JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 101, NO. C8, PAGES 18,377-18,389, AUGUST 15, 1996 Comparison between weather buoy and Comprehensive Ocean-Atmosphere Data Set wind data for the west coast of Canada JosefY. Cherniawskyand William R. Crawford Department of Fisheriesand Oceans,Canadian Hydrographic Service,Institute of Ocean Sciences Sidney, British Columbia, Canada Abstract. We describesurfacewind data from Canadian weather buoys in the northeast Pacific and compare monthly mean wind speedand direction during 1987 to 1992 from these buoys with monthly mean values in the Comprehensive Ocean-Atmosphere Data Set (COADS). The latter is based mainly on data from volunteer observing ships (VOS), binned in 2 ø x 2 ø squares. The number of VOS observations in each squarein Hecate Strait, Dixon Entrance, and Queen Charlotte Soundis too low to producereliable averages.Near other buoyswe find COADS windspeeds to be higher than buoy windspeeds by about1 to 2 ms -•. This difference shows somedependence on wind speedand is in qualitative agreement with previous comparisons between VOS and weather buoy winds. In addition, we find direction-dependent differences in wind direction, with COADS wind vectors to the right of buoy wind vectors for alongshore windsfrom the northwest near the Vancouver Island coast. These resultssuggest that high placement of anemometers (above 25-30 m) on largemodern vessels is partly responsible for higher COADS wind speeds, while rotation of wind vectors alongmountainranges and inadequate VOS coverage in coastal watersis biasing wind direction in COADS data toward the open oceanwinds. 1. Introduction Marine weatherbuoysare currentlymaintained along the Pacific Coastof Canada(Figure1) by the Atmo- spheric Environment Service (AES, Environment Cana- da), Canadian Coast Guard, Institute of Ocean Sciences (lOS, Department of Fisheries andOceans), andAXYS EnvironmentalConsultingof Sidney,British Columbia [Woodand Wells, 1988; Environment Canada,1994]. These buoyswere put in place to improve marine weath- er forecastsand search and rescue capabilities. Buoy data provide a unique history of marine weather on sy- noptic and longertimescales that are not available from any other source. They are also useful for calibration of remotely sensed data from satellites [Cower, 1996]. The number of buoys has gradually increased from an initial deployment of two in September 1987 to 16 in the fall of 1993. Observations of surface wind (from two R.M. Younganemometers at each buoy), air and sea surface temperatures, sea levelpressure (two pres- sure sensors) andotherdata aretransmitted every hour via the GOES environmentalsatellite and archivedby Environment Canada, Pacific Region. Copyright 1996 by the American GeophysicalUnion. Paper number 96JC01687. 0148-0227/96/96JC-01687509.00 We describe herebuoywind data and compare month- ly meanscalar wind speed and direction from 11 buoys with wind data from nearbyvolunteer observing ships (VOS), as provided in the Comprehensive Ocean-Atmo- sphere Data Set(GOADS, Release la, standard Monthly Summary TrimmedGroup4). GOADS wasdescribed by Woodruff et al. [19871 andSlutz et al. [19851. It over- laps in time with the Canadian buoy data from 1987 until 1992. Preliminary resultsfrom this comparison were presented by Cherniawsky et al. [1995]. We are interestedin this comparison for two reasons. First, we need to know about quality and relative merits of thesetwo data sets for providing surface wind forc- ing of coastal oceancirculationmodels. We alsoneedto learn more about buoy-VOS wind data compatibility, asbuoy data are nowroutinely included in the enhanced andinterim versions of GOADS (Release la Documen- tation; ftp from ncardata.ucar.edu:/pub/COADS). In particular, we are concerned with spatial artifacts in future versions of GOADS, for those GOADS grids wherebuoy data coverage overwhelms that from VOS. Previous evaluations of buoy winds and comparisons. between buoy andshipdata were done for different geo- graphical areas and based on individual (not monthly mean) ship andbuoy reports [Quayle, 1984; Gilhousen, 1987;Wilkerson andEarle, 1990; Pierson, 1990].These authors reported buoywind speeds that are systemati- cally lower byabout 1 to 2 ms- • than those reported by VOS vessels. Similar differences alsoappearin our corn- 18,377