Abstract
Long-term changes in upper-air wind speed remain substantially uncertain due to the spatial and temporal sparseness of radiosonde observations. This study estimates wind speed trends at six tropospheric pressure levels (850–200 hPa) over land using global IGRA radiosonde records and six reanalysis datasets. Radiosonde observations indicate a general increase in upper troposphere over the past four decades, except Australia. In contrast, trends in the lower troposphere appear weaker and more spatially heterogeneous. While reanalysis products generally capture the mean state and interannual variability of wind speed in align with radiosondes, they exhibit considerable discrepancy in long-term trends—both in magnitude and direction—particularly at upper levels. Among the datasets, CRA-40 and ERA5 demonstrate relatively closer agreement with radiosonde trends, though notable regional biases remain. These results highlight the persistent inconsistencies between reanalysis products and radiosonde observations, emphasizing the need to improve upper-air observational coverage, data assimilation strategies, and uncertainty quantification. This study provides a systematic comparison framework to enable more robust detection and attribution of wind speed changes under climate change.