Globally averaged wind speeds from archived ship reports show an increasing trend since the 1940s, which previous studies attributed to the gradual shift from visually estimated to anemometer-measured winds, and an increase in mean anemometer height. To test this hypothesis, adjustments to account for these changes were applied to individual ship wind observations in the International Comprehensive Ocean–Atmosphere Data Set (ICOADS). Annual means calculated from adjusted reports were compared to reanalysis near-surface winds in consistently well-sampled regions for the period 1958–2002.
The adjustments to individual ship wind reports for known biases improve the homogeneity over the entire period and remove trends, prior to the early 1980s. However, for the 21-year period 1982–2002, trends remain in the annual mean spatially averaged adjusted winds, of 0.4 ms−1 decade−1 (6% decade−1) for estimated speeds, and of 0.2 ms−1 decade−1 (3% decade−1) for measured speeds. In contrast, co-located reanalysis near-surface wind speeds show little change, 1% decade−1 or less. Anemometers were installed on increasing numbers of ships during the period of study, even as many of these ships continued to report estimated winds. Part of the trend in estimated wind speeds is probably spurious, most likely a result of observers being increasingly influenced by readings from anemometers located well above 10-m. This influence is indicated by the gradual reduction in the tendency to underestimate nighttime marine winds. Results confirm earlier findings that reanalysis near-surface winds were biased low. The paper summarizes methods that could be employed to account for remaining inhomogeneities and thereby improve the quality of the marine wind climate record. Copyright © 2007 Crown in the right of Canada. Published by John Wiley & Sons, Ltd