The Accuracy of Voluntary Observing Ships' Meteorological Observations-Results of the VSOP-NA

Abstract

For the Voluntary Observing Ships Special Observing Project for the North Atlantic (VSOP-NA), the layout, meteorological instrumentation, and observing practices of 45 voluntary observing ships (VOS) operating in the North Atlantic were cataloged. Over a two-year period these ships provided extra information with each observation, and the effect of different observing practices has been quantified by using analysis fields from an atmospheric forecast model as a comparison standard. Biases of order several tenths of a degree Celsius were detected in sea surface temperature data from engine intake thermometers, in dewpoint temperatures from screens (and to a lesser extent, psychrometers), and in air temperatures due to solar heating. Wind speeds from anemometers were high compared to visual winds by about 2 kt for winds up to about 25 kt. The VSOP-NA data do not, however, indicate which is the more accurate. Correction for anemometer height and use of the WMO Commission for Marine Meteorology version of the Beaufort scale reduced this difference significantly. The result of these corrections on mean heat flux estimates was only a few watts per square meter but much greater changes resulted for particular areas and seasons. The project identified observing methods that are to be preferred for future use on the VOS, and demonstrated that the combined use ofVOS data and a forecasting model allowed the detection of biases both in the observations and in the model analyses.

Trends in ship wind speeds adjusted for observation method and height

“…The decrease in the day − night difference in U EL since the early 1980s supports this idea. The larger day − night difference in U EL during the 1960s to early 1980s, of about 0.3 ms −1 , could be explained if observers were underestimating the sea state, and hence the wind speed, at night (Kent et al, 1993). The day − night difference (order 0.1 ms −1 or less) shown by U M10 is more likely to be correct, which suggests that the much larger day − night difference in the visually estimated observations is spurious.…”

“…Kent et al (1993) showed that reports from ships' estimated winds were biased low at night, but the tendency for a low bias was reduced for those ships carrying an anemometer. We found very little day − nighttime bias in measured ship winds.…”

Methods to homogenize wind speeds from ships and buoys

“…Visual estimates of wind speed are harder to make at night (e.g. Kent et al, 1993) and we therefore also analysed night and daytime data separately (Table IV). The data showed a tendency for the daytime wind speed distribution to be slightly stronger than the night-time, although overall the differences are small.…”

“…Ship heading with respect to the prevailing winds and waves may also affect the visual estimate of wind (Thomas and Swail, 2004). Such factors, and others, such as possible variations in the Beaufort scale with time and recruiting country (Lindau, 1995b) and the influence of anemometers on ships reporting U E (Kent et al, 1993), also require further investigation.…”

“…However, significant temporal changes in the size and type of observing platform (including the introduction of moored buoys in the 1970s) and in measurement methods have introduced inhomogeneities to databases of archived marine winds. Inhomogeneities occur for many reasons, including: different observing methods (Quayle, 1980;Peterson and Hasse, 1987;Cardone et al, 1990;Wilkerson and Earle, 1990;Kent et al, 1993Kent et al, , 1998Lindau 1995a;); systematic changes in measurement height (Cardone et al, 1990); airflow distortion over the ship (Dobson, 1981;Yelland et al, 2002;Moat et al, 2005); different averaging times and methods (Dobson, 1981;Pierson, 1983;Gilhousen 1987;Taylor et al, 2002); and the effect of environmental conditions (Dobson, 1981;Kent et al, 1993). Other sources of observational error include: differences from nominal observing time; the anemometer type, calibration, and location; errors in calculation of true wind from the relative wind (Kent et al, 1993;Gulev, 1999;Smith et al, 1999); rounding artefacts; and for Beaufort winds, the stage of development of the waves.…”