Recovery of Near-Surface Velocity from Undrogued Drifters
Stephen E. Pazan, P. P. Niiler
Abstract
Stephen E. Pazan, P. P. Niiler
Abstract
The authors have quality controlled six global datasets of drifting buoy data, made comparisons of 15-m drogued and undrogued buoy observations, and developed a 2D linear regression model of the difference between drogued and undrogued drifter velocity as a function of wind. The data were acquired from 2334 Surface Velocity Program (SVP) drifters, including 1845 SVP drifters after they lost their drogues; 704 AN/WSQ-6 Navy drifter buoys; and 503 First Global GARP Experiment (FGGE) drifter buoys. Meridional and zonal surface wind velocity components from the global synoptic FNMOC model, the global synoptic ECMWF model, and the global synoptic NCEP model were interpolated to naval AN/WSQ-6, WOCE–TOGA buoy, or FGGE buoy positions and date/times in the datasets. Two-day mean buoy drift velocities and positions were computed: 122 101 SVP drifter mean velocities before they lost their drogues and 58 201 SVP drifter mean velocities after they lost their drogues, 21 799 Navy drifter mean velocities, and 42 338 FGGE drifter mean velocities. A regression analysis was made on selected data in selected 2° lat × 8° long bins: Uundrogued = Aundrogued + BundroguedWundrogued, Udrogued = Adrogued + BdroguedWdrogued, where Uundrogued, Udrogued was ensemble mean buoy velocity and W was wind velocity, and the real and imaginary parts of these quantities were the zonal and meridional components, respectively. The difference in these complex valued regression coefficients, Bdifference = Bundrogued − Bdrogued measured the linear response to the wind. Navy and SVPL buoy response to the wind was identical and FGGE buoy response was generally the same; the global weighted mean value of |Bdifference| was 0.0088 ± 0.002. The difference in the complex valued y intercept, Adifference = Aundrogued − Adrogued, was nearly always zero within error. The buoy response to wind was also estimated by b = (Uundrogued − Udrogued)/Wundrogued, where the velocity difference Uundrogued − Udrogued was calculated from ensemble mean buoy velocities and W was ensemble mean wind velocity. The global-weighted mean value of |b| was 0.0097 ± 0.005. The analysis also found that the phase angle of either Bdifference or b, which measures the velocity difference with respect to the wind, was zero within error and not a function of the surface wind or the Coriolis parameter.
OpenAlex reports 102 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
The authors have quality controlled six global datasets of drifting buoy data, made comparisons of 15-m drogued and undrogued buoy observations, and developed a 2D linear regression model of the difference between drogued and undrogued drifter velocity as a function of wind. The data were acquired from 2334 Surface Velocity Program (SVP) drifters, including 1845 SVP drifters after they lost their drogues; 704 AN/WSQ-6 Navy drifter buoys; and 503 First Global GARP Experiment (FGGE) drifter buoys. Meridional and zonal surface wind velocity components from the global synoptic FNMOC model, the global synoptic ECMWF model, and the global synoptic NCEP model were interpolated to naval AN/WSQ-6, WOCE–TOGA buoy, or FGGE buoy positions and date/times in the datasets. Two-day mean buoy drift velocities and positions were computed: 122 101 SVP drifter mean velocities before they lost their drogues and 58 201 SVP drifter mean velocities after they lost their drogues, 21 799 Navy drifter mean velocities, and 42 338 FGGE drifter mean velocities. A regression analysis was made on selected data in selected 2° lat × 8° long bins: Uundrogued = Aundrogued + BundroguedWundrogued, Udrogued = Adrogued + BdroguedWdrogued, where Uundrogued, Udrogued was ensemble mean buoy velocity and W was wind velocity, and the real and imaginary parts of these quantities were the zonal and meridional components, respectively. The difference in these complex valued regression coefficients, Bdifference = Bundrogued − Bdrogued measured the linear response to the wind. Navy and SVPL buoy response to the wind was identical and FGGE buoy response was generally the same; the global weighted mean value of |Bdifference| was 0.0088 ± 0.002. The difference in the complex valued y intercept, Adifference = Aundrogued − Adrogued, was nearly always zero within error. The buoy response to wind was also estimated by b = (Uundrogued − Udrogued)/Wundrogued, where the velocity difference Uundrogued − Udrogued was calculated from ensemble mean buoy velocities and W was ensemble mean wind velocity. The global-weighted mean value of |b| was 0.0097 ± 0.005. The analysis also found that the phase angle of either Bdifference or b, which measures the velocity difference with respect to the wind, was zero within error and not a function of the surface wind or the Coriolis parameter.
Key concepts: Drifter, Buoy, Zonal and meridional, Wind speed, Climatology, Meteorology, Environmental science, Geodesy