Twenty Years of Polar Winds from AVHRR: Validation and Comparison with ERA-40

Recent studies have shown that the Arctic climate has changed markedly over the past 20 years. Two major reanalysis products that can be used for studying recent changes unfortunately exhibit relatively large errors in the wind field over the Arctic where there are few radiosonde data available for assimilation. At least 10 numerical weather prediction centers worldwide have demonstrated that satellite-derived polar winds have a positive impact on global weather forecasts. The impact on reanalyses should be similar. Therefore, a polar wind dataset spanning more than 20 years was generated using Advanced Very High Resolution Radiometer (AVHRR) data. Comparisons with winds from radiosondes show biases in the AVHRR-derived winds of 0.1-0.8 m s(-1), depending on the level. In addition, AVHRR has lower root-mean-square speed errors and speed biases than the 40-yr ECMWF reanalysis product (ERA-40) when compared with rawinsondes not assimilated into the reanalysis. Therefore, it is recommended that the historical AVHRR polar winds be assimilated into future versions of the reanalysis products. The authors also explore possible kinematic reasons for the disparities between ERA-40 and AVHRR wind fields. AVHRR and ERA-40 speed and direction differences for various kinematic flow features are investigated. Results show that, on average, AVHRR winds are faster in jet streams and ridges but are slower in troughs and jet exit regions. The results from this study could lead to a better dynamical understanding of why the reanalysis product produces a less-accurate wind vector field over regions that are void of radiosonde data.