accurate attitude knowledge.
    Footprints move 3.8 km on ground during averaging.
    Step 2: Determine co-located ECMWF direction and thus relative azimuth.
    Step 3: Assuming ECMWF direction is correct, invert geophysical model function (GMF) to obtain retrieved speed.
    Use Remote Sensing Systems Ku2011 GMF, (Ricciardulli and Wentz)
    Step 4: Co-locate SSM/I rain rate measurements for use in rain flagging.
    26% of wind vector observations within 30 minutes of a SSM/I co-location.
    58% with 90 minutes
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 6. Comparison of Spinning and Non-Spinning Wind Profiles
    Here we compare a 1000-km long non-spinning wind speed profile (bottom) with a similar profile (top) obtained when QuikSCAT was spinning.
    Both profiles are compared with co-located ECWMF and SSM/I wind speeds.
    Rainy data is omitted.
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 7. Spectra Comparison
    Non-spinning QuikSCAT data is compared with
    ECMWF (much lower energy at meso-scales)
    Spinning QuiKSCAT data obtained from slices binned in 12.5 km by 12.5 km cells.
    Factor of two better resolution than non-spinning data which uses whole footprints rather than slices.
    2 orders of magnitude more instrument noise than non-spinning data.
    An analytical spectrum is produced assuming a k-2 slope with added white noise and a low pass filter representing the typical resolution of the backscatter measurements used in the retrievals.
    The analytical spectra for the spinning and nonspinning QuikSCAT wind speeds are consistent with the observed spectra.
    Nonspinning Spectra were computed using data from the month of June 2011.
    Spinning QuikSCAT spectra were computed using a full year (2008) of wind data.
    Wind speed spectra are computed for four different spatial regions.
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 8. Equations for Analytical Spectra
    Gaussian filter models antenna spatial response
    Half power contour
    Analytical Spectrum S(k)
    A = constant scale factor used to match the observed magnitude of the spectra
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 9. Spectra by region
    Freilich and Chelton, Journal of Physical Oceanography, 1986
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 10. 2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 11. Spectra of regions 3 and 4
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 12. Results from Spectra Study
    Noise floor for non-spinning data is ~ 2 orders of magnitude lower than for spinning data.
    The non-spinning wind speed spectrum (red) is consistent with a spectrum (black solid curve) with 30-km measurement resolution and k-2 slope.
    The noise in the new 12.5 km JPL reprocessing (cyan) of the nominal (spinning) QuikSCAT data is less than that of the current (blue) JPL data set.
    The spinning wind speed spectrum (cyan) is consistent with a spectrum (black dashed curve) with 15-km measurement resolution and k-2 slope.
    The spectra of the non-spinning wind speeds (red) are similar to the spectra of the backscatter (green).
    Interesting Observation: There is excess energy in all the observed spectra at 80-km scale for regions 2 and 3 as compared to regions 1 and 4.
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 13. Map of Difference of Non-Spinning Wind Speeds from ECWMF
    The speed bias (top) of the non-spinning QuikSCAT speeds w.r.t. ECMWF winds shows prominent discontinuities around + or – 40 deg latitude and features of rain.
    The standard deviation (bottom) of the difference also shows the effect of rain.
    All data was included in these plots. No rain flagging was applied.
    Speed Bias (Non-spinning QuikSCAT – ECMWF m/s)
    Standard Deviation of Difference (m/s)
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 14. Rain Effects on Retrieved Winds
    The top panel shows the average SSM/I rain rate from http://www.ssmi.com for the month of June 2010.
    The standard deviation of the difference between QuikSCAT non-spinning wind speeds and ECMWF wind speeds is highly correlated with rain.
    SSM/I Rain Rate, June 2010
    0.0 1.5 3.0 mm/hr
    Standard Deviation of Wind Speed Difference form ECMWF, June 2011
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 15. Overall Westward Bias
    The prominent latitudinal discontinuity in the wind speed bias (top) is due to a 0.3 m/s westward bias in the zonal component of the scatterometer winds w.r.t the ECMWF winds that has been observed for multiple scatterometers (Hristova and Rodriguez).
    The middle panel shows the residual bias when 0.3 m/s is added to the zonal component of the QuikSCAT winds.
    The bottom panel is the average zonal winds. Note the sign change between the tropics and the high latitudes. This sign change is why a constant offset in zonal winds yields a speed reduction in high latitudes and a speed increase in the tropics.
    Speed Bias (QuikSCAT –ECMWF m/s)
    Residual Speed Bias with 0.3 m/s Westward Bias Removed
    Zonal Component of QuikSCAT non-spinning winds (m/s)
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 16. Overall Westward Bias
    After the 0.3 m/s westward bias is removed from the QuikSCAT winds, the remaining bias is highly correlated with the distribution of rain.
    Arguably, the “westward scatterometer bias” is actually an eastward ECMWF bias (Hristova and Rodriguez).
    Speed Bias (QuikSCAT –ECMWF m/s) with 0.3 m/s Eastward Bias removed
    SSM/I Rain Rate, June 2010
    0.0 1.5 3.0 mm/hr
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 17. Ocean Current Effects
    OSCAR monthly average current June 2011 (m/s)
    After omitting wind vectors that are co-located with nonzero SSM/I rain rates (within 30 minutes), we compared the residual differences from ECMWF with OSCAR ocean currents.
    The residual differences when binned by OSCAR ocean current speed are consistent with the root sum square of a nominal ECMWF error (1 m/s) and the OSCAR current.
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 18. Conclusions
    Because QuikSCAT stopped spinning, nominal operation ceased in November 2009.
    The current non-spinning state of QuikSCAT
    limits global coverage to once a month;
    disallows wind direction determination due to single azimuth looks;
    makes rain detection difficult due to single polarization availability.
    Nonetheless, the currently acquired data has a unique feature.
    Due to the large number of looks, non-spinning QuikSCAT winds have negligibly small errors due to instrument noise.
    NonspinningQuikSCAT winds can be useful for
    analyzing the differences between scatterometer winds and numerical wind products.
    detecting small effects on sigma-0 that are harder to observe under higher noise conditions.
    monitoring long term trends in 6:00 AM / 6:00 PM local time wind speeds.
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds
  • 19. References
    Ricciardulli and Wentz, (Ku2011 Ku-band Geophysical Model Function) manuscript in preparation, technical report on http://www.ssmi.com
    Hristova-Veleva, S. M., and E. Rodriguez, 2010: “SST-Induced Surface Wind Response: Comparison of QuikSCAT and ASCAT depiction of the phenomenon”, OVWST meeting, Barcelona, Spain, May 2010
    2011/07/28 IGARSS 2011
    QuikSCAT Nonspinning Winds