Work with Drifters

Work with east Pacific drifters

  1. Figures for comparing drifter and XBT geostrophic velocities
    Work done for the paper Mean three-dimensional circulation in the northeast tropical Pacific" (Kessler, 2002):

    Drifter tracks obtained from AOML as kriged positions. 1045 drifter tracks were found in the region 120°E to the coast, 25°S-30°N, and these contained 485752 kriged 6-hourly samples (drogue-on drifters only).
    Velocity components were gridded to a 1°x1°xmonthly grid using Gaussian mapping with scales of 2°longitude, 1°latitude and 2 months, to make an average annual cycle.

    1. Count by gridboxes  Months with drifter obs each gridbox  Average number per month each gridbox
    2. Count by years: Count  Percent  Overlay SOI
    3. Compare drifter and geostrophic currents:
      1. Components: Ug, Ud and differences  Add Ekman currents  Use R&N H* for H_E
      2. Vectors: Ug, Ud and difference  Add winds rotated 90°  U_g, U_d, Difference and Ekman (R&N H*)
      3. Seasonality: Components  Vectors  Dense vectors  Seasonal change
      4. XBT, ADCP and drifter currents at 110°W and 95°W
    4. Ralph&Niiler Ekman depth and consequences

  2. Work done trying to repeat the Poulain (JPO, 1993) result:
    1. Figures from Poulain:
      Poulain binned drifter velocities (using a Lagrangian averaging) on a 1/20th degree meridional scale, averaged over 150°W-90°W. He used about (not stated in the paper) 7000 drifters within 1°-1°N, during 1979-1991.
      1. Fig. 1. Meridional divergence at the equator. Using 1° binning.
      2. Fig. 2. Mean meridional velocity. Using 1/20th degree binning.
      3. Fig. 3. Equatorial divergence as a function of latitude band width

      Work done trying to replicate this result, using kriged positions east of 120°W (same data as in part 1 above):

    2. Number of observations: By latitude band   By longitude within 1°S-1°N. (Total 13459 within 1°S-1°N)
    3. Mean v (compare binning): Gross comparison (10°S-10°N)   Detail of 2°S-2°N   
    4. dv/dy: 1° binning   Compare different binning choices

      Bottom line: I can't duplicate Poulain's result.

      • Poulain shows mean v reaching about 4 cm/s within about 0.2° from the equator. Mine don't reach that level until about 0.4-0.5°. (Compare 2.1b and second of 2.3).
      • Poulain shows dv/dy about 10 (x10**-7/s) with 1° binning, rising to about 25 with 1/10th° binning and 35 with 1/20th° binning. I get 5 with 1° binning and perhaps 15 with 1/10th° binning. (Compare 2.1c with the second of 2.4). Thus my values are just about half of his. I find a doubling of dv/dy going from 1° to 1/5th° binning, but little change beyond that (my 1/20th° binning is too noisy to take derivatives).
      Differences that might lead to these discrepancies:
      • My data only extends to 120°W. (But Poulain's Fig 1a suggests that this should not be a problem).
      • I used the plain kriged 6-hr velocities, while Poulain used a Lagrangian averaging technique. However, he was trying to maximize use of his many fewer observations, while I have 10 more years of data and about twice as many total obs (even though the longitude band is half as wide).
        Compare dv/dy for Poulain's period (1979-91) with the entire record. Not much difference.
      So why can't I duplicate Poulain's result????

    Other relevant pages (Re Kessler 2002 paper):
  3. Main EPAC page
  4. Winds
  5. Rossby model
  6. Sverdrup Balance
  7. Mountain-gap winds (NSCAT!)
  8. Quikscat winds
  9. Paper figures page
  10. Bv=fdw/dz page
  11. Coastal BC page
  12. Talk pages: AGU  Feb 01 talks