• Accadia, C., S. Zecchetto, A. Lavagnini, and A. Speranza, 2007: Comparison of 10-m wind forecasts from a regional area model and QuikSCAT scatterometer wind observations over the Mediterranean Sea. Mon. Wea. Rev., 135, 19451960, https://doi.org/10.1175/MWR3370.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Alsabah, R., A. Al-Sabbagh, and J. Zec, 2018: RapidScat backscatter measurement validation. J. Appl. Remote Sens., 12, 034005, https://doi.org/10.1117/1.JRS.12.034005.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Alves, J. H. G. M., and I. R. Young, 2003: On estimating extreme wave heights using combined Geosat, TOPEX/Poseidon and ERS-1 altimeter data. Appl. Ocean Res., 25, 167186, https://doi.org/10.1016/j.apor.2004.01.002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bender, L. C., III, N. L. Guinasso Jr., J. N. Walpert, and S. D. Howden, 2010: A comparison of methods for determining significant wave heights—Applied to a 3-m discus buoy during Hurricane Katrina. J. Atmos. Oceanic Technol., 27, 10121028, https://doi.org/10.1175/2010JTECHO724.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bentamy, A., 2008: Characterization of ASCAT measurements based on buoy and QuikSCAT wind vector observations. Ocean Sci., 5, 77101, https://doi.org/10.5194/os-4-265-2008.

    • Search Google Scholar
    • Export Citation
  • Bentamy, A., Y. Quilfen, and P. Flament, 2002: Scatterometer wind fields: A new release over the decade 1991-2001. Can. J. Remote Sens., 28, 431449, https://doi.org/10.5589/m02-041.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chakraborty, A., R. Kumar, and A. Stoffelen, 2013: Validation of ocean surface winds from the OceanSat-2 scatterometer using triple collocation. Remote Sens. Lett., 4, 8493, https://doi.org/10.1080/2150704X.2012.693967.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crapolicchio, R., G. De Chiara, A. Elyouncha, P. Lecomte, X. Neyt, A. Paciucci, and M. Talone, 2012: ERS-2 scatterometer: Mission performances and current reprocessing achievements. IEEE Trans. Geosci. Remote Sens., 50, 24272448, https://doi.org/10.1109/TGRS.2011.2179808.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Donelan, M. A., 1982: The dependence of the aerodynamic drag coefficient on wave parameters. Proc. First Int. Conf. on Meteorology and Air–Sea Interaction of the Coastal Zone, Boston, MA, Amer. Meteor. Soc., 381–387.

  • Durrant, T. H., D. J. M. Greenslade, and I. Simmonds, 2009: Validation of Jason-1 and Envisat remotely sensed wave heights. J. Atmos. Oceanic Technol., 26, 123134, https://doi.org/10.1175/2008JTECHO598.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ebuchi, N., H. C. Graber, and M. J. Caruso, 2002: Evaluation of wind vectors observed by QuikSCAT/SeaWinds using ocean buoy data. J. Atmos. Oceanic Technol., 19, 20492062, https://doi.org/10.1175/1520-0426(2002)019<2049:EOWVOB>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Elyouncha, A., and X. Neyt, 2013a: C-band satellite scatterometer intercalibration. IEEE Trans. Geosci. Remote Sens., 51, 14781491, https://doi.org/10.1109/TGRS.2012.2217381.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Elyouncha, A., and X. Neyt, 2013b: Inter-calibration of MetOp-A and MetOp-B scatterometers using ocean measurements. Proc. SPIE, 8888, 888806, https://doi.org/10.1117/12.2033927.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Evans, D., C. Conrad, and F. Paul, 2003: Handbook of automated data quality control checks and procedures of the National Data Buoy Center. NOAA/NDBC Tech. Doc. 03-02, 44 pp.

  • Figa-Saldaña, J., J. J. Wilson, E. Attema, R. Gelsthorpe, M. Drinkwater, and A. Stoffelen, 2002: The Advanced Scatterometer (ASCAT) on the Meteorological Operational (MetOp) platform: A follow on for European wind scatterometers. Can. J. Remote Sens., 28, 404412, https://doi.org/10.5589/m02-035.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Guan, C., and L. Xie, 2004: On the linear parameterization of drag coefficient over sea surface. J. Phys. Oceanogr., 34, 28472851, https://doi.org/10.1175/JPO2664.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Harper, W. V., 2014: Reduced major axis regression: Teaching alternatives to least squares. Ninth Int. Conf. on Teaching Statistics, Flagstaff, AZ, International Statistical Institute, https://iase-web.org/icots/9/proceedings/pdfs/ICOTS9_C131_HARPER.pdf.

  • Holbach, H. M., and M. A. Bourassa, 2017: Platform and across-swath comparison of vorticity spectra from QuikSCAT, ASCAT-A, OSCAT, and ASCAT-B scatterometers. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 10, 22052213, https://doi.org/10.1109/JSTARS.2016.2642583.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Holland, P. W., and R. E. Welsch, 1977: Robust regression using iteratively reweighted least-squares. Commun. Stat. Theory Methods, 6, 813827, https://doi.org/10.1080/03610927708827533.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • IMOS, 2015a: IMOS netCDF conventions, version 1.4. University of Tasmania Rep., 82 pp., https://s3-ap-southeast-2.amazonaws.com/content.aodn.org.au/Documents/IMOS/Conventions/IMOS_NetCDF_Conventions.pdf.

  • IMOS, 2015b: IMOS netCDF file naming convention, version 1.5. University of Tasmania Rep., 22 pp., https://s3-ap-southeast-2.amazonaws.com/content.aodn.org.au/Documents/IMOS/Conventions/IMOS_NetCDF_File_Naming_Convention.pdf.

  • Jensen, R., V. Swail, R. Bouchard, R. Riley, T. Hesser, M. Blaseckie, and C. MacIsaac, 2015: Field Laboratory for Ocean Sea State Investigation and Experimentation: FLOSSIE: Intra-measurement evaluation of 6N wave buoy systems. 14th Int. Workshop on Wave Hindcasting and Forecasting/Fifth Coastal Hazard Symp, Key West, FL, JCOMM.

  • Large, W. G., J. Morzel, and G. B. Crawford, 1995: Accounting for surface wave distortion of the marine wind profile in low-level ocean storms wind measurements. J. Phys. Oceanogr., 25, 29592971, https://doi.org/10.1175/1520-0485(1995)025<2959:AFSWDO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Madsen, N. M., and D. G. Long, 2016: Calibration and validation of the RapidScat scatterometer using tropical rainforests. IEEE Trans. Geosci. Remote Sens., 54, 28462854, https://doi.org/10.1109/TGRS.2015.2506463.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pensieri, S., R. Bozzano, and M. E. Schiano, 2010: Comparison between QuikSCAT and buoy wind data in the Ligurian Sea. J. Mar. Syst., 81, 286296, https://doi.org/10.1016/j.jmarsys.2010.01.004.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pickett, M. H., W. Tang, L. K. Rosenfeld, and C. H. Wash, 2003: QuikSCAT satellite comparisons with nearshore buoy wind data off the U.S. West Coast. J. Atmos. Oceanic Technol., 20, 18691879, https://doi.org/10.1175/1520-0426(2003)020<1869:QSCWNB>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Portabella, M., A. Stoffelen, J. Verspeek, A. Verhoef, and J. Vogelzang, 2007: ASCAT scatterometer ocean calibration. Int. Geoscience and Remote Sensing Symp., Barcelona, Spain, IEEE, 2539–2542, https://doi.org/10.1109/IGARSS.2007.4423361.

    • Crossref
    • Export Citation
  • Priestley, C. H. B., 1959: Turbulent Transfer in the Lower Atmosphere. University of Chicago Press, 130 pp.

  • Ribal, A., and I. R. Young, 2019: 33 years of globally calibrated wave height and wind speed data based on altimeter observations. Sci. Data, 6, 77, https://doi.org/10.1038/s41597-019-0083-9.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Satheesan, K., A. Sarkar, A. Parekh, M. R. R. Kumar, and Y. Kuroda, 2007: Comparison of wind data from QuikSCAT and buoys in the Indian Ocean. Int. J. Remote Sens., 28, 23752382, https://doi.org/10.1080/01431160701236803.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Spencer, M. W., C. Wu, and D. G. Long, 2000: Improved resolution backscatter measurements with the SeaWinds pencil-beam scatterometer. IEEE Trans. Geosci. Remote Sens., 38, 89104, https://doi.org/10.1109/36.823904.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stoffelen, A., 1999: A simple method for calibration of a scatterometer over the ocean. J. Atmos. Oceanic Technol., 16, 275282, https://doi.org/10.1175/1520-0426(1999)016<0275:ASMFCO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Takbash, A., I. R. Young, and Ø. Breivik, 2019: Global wind speed and wave height extremes derived from long-duration satellite records. J. Climate, 32, 109126, https://doi.org/10.1175/JCLI-D-18-0520.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Taylor, P. K., and M. J. Yelland, 2001: Comments on “On the effect of ocean waves on the kinetic energy balance and consequences for the inertial dissipation technique.” J. Phys. Oceanogr., 31, 25322536, https://doi.org/10.1175/1520-0485(2001)031<2532:COOTEO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Thomas, B. R., E. C. Kent, and V. R. Swail, 2005: Methods to homogenize wind speeds from ships and buoys. Int. J. Climatol., 25, 979995, https://doi.org/10.1002/joc.1176.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Verspeek, J., A. Stoffelen, M. Portabella, A. Verhoef, and J. Vogelzang, 2008: ASCAT scatterometer ocean calibration. Int. Geoscience and Remote Sensing Symp., Boston, MA, IEEE, V248–V251, https://doi.org/10.1109/IGARSS.2008.4780074.

    • Crossref
    • Export Citation
  • Vinoth, J., and I. R. Young, 2011: Global estimates of extreme wind speed and wave height. J. Climate, 24, 16471665, https://doi.org/10.1175/2010JCLI3680.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wentz, F. J., 1983: A model function for ocean microwave brightness temperatures. J. Geophys. Res., 88, 18921908, https://doi.org/10.1029/JC088iC03p01892.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yang, X., X. Li, W. G. Pichel, and Z. Li, 2011: Comparison of ocean surface winds from ENVISAT ASAR, MetOp ASCAT Scatterometer, buoy measurements, and NOGAPS model. IEEE Trans. Geosci. Remote Sens., 49, 47434750, https://doi.org/10.1109/TGRS.2011.2159802.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Young, I. R., 1999: Seasonal variability of the global ocean wind and wave climate. Int. J. Climatol., 19, 931950, https://doi.org/10.1002/(SICI)1097-0088(199907)19:9<931::AID-JOC412>3.0.CO;2-O.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Young, I. R., and M. A. Donelan, 2018: On the determination of global ocean wind and wave climate from satellite observations. Remote Sens. Environ., 215, 228241, https://doi.org/10.1016/j.rse.2018.06.006.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Young, I. R., and A. Ribal, 2019: Multiplatform evaluation of global trends in wind speed and wave height. Science, 364, 548552, https://doi.org/10.1126/science.aav9527.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Young, I. R., A. V. Babanin, and S. Zieger, 2011: Response to comment on “Global trends in wind speed and wave height.” Science, 334, 905, https://doi.org/10.1126/science.1210548.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Young, I. R., E. Sanina, and A. V. Babanin, 2017: Calibration and cross validation of a global wind and wave database of altimeter, radiometer, and scatterometer measurements. J. Atmos. Oceanic Technol., 34, 12851306, https://doi.org/10.1175/JTECH-D-16-0145.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zeng, L., and R. A. Brown, 1998: Scatterometer observations at high wind speeds. J. Appl. Meteor., 37, 14121420, https://doi.org/10.1175/1520-0450(1998)037<1412:SOAHWS>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zieger, S., 2010: Long term trends in ocean wind speed and wave height. Ph.D. thesis, Swinburne University of Technology, 177 pp.

  • Zieger, S., J. Vinoth, and I. R. Young, 2009: Joint calibration of multiplatform altimeter measurements of wind speed and wave height over the past 20 years. J. Atmos. Oceanic Technol., 26, 25492564, https://doi.org/10.1175/2009JTECHA1303.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zieger, S., A. V. Babanin, W. Erick Rogers, and I. R. Young, 2015: Observation-based source terms in the third-generation wave model WAVEWATCH. Ocean Modell., 96, 225, https://doi.org/10.1016/j.ocemod.2015.07.014.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery
    Fig. 1.

    Characteristics of the four groups of scatterometers, including antenna configuration, polarization, swath configuration, incidence angles, and missions.

  • View in gallery
    Fig. 2.

    Durations of all scatterometer data from the seven satellite missions.

  • View in gallery
    Fig. 3.

    Locations of the NODC buoys (blue dots) used in this study in which only buoys more than 50 km offshore are used. Green shaded regions indicate the locations for MetOp-A and MetOp-B cross validations (see Fig. 10a).

  • View in gallery
    Fig. 4.

    Calibration of scatterometer wind speed against NODC buoy data. Shown are the 1:1 agreement (dashed diagonal line) and the RMA regression (thick solid line). Contours show the density of matchup data points, which has been normalized such that the maximum value is 1.0. Contours are drawn at 0.9, 0.7, 0.5, 0.4, 0.3, 0.2, 0.1, and 0.05. Dots represent outliers excluded from the RMA regression.

  • View in gallery
    Fig. 5.

    Q–Q plots between the scatterometer and NODC buoy data for wind speed after the calibration was applied.

  • View in gallery
    Fig. 6.

    Q–Q plots between the scatterometer and NODC buoy data for wind speed after the linear calibration and high-wind correction were applied.

  • View in gallery
    Fig. 7.

    Comparison between scatterometer and NODC for wind direction. The 1:1 agreement line is shown (thick solid line). Contours show the density of matchup data points, which has been normalized such that the maximum value is 1.0. Contours are drawn at 0.9, 0.7, 0.5, 0.4, 0.3, 0.2, 0.1, and 0.05.

  • View in gallery
    Fig. 8.

    Locations of offshore platforms used for validation of scatterometer data.

  • View in gallery
    Fig. 9.

    Q–Q plots between the calibrated scatterometer and platform data for wind speed.

  • View in gallery
    Fig. 10.

    Cross-validation matchup plots between the scatterometers for wind speed. Shown are the 1:1 agreement (dashed diagonal line) and the RMA regression (thick solid line). Contours show the density of matchup data points, which has been normalized such that the maximum value is 1.0. Contours are drawn at 0.9, 0.7, 0.5, 0.4, 0.3, 0.2, 0.1, and 0.05. Dots represent outliers excluded from the RMA regression.

  • View in gallery
    Fig. 11.

    Q–Q plots between the scatterometers for wind speed. (a) MetOp-AMetOp-B, (b) MetOp-A–RapidScat, (c) MetOp-B–RapidScat, (d) MetOp-BOceanSat-2, (e) MetOp-A–QuikSCAT, and (f) OceanSat-2ERS-2.

  • View in gallery
    Fig. 12.

    Scatterometer–scatterometer difference for wind speed as a function of time. (a) MetOp-AMetOp-B, (b) MetOp-A–RapidScat, (c) MetOp-B–RapidScat, (d) MetOp-BOceanSat-2, (e) MetOp-A–QuikSCAT, and (f) OceanSat-2ERS-2.

  • View in gallery
    Fig. 13.

    Cross-validation matchup plots between scatterometers and altimeters for wind speed. Shown are the 1:1 agreement (dashed diagonal line) and the RMA regression (thick solid line). Contours show the density of matchup data points, which has been normalized such that the maximum value is 1.0. Contours are drawn at 0.9, 0.7, 0.5, 0.4, 0.3, 0.2, 0.1, and 0.05. Dots represent outliers excluded from the RMA regression.

All Time Past Year Past 30 Days
Abstract Views 69 0 0
Full Text Views 801 374 41
PDF Downloads 636 217 9

Calibration and Cross Validation of Global Ocean Wind Speed Based on Scatterometer Observations

Agustinus RibalDepartment of Infrastructure Engineering, University of Melbourne, Parkville, Victoria, Australia, and Department of Mathematics, Faculty of Mathematics and Natural Sciences, Hasanuddin University, Makassar, Indonesia

Search for other papers by Agustinus Ribal in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0002-3286-5565
and
Ian R. YoungDepartment of Infrastructure Engineering, University of Melbourne, Parkville, Victoria, Australia

Search for other papers by Ian R. Young in
Current site
Google Scholar
PubMed
Close
Free access

Abstract

Global ocean wind speed observed from seven different scatterometers, namely, ERS-1, ERS-2, QuikSCAT, MetOp-A, OceanSat-2, MetOp-B, and Rapid Scatterometer (RapidScat) were calibrated against National Data Buoy Center (NDBC) data to form a consistent long-term database of wind speed and direction. Each scatterometer was calibrated independently against NDBC buoy data and then cross validation between scatterometers was performed. The total duration of all scatterometer data is approximately 27 years, from 1992 until 2018. For calibration purposes, only buoys that are greater than 50 km offshore were used. Moreover, only scatterometer data within 50 km of the buoy and for which the overpass occurred within 30 min of the buoy recording data were considered as a “matchup.” To carry out the calibration, reduced major axis (RMA) regression has been applied where the regression minimizes the size of the triangle formed by the vertical and horizontal offsets of the data point from the regression line and the line itself. Differences between scatterometer and buoy data as a function of time were investigated for long-term stability. In addition, cross validation between scatterometers and independent altimeters was also performed for consistency. The performance of the scatterometers at high wind speeds was examined against buoy and platform measurements using quantile–quantile (Q–Q) plots. Where necessary, corrections were applied to ensure scatterometer data agreed with the in situ wind speed for high wind speeds. The resulting combined dataset is believed to be unique, representing the first long-duration multimission scatterometer dataset consistently calibrated, validated and quality controlled.

© 2020 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Ian R. Young, ian.young@unimelb.edu.au

Abstract

Global ocean wind speed observed from seven different scatterometers, namely, ERS-1, ERS-2, QuikSCAT, MetOp-A, OceanSat-2, MetOp-B, and Rapid Scatterometer (RapidScat) were calibrated against National Data Buoy Center (NDBC) data to form a consistent long-term database of wind speed and direction. Each scatterometer was calibrated independently against NDBC buoy data and then cross validation between scatterometers was performed. The total duration of all scatterometer data is approximately 27 years, from 1992 until 2018. For calibration purposes, only buoys that are greater than 50 km offshore were used. Moreover, only scatterometer data within 50 km of the buoy and for which the overpass occurred within 30 min of the buoy recording data were considered as a “matchup.” To carry out the calibration, reduced major axis (RMA) regression has been applied where the regression minimizes the size of the triangle formed by the vertical and horizontal offsets of the data point from the regression line and the line itself. Differences between scatterometer and buoy data as a function of time were investigated for long-term stability. In addition, cross validation between scatterometers and independent altimeters was also performed for consistency. The performance of the scatterometers at high wind speeds was examined against buoy and platform measurements using quantile–quantile (Q–Q) plots. Where necessary, corrections were applied to ensure scatterometer data agreed with the in situ wind speed for high wind speeds. The resulting combined dataset is believed to be unique, representing the first long-duration multimission scatterometer dataset consistently calibrated, validated and quality controlled.

© 2020 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Ian R. Young, ian.young@unimelb.edu.au

1. Introduction

Accurate and high-resolution ocean wind data are critically important for many applications such as regional weather forecasting, ocean energy development, validation of numerical models, marine disaster monitoring and wind climatology (Ribal and Young 2019; Yang et al. 2011; Young and Ribal 2019; Young et al. 2017). Today, with the number of satellite platforms that have been launched, high-resolution ocean wind data are available through altimeters, radiometers, scatterometers, and synthetic aperture radars (Young et al. 2017). However, to form consistent long-duration records, such satellite data observations must be consistently calibrated, validated, and cross validated. Examples of multiplatform datasets for both altimeter and radiometer observations, calibrated, and cross validated in a consistent manner include Young et al. (2017) and Ribal and Young (2019).

Radiometers measure brightness temperature over a broad swath, which is related to wind speed through a complicated radiative transfer equation (Wentz 1983). Altimeters measure the radar cross section σ0 (ratio of transmitted to received radar energy) along a nadir track below the satellite, which is then related to wind speed. Scatterometers also measure radar cross section but over a broad swatch that depends on its antenna configuration. As a result of the antenna configuration, and the fact that measurements are made at a range of azimuth angles, scatterometers can measure both wind speed and direction. The resolution of the scatterometer measurements within the broad ground-track swath is typically either 25 or 12.5 km (Crapolicchio et al. 2012; Figa-Saldaña et al. 2002; Spencer et al. 2000).

Cross-platform evaluation between scatterometer radar cross section σ0 measurements have been undertaken in a number of studies (Alsabah et al. 2018; Madsen and Long 2016; Portabella et al. 2007; Stoffelen 1999) while other studies have concentrated on comparison of the derived wind speed against buoy data (Accadia et al. 2007; Bentamy 2008; Bentamy et al. 2002; Ebuchi et al. 2002; Pensieri et al. 2010; Pickett et al. 2003; Satheesan et al. 2007; Verspeek et al. 2008). Note that σ0 can be related to wind speed at a reference height of 10 m U10 through a geophysical model function (GMF). In addition, some calibrations have also been performed against other scatterometer missions—so-called intercalibration (Elyouncha and Neyt 2013a,b; Holbach and Bourassa 2017; Yang et al. 2011). These calibration studies tend to be confined to a single scatterometer platform or, occasionally a small number of scatterometers. We are unaware of a consistent multimission calibration across a large number of platforms. Such a consistent multiplatform dataset is necessary for long-duration climate studies (Young and Ribal 2019; Young et al. 2011).

In this study, the wind speed from seven different scatterometer missions, namely ERS-1, ERS-2, QuikSCAT, MetOp-A, OceanSat-2, MetOp-B, and Rapid Scatterometer (RapidScat) (expressed in the order of launch) have been calibrated, validated and cross validated. The calibration is based on National Data Buoy Center (NDBC) buoy data. Once the calibrations have been performed for all scatterometers, the data are validated at high wind speeds using wind data from offshore oil platform observations sourced from the Norwegian Meteorological Institute. Since scatterometers also measure wind direction, in contrast to radiometers and altimeters, wind direction from the scatterometers and buoy data were also compared. Furthermore, the differences between scatterometer observations and buoy measurements as a function of time are also investigated to determine long-term stability of the measurements. In addition, for stability and consistency checks, cross validation between different scatterometers as well as between scatterometers and altimeters were also carried out.

This paper is arranged as follows. Following this introduction, a brief description of scatterometers and the sources of the original datasets are presented in section 2. This is followed in sections 3 and 4 by a description of the NDBC buoy data used and the calibration procedures for the scatterometers against buoy data. Section 5 presents the validation of the calibrated scatterometer data with respect to the platform measurements. Furthermore, cross validation between different scatterometers and the cross validation between scatterometers and altimeters are presented in sections 6 and 7, respectively. Finally, conclusions are drawn in section 8.

2. Scatterometer description and data

a. Scatterometer overview

In general, the scatterometers used in this work can be classified based on the frequency band in which they operated, namely, Ku band and C band. C-band scatterometers including ERS-1, ERS-2, MetOp-A, and MetOp-B were launched by the European Space Agency (ESA). Ku-band scatterometers include QuikSCAT and RapidScat, which were launched by the National Aeronautics and Space Administration (NASA), and OceanSat-2, which was launched by the Indian Space Research Organization (ISRO). A further grouping of scatterometers can be made based on antenna configuration, polarization and ground swath configuration. This yields four groups: European Space Agency Scatterometer (ESCAT), SeaWinds, Advanced Scatterometer (ASCAT), and OceanSat Scatterometer (OSCAT; scatterometer onboard OceanSat-2). The details of each scatterometer group is presented below.

The ESA’s ESCAT scatterometer operated in the C band at a frequency of 5.3 GHz. The ESCAT was carried on both ERS-1 and ERS-2. ERS-1 was launched on 17 July 1991 but scatterometer data from the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT) is only available from 1992. Similarly, ERS-2 was launched on 21 April 1995 but the data sourced from ESA dates from 1996. The scatterometers carried on ERS-1 and ERS-2 not only had the same frequency but also had the same antenna configuration, incidence antenna angle and polarization as shown in Fig. 1. Moreover, both of them have the same repeat mission, altitude, and inclination angle, which are 35 days, 785 km, and 98.5°, respectively, as presented in Table 1. The azimuth angles are 45°, 90°, and 135° for the forward, sideways, and afterward antennas, respectively (Fig. 1) (Crapolicchio et al. 2012).

Fig. 1.
Fig. 1.

Characteristics of the four groups of scatterometers, including antenna configuration, polarization, swath configuration, incidence angles, and missions.

Citation: Journal of Atmospheric and Oceanic Technology 37, 2; 10.1175/JTECH-D-19-0119.1

Table 1.

Scatterometer operating characteristics for seven scatterometers, including repeat mission period, orbit parameters, antenna properties, launch date, operational time for which data are available, and responsible organization.

Table 1.

The SeaWinds scatterometer was developed by NASA and uses the Ku band with a frequency of 13.4 GHz. SeaWinds was carried on QuikSCAT and RapidScat. These scatterometers were launched on 19 June 1999 and on 20 September 2014, respectively (Ebuchi et al. 2002). The QuikSCAT mission ended on 23 November 2009 due to an age-related mechanical failure. This instrument was then replaced by RapidScat. However, the later scatterometer only has data available until 2016. It should be mentioned that, unlike other scatterometers, RapidScat is the only scatterometer that was in a non-sun-synchronous orbit, meaning that RapidScat passes over different locations on the ground at the same solar time of day. The antenna configuration of these scatterometers uses a rotating dish with two spot beams that conically sweep producing a circular pattern on the surface as shown in Fig. 1. The swaths for the inner and the outer beams are 1400 and 1800 km, respectively, centered on the nadir. The repeat mission for QuikSCAT is four days, while the repeat mission for RapidScat is irregular due to its non-sun-synchronous orbit (Madsen and Long 2016). Hence, the inclination and the altitude for QuikSCAT and RapidScat are also different, with the inclination and the altitude for QuikSCAT being 98.6° and 803 km, respectively, while the inclination of RapidScat is 51.6° and its altitude varied from 375 to 435 km as can be seen from Table 1 (Ebuchi et al. 2002; Satheesan et al. 2007).

ASCAT was designed based on the ESCAT scatterometer. Hence, ASCAT and ESCAT have some common features such as azimuth angle, polarization, radar frequency, and the operational product resolution (25 km). However, ASCAT introduced some new key additional features such as an increasing spatial coverage by using a double swath, an increased incidence angle range, improved instrument design for higher stability and reliability, and an improved onboard processing concept for lower data rates. Moreover, the incidence angles and the swath size were also increased from 18°–59° to 25°–65° and from 500 to 550 km, respectively, as shown in Fig. 1 (Figa-Saldaña et al. 2002). In addition, the altitude of ASCAT was also increased from 785 km for ESCAT to 817 km. Similarly, the inclination of ASCAT was slightly increased from 98.5° for ESCAT to 98.7° as summarized in Table 1 (Elyouncha and Neyt 2013a; Figa-Saldaña et al. 2002). The swaths of ASCAT are separated by approximately 360 km from the satellite ground track while the distance of the swath to the satellite track of ESCAT is 200 km (Crapolicchio et al. 2012). The ASCAT scatterometer was used on two different satellites, namely, MetOp-A and MetOp-B, which were launched on 19 October 2006 and on 17 September 2012, respectively. These scatterometer missions are still ongoing. It should be noted that MetOp-B follows MetOp-A with a 49-min delay in a tandem configuration (Elyouncha and Neyt 2013b).

The fourth scatterometer group is OSCAT, which was developed by the ISRO and launched on OceanSat-2 on 23 September 2009. The OSCAT scatterometer is also carried onboard ScatSat-1. However, since the data from this mission are not available in the public domain, it has not been included in this present work. The OSCAT scatterometer carried on OceanSat-2 operates in the Ku band at a frequency of 13.515 GHz. Its altitude and inclination are 720 km and 98.25°, respectively. The repeat cycle is 2 days and the incidence angles are 48.9° and 57.6° for HH and VV polarization, respectively. The antenna configuration of this scatterometer is the same as SeaWinds but the outer beam swath has been increased by 40 km from 1800 to 1840 km where the details can be found in Table 1 and Fig. 1 (Chakraborty et al. 2013).

b. Scatterometer data sources

The scatterometer data were sourced from three different public domain sites: QuikSCAT, MetOp-A, OceanSat-2, MetOp-B, and RapidScat were obtained from PO.DAAC (https://podaac.jpl.nasa.gov/); ERS-1 and ERS-2 were sourced from EUMETSAT (http://archive.eumetsat.int) and ESA (https://earth.esa.int), respectively. The total duration of all scatterometer missions is approximately 27 years, from 1992 until 2018 where the period of each scatterometer is presented in Fig. 2 and Table 1. The resolution of ERS-2, QuikSCAT, OceanSat-2, and RapidScat is 12.5 km while the data resolution of the other scatterometers, namely, ERS-1, MetOp-A, and MetOp-B, is 25 km.

Fig. 2.
Fig. 2.

Durations of all scatterometer data from the seven satellite missions.

Citation: Journal of Atmospheric and Oceanic Technology 37, 2; 10.1175/JTECH-D-19-0119.1

Each of the original scatterometer datasets contains a range of data quality flags; however, these flags are not consistent across the datasets. With the exception of ERS-2 and OceanSat-2, the datasets have three quality flags for wind speed, defining “good” data, “low wind speed” data, and “high wind speed” data. ERS-2 has only one quality flag for “good” wind speed while OceanSat-2 has five quality flags for “good” wind speed. Low and high wind speeds are defined as wind speeds that are less than 3 m s−1 and greater than 30 m s−1, respectively. Even though ERS-1, MetOp-A, and MetOp-B scatterometer data were obtained from different sources, they have the same quality flags for “good,” “low,” and “high” wind speed. Similarly, although QuikSCAT and RapidScat both classify the wind speed into three categories, they use different numerical values for the flags. Therefore, care needs to be taken in processing the various datasets. In compiling the present dataset, these flags have been simplified to a consistent set across all scatterometer missions (see appendixes A and B). It should be noted that high and low wind speed in the original dataset have been flagged as “Good_data” in the present database.

3. NDBC buoy data

As shown in Fig. 2, the duration of the combined scatterometer dataset is approximately 27 years. For calibration purposes, we require a high-quality database of buoy wind speed and direction over this period. In addition, the buoy data should be relatively far from land, in order to reduce land/island contamination due to the size of scatterometer vector cell (12.5 or 25 km). The most extensive dataset that satisfies these parameters is the NDBC buoy archive. These data have been quality controlled, and archived by the National Oceanographic Data Center (NODC; https://data.nodc.noaa.gov/thredds/catalog/ndbc/cmanwx/catalog.html), where they are available in the public domain under NOAA’s National Centers for Environmental Information (NCEI). Another reason to choose NDBC buoy data is due to the fact that it is a long-duration record and generally regarded as being consistent over time (Durrant et al. 2009). Even though the NDBC buoy locations are geographically limited to the Northern Hemisphere, it has been shown using altimeter data that this does not have a significant impact of the mean calibration (Young and Donelan 2018). Moreover, although the validity of high-wind-speed buoy data has been questioned (Alves and Young 2003; Bender et al. 2010; Jensen et al. 2015; Large et al. 1995; Taylor and Yelland 2001; Vinoth and Young 2011; Zeng and Brown 1998), the data have been widely used for validation of model results and calibration of satellite observation and have been found to be high quality (Evans et al. 2003; Ribal and Young 2019; Zieger et al. 2015).

To avoid land/island contamination, only NODC moored buoy data that were more than 50 km from the coastline were used as a source for calibration wind speed (Zieger et al. 2009). The NODC data after 2011, contained a series of quality flags for wind speed, namely 0, 1, 2, and 3, which represent quality_good, out_of_range, sensor_nonfunctional, and questionable, respectively. To obtain high-quality buoy wind speed data, only values flagged “0” were used for the calibration of the scatterometers. Although NODC data before 2011 do not have quality flags, examination of the data indicates few clear outliers (<0.5%; see Table 2). The locations of the buoy data used in the calibration are shown in Fig. 3.

Table 2.

Calibration relationships for scatterometer wind speed, obtained from the RMA regression. U10* is the calibrated value and U10 is the uncalibrated data. Also shown are the confidence limits on the regression, number of points n, and the percentage of outliers from the robust regression.

Table 2.
Fig. 3.
Fig. 3.

Locations of the NODC buoys (blue dots) used in this study in which only buoys more than 50 km offshore are used. Green shaded regions indicate the locations for MetOp-A and MetOp-B cross validations (see Fig. 10a).

Citation: Journal of Atmospheric and Oceanic Technology 37, 2; 10.1175/JTECH-D-19-0119.1

Buoy wind speeds are measured at a variety of different heights depending on the anemometer height Uz. For calibration purposes, the wind speed is required at a standard reference height of 10 m U10. Assuming a neutral-stability-logarithm boundary layer (e.g., Priestley 1959; Young 1999), U10 can be approximated by
U10=Uzκ2Cd1ln(z/z0),
where κ is the von Kármán constant, which is approximately 0.4, Cd is the drag coefficient, and z0 is the roughness length. Measurements of Cd over the ocean yield results with scatter over an order of magnitude, and much research has focused on the wind speed and sea-state dependence of Cd (Donelan 1982; Guan and Xie 2004; Young 1999). In this work, Cd = 1.2 × 10−3 and z0 = 9.7 × 10−5 m have been assumed. As mentioned in previous studies (Young et al. 2017), a different assumption of Cd does not have a major impact on the final satellite wind speed (Zieger et al. 2009). For a more detailed description of NOAA buoy data, one can refer to Zieger (2010). This choice of boundary layer correction is consistent with previous altimeter calibrations (Zieger et al. 2009). Alternative approaches to address boundary layer correction have been proposed by, for example, Thomas et al. (2005) and Yang et al. (2011).

4. Calibration against NDBC buoy data

To calibrate scatterometer wind speed, it is necessary to determine cases where buoy anemometer measurements were made at approximately the same time as there was a satellite overflight—termed a matchup. A matchup is considered to have occurred if the following criteria are satisfied:

  1. The scatterometer wind vector cell location is within 50 km of the buoy location and the time difference between the buoy measurement and the scatterometer overflight is less than 30 min.

  2. Only buoys that are more than 50 km from the coastline were used so as to avoid land/islands contamination for both buoy and scatterometer.

  3. A minimum of five scatterometer wind vector cells were required within a 50-km-radius region around the buoy.

  4. Large variability in scatterometer wind speeds were excluded. Matchups were excluded if σ(U10)/U¯10>0.2, where σ(U10) and U¯10 are the standard deviation and mean, respectively, of scatterometer wind speed vector cells, within a 50-km radius around the buoy.

For each scatterometer, matchup data across all the buoys were pooled and a linear regression analysis performed between the buoy and scatterometer wind speeds U10. It should be noted that wind speeds greater than 60 m s−1 were excluded from the analysis, as such data are questionable (a very rare occurrence in the dataset). For calibration purposes, buoy data are usually considered as “ground truth.” However, buoy data are not free from errors due to their sampling variability and instrument accuracy (Young et al. 2017; Zieger et al. 2009). As a result, a conventional linear regression analysis is not appropriate. To this end, a reduced major axis (RMA) regression was used (Harper 2014). This regression minimizes the triangular area bounded by the vertical and horizontal offsets between the data point and the regression line and the cord of the regression line. This is in contrast to a conventional regression, which minimizes the vertical axis offset from the regression line. In addition, standard least squares regression analysis is highly sensitive to outliers. Such outliers can be removed by the use of robust regression (Holland and Welsch 1977). Robust regression assigns a weight to each point, with values between 0 and 1. Points with a value less than 0.01 were designated as outliers and removed from the analysis before applying the RMA regression analysis.

To evaluate the performance of the calibration, four different statistical parameters were evaluated, namely, bias B, root-mean-square error (RMSE), Pierson’s correlation coefficient ρ, and scatter index (SI). These parameters were evaluated using the following relationships where M and O stand for model and observation (Ribal and Young 2019; Zieger et al. 2015), respectively, and N is the number of matchup points:
B=1Ni=1N(MiOi),
RMSE=1Ni=1N(MiOi)2,
SI=1Ni=1N(MiOiB)21Ni=1NOi,
ρ=cov(M,O)cov(M)cov(O).

The matchup data across all the NDBC buoys were pooled for each of the seven scatterometers and the linear RMA regression undertaken. Figure 4 shows example results for four cases (ERS-1, MetOp-A, QuikSCAT, and OceanSat-2).

Fig. 4.
Fig. 4.

Calibration of scatterometer wind speed against NODC buoy data. Shown are the 1:1 agreement (dashed diagonal line) and the RMA regression (thick solid line). Contours show the density of matchup data points, which has been normalized such that the maximum value is 1.0. Contours are drawn at 0.9, 0.7, 0.5, 0.4, 0.3, 0.2, 0.1, and 0.05. Dots represent outliers excluded from the RMA regression.

Citation: Journal of Atmospheric and Oceanic Technology 37, 2; 10.1175/JTECH-D-19-0119.1

It is clear that the scatterometer values of U10 agree well with the buoy data, with only small deviations for the 1:1 correlation line. Young et al. (2017) have previously investigated the calibration of QuikSCAT and the result shown in Fig. 4c is almost identical to this previous calibration (see Fig. 6b of Young et al. 2017). This occurs despite the fact that the source of the data and presumably the processing is different [PO.DAAC here vs Remote Sensing Systems for Young et al. (2017)].

Although scatterplots such as those shown in Fig. 4 provide a good overall assessment of platform performance, they offer little insight into the performance at high wind speed, where is a relatively small amount of buoy data. In these cases, it is more insightful to examine quantile–quantile (Q–Q) plots between the buoy data and scatterometer data. Based on such an analysis, it was clear that three of the scatterometers overestimate high wind speed compared to buoys, namely, QuikSCAT, OceanSat-2 and RapidScat. Figure 5 shows Q–Q results for QuikSCAT and RapidScat. Interestingly, all three scatterometers operate in the Ku band.

Fig. 5.
Fig. 5.

Q–Q plots between the scatterometer and NODC buoy data for wind speed after the calibration was applied.

Citation: Journal of Atmospheric and Oceanic Technology 37, 2; 10.1175/JTECH-D-19-0119.1

Young et al. (2017) noted very similar high-wind-speed behavior for radiometers and Takbash et al. (2019) proposed the empirical correction:
U10*=U10[1a(U10b)0.79],
where U10 is the scatterometer/radiometer wind speed after linear calibration is applied, U10* is the corrected wind speed, and a and b are constants that depend on the individual scatterometer. The values of a and b were determined based on the Q–Q plot analysis between buoy measurement and scatterometer data after the linear calibration was applied. This relationship was tested for the QuikSCAT, OceanSat-2, and RapidScat scatterometers, which overestimated wind speed compared to buoys. Figure 6 shows example Q–Q plots for the same four scatterometers shown in Fig. 4. In the cases of QuikSCAT and OceanSat-2 (Figs. 6c,d) the empirical correction (6) results in much better high-wind-speed agreement with the buoy data. The final scatterometer calibration relations, including the high-wind-speed correction equations are summarized in Table 2.
Fig. 6.
Fig. 6.

Q–Q plots between the scatterometer and NODC buoy data for wind speed after the linear calibration and high-wind correction were applied.

Citation: Journal of Atmospheric and Oceanic Technology 37, 2; 10.1175/JTECH-D-19-0119.1

As noted above, scatterometers also measure wind direction. Hence, comparisons of wind direction between buoy and scatterometer data were also undertaken for each of the scatterometers. The same collocation process used for wind speed was again adopted for wind direction. That is, only wind direction data within 50 km of buoys and overpasses within 30 min of the buoy data recording were used for matchups. For all seven scatterometers, there was excellent agreement with the buoys for wind direction (see statistics on the Fig. 7). Figure 7 again shows scatterplots for the four scatterometers considered earlier (ERS-1, MetOp-A, QuikSCAT, and OceanSat-2). Based on these results, no attempt was made to calibrate or alter the wind direction measurements provided for each of the scatterometers.

Fig. 7.
Fig. 7.

Comparison between scatterometer and NODC for wind direction. The 1:1 agreement line is shown (thick solid line). Contours show the density of matchup data points, which has been normalized such that the maximum value is 1.0. Contours are drawn at 0.9, 0.7, 0.5, 0.4, 0.3, 0.2, 0.1, and 0.05.

Citation: Journal of Atmospheric and Oceanic Technology 37, 2; 10.1175/JTECH-D-19-0119.1

5. Validation against platform data measurements

To perform an independent validation of the calibrated scatterometer data, we utilized platform data provided by the Norwegian Meteorological Institute for offshore oil platforms at the locations shown in Fig. 8. As can be seen from the figure, there are 10 different locations in which platform data are available. The time period of the platform measurement is from 1999 until 2016. These data have been extensively studied by the Norwegian oil industries and found to be reliable (Takbash et al. 2019). To perform the validation, again, the same matchup criteria as for the buoy data have be applied. That is, only scatterometer data that are within 50 km of the platform and with a time difference less than 30 min are used. Once, the matchups were obtained, the platform measurement data were compared with the calibrated scatterometer data, using the calibration relationships as presented in Table 2.

Fig. 8.
Fig. 8.

Locations of offshore platforms used for validation of scatterometer data.

Citation: Journal of Atmospheric and Oceanic Technology 37, 2; 10.1175/JTECH-D-19-0119.1

The advantage of using platform data for this validation is that the characteristics of the data are quite different to buoy data. Typically, the data are recorded at a much greater height than the buoy data (again a boundary layer correction was applied). In addition, platform data do not suffer from issues of sheltering of the anemometer in high sea states, which has cast doubts on the accuracy of buoy measurements at high wind speeds (Bender et al. 2010; Jensen et al. 2015; Large et al. 1995; Taylor and Yelland 2001; Zeng and Brown 1998). Platform data do, however, potentially suffer from blockage effects caused by the platform. In this regard, the present dataset has been extensively validated to reliably define the 10 m wind speed U10 by the Norwegian oil industry.

Scatter and Q–Q plots between the scatterometer and platform data were undertaken for each of the scatterometers and Fig. 9 shows example Q–Q plots for MetOp-A and MetOp-B. For all cases, the Q–Q plots showed good agreement between calibrated scatterometer and platform data. All scatterometers measured slightly higher values than the platform data, indicating that the platform data are generally slightly lower than the buoy data. Importantly, however, the Q–Q plots show good agreement at high wind speeds for all calibrated scatterometers, indicating the high-wind-speed corrections applied for QuikSCAT, OceanSat-2, and RapidScat are appropriate.

Fig. 9.
Fig. 9.

Q–Q plots between the calibrated scatterometer and platform data for wind speed.

Citation: Journal of Atmospheric and Oceanic Technology 37, 2; 10.1175/JTECH-D-19-0119.1

6. Cross validation between scatterometers

Cross validation between the scatterometers was carried out to check the consistency and stability of the calibrated scatterometers. Based on Fig. 2, there are nine possible cross validations that can be performed: RapidScat–MetOp-A, RapidScat–MetOp-B, OceanSat-2MetOp-A, OceanSat-2MetOp-B, QuikSCAT–MetOp-A, QuikSCAT–ERS-2, ERS-2OceanSat-2, ERS-2MetOp-A, and MetOp-AMetOp-B. The same matchup criteria as for the buoys was again applied for the cross validation (i.e., 50-km spatial and 30-min temporal separations). However, MetOp-A and MetOp-B are in the same tandem orbit but with a 49-min time delay (Elyouncha and Neyt 2013b). Therefore, the 30-min separation criteria would result in no matchups. To address this issue, the collocation time between MetOp-A and MetOp-B was increased from 30 to 60 min. As the number of resulting matchups between MetOp-A and MetOp-B is enormous, only eighteen 15° × 15° locations around the world were selected to extract matchups (see Fig. 3). After applying all the quality criteria, this resulted in 1 769 942 matchups between MetOp-A and MetOp-B. Figure 10 shows example cross validations for MetOp-AMetOp-B, MetOp-A–RapidScat, MetOp-B–RapidScat, MetOp-BOceanSat-2, MetOp-A–QuikSCAT, and OceanSat-2ERS-2. The agreement between the instruments is generally excellent (see statistics on the Fig. 10). A further analysis of the corresponding Q–Q plots are shown in Fig. 11. Again, the agreement is excellent, including at high wind speeds. In addition, the difference between the scatterometers as a function of time is shown in Fig. 12, indicating there are no changes in the calibration as a function of time.

Fig. 10.
Fig. 10.

Cross-validation matchup plots between the scatterometers for wind speed. Shown are the 1:1 agreement (dashed diagonal line) and the RMA regression (thick solid line). Contours show the density of matchup data points, which has been normalized such that the maximum value is 1.0. Contours are drawn at 0.9, 0.7, 0.5, 0.4, 0.3, 0.2, 0.1, and 0.05. Dots represent outliers excluded from the RMA regression.

Citation: Journal of Atmospheric and Oceanic Technology 37, 2; 10.1175/JTECH-D-19-0119.1

Fig. 11.
Fig. 11.

Q–Q plots between the scatterometers for wind speed. (a) MetOp-AMetOp-B, (b) MetOp-A–RapidScat, (c) MetOp-B–RapidScat, (d) MetOp-BOceanSat-2, (e) MetOp-A–QuikSCAT, and (f) OceanSat-2ERS-2.

Citation: Journal of Atmospheric and Oceanic Technology 37, 2; 10.1175/JTECH-D-19-0119.1

Fig. 12.
Fig. 12.

Scatterometer–scatterometer difference for wind speed as a function of time. (a) MetOp-AMetOp-B, (b) MetOp-A–RapidScat, (c) MetOp-B–RapidScat, (d) MetOp-BOceanSat-2, (e) MetOp-A–QuikSCAT, and (f) OceanSat-2ERS-2.

Citation: Journal of Atmospheric and Oceanic Technology 37, 2; 10.1175/JTECH-D-19-0119.1

It is clear in Figs. 10 and 12 that there is some difference in mean values of some of the scatterometers. The cross validations of MetOp-A–RapidScat, MetOp-BOceanSat-2, MetOp-B–RapidScat, and MetOp-A–QuikSCAT all show such a difference. In each case, these represent comparisons of C-band scatterometers versus Ku-band scatterometers, with the Ku-band scatterometers giving slightly lower wind speeds. As shown by Young and Donelan (2018), satellite remote sensing measurements of wind speed are impacted by boundary layer stability, with the magnitude of the impact influenced by the frequency of the instrument under consideration. Although all the scatterometers were calibrated against an extensive buoy dataset, the cross-validation matchups are across a much wider range of latitudes and hence a more extensive span of air and water temperatures (i.e., different atmospheric stability). This accounts for the difference seen here, with similar differences previously reported for altimeter and radiometer data (Young and Donelan 2018).

7. Cross validation between scatterometers and altimeters

In addition to the cross validation between scatterometers, a further consistency and stability check was undertaken by cross validating against altimeter data. For this purpose, the calibrated altimeter dataset of Ribal and Young (2019) was used. There are a large number of possible validation overlaps between these two satellite datasets. For the present purposes, a total of 22 combinations of altimeter and scatterometer were considered. These combinations involved six scatterometers: ERS-1, QuikSCAT, MetOp-A, OceanSat-2, MetOp-B, and RapidScat and seven altimeters: TOPEX, Jason-1, Jason-2, CryoSat-2, SARAL, Jason-3, and Sentinel-3A. The same collocation criteria as previously applied have been used (i.e., 50-km spatial and 30-min temporal separations). Figure 13 shows examples of six combinations, again showing good agreement. In each case, however, the scatterometers produce slightly lower values of wind speed than the altimeters. As noted above, it is believed that this is the result of atmospheric stability and the fact that the altimeters and scatterometers operate in different frequency bands.

Fig. 13.
Fig. 13.

Cross-validation matchup plots between scatterometers and altimeters for wind speed. Shown are the 1:1 agreement (dashed diagonal line) and the RMA regression (thick solid line). Contours show the density of matchup data points, which has been normalized such that the maximum value is 1.0. Contours are drawn at 0.9, 0.7, 0.5, 0.4, 0.3, 0.2, 0.1, and 0.05. Dots represent outliers excluded from the RMA regression.

Citation: Journal of Atmospheric and Oceanic Technology 37, 2; 10.1175/JTECH-D-19-0119.1

The performance of the cross validations was analyzed using four different statistical parameters as given in Eqs. (2)(5). The values of the statistical parameters for all 22 cross validations are provided in the Tables 36. Note that C-2, TP, J-1, J-2, J-3, SA, and S-3 are abbreviated forms of the altimeters CryoSat-2, TOPEX, Jason-1, Jason-2, Jason-3, SARAL, and Sentinel-3A, respectively. Similarly, ER, MA, MB, OC, QU, and RA are abbreviated forms of scatterometers ERS-1, MetOp-A, MetOp-B, OceanSat-2, QuikSCAT, and RapidScat, respectively. Moreover, unlike Tables 35, Table 6 has to be read from row to column as bias is not a commutative parameter as shown in Eq. (2).

Table 3.

Root-mean-square error for cross validation between scatterometers and altimeters (m s−1).

Table 3.
Table 4.

Scatter index for cross validation between scatterometers and altimeters.

Table 4.
Table 5.

Correlation coefficient for cross validation between scatterometers and altimeters.

Table 5.
Table 6.

Bias for cross validation between scatterometers and altimeters (m s−1).

Table 6.

As shown in Tables 36, all four statistical parameters that have been used to justify the performance of the cross validation show good results, with the scatter index for all cases less than 0.13 and the correlation coefficient more than 0.95. Although some of the RMSE values are more than 1 m s−1, they are still acceptable as the other parameters are still good. This is particularly the case for OceanSat-2. There is a consistent negative bias (scatterometer lower than altimeter). This is consistent with the results shown in Figs. 10 and 12 and attributed above to the impact of atmospheric stability.

The resulting dataset has been archived with the Australian Ocean Data Network (AODN) and is available for public domain access. Details of this archive are provided in appendixes A and B.

8. Conclusions

Global ocean wind speed measured from seven different scatterometers have been calibrated, validated, and cross validated in this study, creating a consistent dataset spanning 27 years, from 1992 until 2018. This combined dataset is believed to be the first long-duration, multimission scatterometer dataset developed. Each scatterometer has been calibrated against in situ buoy data, from buoys that are more than 50 km from the coastline to avoid potential land contamination. To show the robust nature of the calibrations, the data have been validated against independent wind measurements obtained from offshore oil platforms. This validation shows that the calibrated scatterometer data are consistent with the platform data both for mean conditions and extreme wind speeds.

To check the consistency and long-term stability of the data, cross validations between scatterometers as well as between scatterometers and altimeters have also been performed. Nine combinations of scatterometers were cross validation showing consistent measurements between all platforms. Similarly, 22 combinations of altimeters and scatterometers involving six calibrated scatterometers and seven calibrated altimeters were cross validated. The results show that calibrated scatterometer data and calibrated altimeter data are in excellent agreement (see statistics on the Fig. 13). Comparisons between altimeter and scatterometer were quantified using four different statistical parameters, all of which showed excellent agreement (see statistics on the Fig. 13).

The combined dataset is a resource that can be used for a variety of purposes, including studies of climatology, long-term trends in wind speed and direction, and the determination of extreme wind speeds (see appendixes A and B).

Acknowledgments

The authors acknowledge ongoing support from the Australian Research Council through Grant DP160100738 and the Integrated Marine Observing System (IMOS) for support in development of this database. As noted in the paper, the original data were sourced from a range of public archives. These repositories are gratefully acknowledged. The platform wind data were supplied by Oyvind Breivik of the Norwegian Meteorological Institute.

APPENDIX A

Data Description

In the present database, eight physical parameters are archived as outlined in the following table. The data have been binned into 1° × 1° bins in a similar manner to the altimeter database of Ribal and Young (2019). Unlike altimeter measurements, scatterometers measure over a swath. As a result, the data in the database are stored in the form of a two-dimensional matrix in which the row index represents the number of points in the swath in the cross-track direction and the column index represents the measurement time (Table A1).

Table A1.

List of all parameters in the present scatterometer database.

Table A1.

In providing the data in netCDF format, we follow the Integrated Marine Observing System (IMOS) data protocol upon which the project is based (IMOS 2015a,b). In particular, the quality flags in the database have followed the IMOS standard flag system where 1, 2, 4, and 9 represent Good_data, Probably_good_data, Bad_data, and Missing_data, respectively. It should be noted that high- and low-wind-speed data flagged in the respective source datasets have been flagged as “Good_data” in the present database.

The file names follow the format IMOS_SRS-Surface-Waves_M_Wind-SCATTEROMETER_FV02_Lat-Lon-DM00.nc, where

  1. IMOS is the name of the project

  2. SRS-Surface-Waves represents the present facility

  3. M signifies meteorological related parameters

  4. Wind represents wind only

  5. SCATTEROMETER is the name of scatterometer (variable)

  6. FV02 represents the version of the file

  7. Lat is the latitude north or south of the most southern border of the 1° file (variable)

  8. Lon is the longitude of the most western border of the 1° file (variable)

  9. DM00 is the first version of delayed mode product

APPENDIX B

Data Location

The data can be accessed in three different ways from the archive.

a. AODN graphical portal

The database will be updated at approximately 6-month intervals and a dynamic archive is maintained at the Australian Ocean Data Network (AODN). The AODN portal can be accessed online (https://portal.aodn.org.au/). The user can access the data graphically from the portal. To find the data, the following navigation is recommended:

  1. Click the “Get Ocean Data Now” button.

  2. Scroll to the keyword search box at the bottom left of the screen. Enter the keywords “scatterometer winds.”

  3. Click on the thumbnail map of the world to the right. The graphical interface that opens allows the user to scroll to any area of the world and define a region to download with the mouse. The specific satellites to download can be specified in the menu to the left.

b. Direct interface

The dynamic archive can also be accessed directly as an Amazon S3 archive. It is recommended that this is done using software such as Cyberduck. Instructions to set up such a server can be found online (https://help.aodn.org.au/downloading-data-from-servers/amazon-s3-servers/; once access to the S3 server is gained, the user should navigate to IMOS/SRS/Surface-Waves/Wind-Scatterometry-DM00).

c. AODN thredds

The data can also be accessed from AODN thredds, which is very useful for Linux users (http://thredds.aodn.org.au/thredds/catalog/IMOS/SRS/Surface-Waves/Wind-Scatterometry-DM00/catalog.html).

REFERENCES

  • Accadia, C., S. Zecchetto, A. Lavagnini, and A. Speranza, 2007: Comparison of 10-m wind forecasts from a regional area model and QuikSCAT scatterometer wind observations over the Mediterranean Sea. Mon. Wea. Rev., 135, 19451960, https://doi.org/10.1175/MWR3370.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Alsabah, R., A. Al-Sabbagh, and J. Zec, 2018: RapidScat backscatter measurement validation. J. Appl. Remote Sens., 12, 034005, https://doi.org/10.1117/1.JRS.12.034005.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Alves, J. H. G. M., and I. R. Young, 2003: On estimating extreme wave heights using combined Geosat, TOPEX/Poseidon and ERS-1 altimeter data. Appl. Ocean Res., 25, 167186, https://doi.org/10.1016/j.apor.2004.01.002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bender, L. C., III, N. L. Guinasso Jr., J. N. Walpert, and S. D. Howden, 2010: A comparison of methods for determining significant wave heights—Applied to a 3-m discus buoy during Hurricane Katrina. J. Atmos. Oceanic Technol., 27, 10121028, https://doi.org/10.1175/2010JTECHO724.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bentamy, A., 2008: Characterization of ASCAT measurements based on buoy and QuikSCAT wind vector observations. Ocean Sci., 5, 77101, https://doi.org/10.5194/os-4-265-2008.

    • Search Google Scholar
    • Export Citation
  • Bentamy, A., Y. Quilfen, and P. Flament, 2002: Scatterometer wind fields: A new release over the decade 1991-2001. Can. J. Remote Sens., 28, 431449, https://doi.org/10.5589/m02-041.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chakraborty, A., R. Kumar, and A. Stoffelen, 2013: Validation of ocean surface winds from the OceanSat-2 scatterometer using triple collocation. Remote Sens. Lett., 4, 8493, https://doi.org/10.1080/2150704X.2012.693967.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crapolicchio, R., G. De Chiara, A. Elyouncha, P. Lecomte, X. Neyt, A. Paciucci, and M. Talone, 2012: ERS-2 scatterometer: Mission performances and current reprocessing achievements. IEEE Trans. Geosci. Remote Sens., 50, 24272448, https://doi.org/10.1109/TGRS.2011.2179808.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Donelan, M. A., 1982: The dependence of the aerodynamic drag coefficient on wave parameters. Proc. First Int. Conf. on Meteorology and Air–Sea Interaction of the Coastal Zone, Boston, MA, Amer. Meteor. Soc., 381–387.

  • Durrant, T. H., D. J. M. Greenslade, and I. Simmonds, 2009: Validation of Jason-1 and Envisat remotely sensed wave heights. J. Atmos. Oceanic Technol., 26, 123134, https://doi.org/10.1175/2008JTECHO598.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ebuchi, N., H. C. Graber, and M. J. Caruso, 2002: Evaluation of wind vectors observed by QuikSCAT/SeaWinds using ocean buoy data. J. Atmos. Oceanic Technol., 19, 20492062, https://doi.org/10.1175/1520-0426(2002)019<2049:EOWVOB>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Elyouncha, A., and X. Neyt, 2013a: C-band satellite scatterometer intercalibration. IEEE Trans. Geosci. Remote Sens., 51, 14781491, https://doi.org/10.1109/TGRS.2012.2217381.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Elyouncha, A., and X. Neyt, 2013b: Inter-calibration of MetOp-A and MetOp-B scatterometers using ocean measurements. Proc. SPIE, 8888, 888806, https://doi.org/10.1117/12.2033927.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Evans, D., C. Conrad, and F. Paul, 2003: Handbook of automated data quality control checks and procedures of the National Data Buoy Center. NOAA/NDBC Tech. Doc. 03-02, 44 pp.

  • Figa-Saldaña, J., J. J. Wilson, E. Attema, R. Gelsthorpe, M. Drinkwater, and A. Stoffelen, 2002: The Advanced Scatterometer (ASCAT) on the Meteorological Operational (MetOp) platform: A follow on for European wind scatterometers. Can. J. Remote Sens., 28, 404412, https://doi.org/10.5589/m02-035.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Guan, C., and L. Xie, 2004: On the linear parameterization of drag coefficient over sea surface. J. Phys. Oceanogr., 34, 28472851, https://doi.org/10.1175/JPO2664.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Harper, W. V., 2014: Reduced major axis regression: Teaching alternatives to least squares. Ninth Int. Conf. on Teaching Statistics, Flagstaff, AZ, International Statistical Institute, https://iase-web.org/icots/9/proceedings/pdfs/ICOTS9_C131_HARPER.pdf.

  • Holbach, H. M., and M. A. Bourassa, 2017: Platform and across-swath comparison of vorticity spectra from QuikSCAT, ASCAT-A, OSCAT, and ASCAT-B scatterometers. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 10, 22052213, https://doi.org/10.1109/JSTARS.2016.2642583.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Holland, P. W., and R. E. Welsch, 1977: Robust regression using iteratively reweighted least-squares. Commun. Stat. Theory Methods, 6, 813827, https://doi.org/10.1080/03610927708827533.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • IMOS, 2015a: IMOS netCDF conventions, version 1.4. University of Tasmania Rep., 82 pp., https://s3-ap-southeast-2.amazonaws.com/content.aodn.org.au/Documents/IMOS/Conventions/IMOS_NetCDF_Conventions.pdf.

  • IMOS, 2015b: IMOS netCDF file naming convention, version 1.5. University of Tasmania Rep., 22 pp., https://s3-ap-southeast-2.amazonaws.com/content.aodn.org.au/Documents/IMOS/Conventions/IMOS_NetCDF_File_Naming_Convention.pdf.

  • Jensen, R., V. Swail, R. Bouchard, R. Riley, T. Hesser, M. Blaseckie, and C. MacIsaac, 2015: Field Laboratory for Ocean Sea State Investigation and Experimentation: FLOSSIE: Intra-measurement evaluation of 6N wave buoy systems. 14th Int. Workshop on Wave Hindcasting and Forecasting/Fifth Coastal Hazard Symp, Key West, FL, JCOMM.

  • Large, W. G., J. Morzel, and G. B. Crawford, 1995: Accounting for surface wave distortion of the marine wind profile in low-level ocean storms wind measurements. J. Phys. Oceanogr., 25, 29592971, https://doi.org/10.1175/1520-0485(1995)025<2959:AFSWDO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Madsen, N. M., and D. G. Long, 2016: Calibration and validation of the RapidScat scatterometer using tropical rainforests. IEEE Trans. Geosci. Remote Sens., 54, 28462854, https://doi.org/10.1109/TGRS.2015.2506463.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pensieri, S., R. Bozzano, and M. E. Schiano, 2010: Comparison between QuikSCAT and buoy wind data in the Ligurian Sea. J. Mar. Syst., 81, 286296, https://doi.org/10.1016/j.jmarsys.2010.01.004.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pickett, M. H., W. Tang, L. K. Rosenfeld, and C. H. Wash, 2003: QuikSCAT satellite comparisons with nearshore buoy wind data off the U.S. West Coast. J. Atmos. Oceanic Technol., 20, 18691879, https://doi.org/10.1175/1520-0426(2003)020<1869:QSCWNB>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Portabella, M., A. Stoffelen, J. Verspeek, A. Verhoef, and J. Vogelzang, 2007: ASCAT scatterometer ocean calibration. Int. Geoscience and Remote Sensing Symp., Barcelona, Spain, IEEE, 2539–2542, https://doi.org/10.1109/IGARSS.2007.4423361.

    • Crossref
    • Export Citation
  • Priestley, C. H. B., 1959: Turbulent Transfer in the Lower Atmosphere. University of Chicago Press, 130 pp.

  • Ribal, A., and I. R. Young, 2019: 33 years of globally calibrated wave height and wind speed data based on altimeter observations. Sci. Data, 6, 77, https://doi.org/10.1038/s41597-019-0083-9.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Satheesan, K., A. Sarkar, A. Parekh, M. R. R. Kumar, and Y. Kuroda, 2007: Comparison of wind data from QuikSCAT and buoys in the Indian Ocean. Int. J. Remote Sens., 28, 23752382, https://doi.org/10.1080/01431160701236803.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Spencer, M. W., C. Wu, and D. G. Long, 2000: Improved resolution backscatter measurements with the SeaWinds pencil-beam scatterometer. IEEE Trans. Geosci. Remote Sens., 38, 89104, https://doi.org/10.1109/36.823904.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stoffelen, A., 1999: A simple method for calibration of a scatterometer over the ocean. J. Atmos. Oceanic Technol., 16, 275282, https://doi.org/10.1175/1520-0426(1999)016<0275:ASMFCO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Takbash, A., I. R. Young, and Ø. Breivik, 2019: Global wind speed and wave height extremes derived from long-duration satellite records. J. Climate, 32, 109126, https://doi.org/10.1175/JCLI-D-18-0520.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Taylor, P. K., and M. J. Yelland, 2001: Comments on “On the effect of ocean waves on the kinetic energy balance and consequences for the inertial dissipation technique.” J. Phys. Oceanogr., 31, 25322536, https://doi.org/10.1175/1520-0485(2001)031<2532:COOTEO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Thomas, B. R., E. C. Kent, and V. R. Swail, 2005: Methods to homogenize wind speeds from ships and buoys. Int. J. Climatol., 25, 979995, https://doi.org/10.1002/joc.1176.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Verspeek, J., A. Stoffelen, M. Portabella, A. Verhoef, and J. Vogelzang, 2008: ASCAT scatterometer ocean calibration. Int. Geoscience and Remote Sensing Symp., Boston, MA, IEEE, V248–V251, https://doi.org/10.1109/IGARSS.2008.4780074.

    • Crossref
    • Export Citation
  • Vinoth, J., and I. R. Young, 2011: Global estimates of extreme wind speed and wave height. J. Climate, 24, 16471665, https://doi.org/10.1175/2010JCLI3680.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wentz, F. J., 1983: A model function for ocean microwave brightness temperatures. J. Geophys. Res., 88, 18921908, https://doi.org/10.1029/JC088iC03p01892.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yang, X., X. Li, W. G. Pichel, and Z. Li, 2011: Comparison of ocean surface winds from ENVISAT ASAR, MetOp ASCAT Scatterometer, buoy measurements, and NOGAPS model. IEEE Trans. Geosci. Remote Sens., 49, 47434750, https://doi.org/10.1109/TGRS.2011.2159802.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Young, I. R., 1999: Seasonal variability of the global ocean wind and wave climate. Int. J. Climatol., 19, 931950, https://doi.org/10.1002/(SICI)1097-0088(199907)19:9<931::AID-JOC412>3.0.CO;2-O.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Young, I. R., and M. A. Donelan, 2018: On the determination of global ocean wind and wave climate from satellite observations. Remote Sens. Environ., 215, 228241, https://doi.org/10.1016/j.rse.2018.06.006.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Young, I. R., and A. Ribal, 2019: Multiplatform evaluation of global trends in wind speed and wave height. Science, 364, 548552, https://doi.org/10.1126/science.aav9527.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Young, I. R., A. V. Babanin, and S. Zieger, 2011: Response to comment on “Global trends in wind speed and wave height.” Science, 334, 905, https://doi.org/10.1126/science.1210548.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Young, I. R., E. Sanina, and A. V. Babanin, 2017: Calibration and cross validation of a global wind and wave database of altimeter, radiometer, and scatterometer measurements. J. Atmos. Oceanic Technol., 34, 12851306, https://doi.org/10.1175/JTECH-D-16-0145.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zeng, L., and R. A. Brown, 1998: Scatterometer observations at high wind speeds. J. Appl. Meteor., 37, 14121420, https://doi.org/10.1175/1520-0450(1998)037<1412:SOAHWS>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zieger, S., 2010: Long term trends in ocean wind speed and wave height. Ph.D. thesis, Swinburne University of Technology, 177 pp.

  • Zieger, S., J. Vinoth, and I. R. Young, 2009: Joint calibration of multiplatform altimeter measurements of wind speed and wave height over the past 20 years. J. Atmos. Oceanic Technol., 26, 25492564, https://doi.org/10.1175/2009JTECHA1303.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zieger, S., A. V. Babanin, W. Erick Rogers, and I. R. Young, 2015: Observation-based source terms in the third-generation wave model WAVEWATCH. Ocean Modell., 96, 225, https://doi.org/10.1016/j.ocemod.2015.07.014.

    • Crossref
    • Search Google Scholar
    • Export Citation
Save