Biases and Uncertainties in Radiosonde‐Based Precipitable Water Vapor

Precipitable water vapor (PWV) plays an important role in weather and climate phenomena. Accurate PWV measurements based on radiosonde observations are essential for assessing remote‐sensing measurements and long‐term trends. This study examines the performance of PWV algorithms, including their intrinsic biases and their sensitivity to radiosonde and data‐reporting errors. It is found that approximating gravity by standard gravity and simultaniously ignoring the water‐vapor enhancement factor in air underestimates PWV by about 0.3% at the poles, 0.5% at 45° latitude, and 0.8% at the equator. In addition, air temperature encoding in alphanumeric upper‐air reports causes an underestimation of around 0.4%. Error propagation analysis shows that, under typical mid‐latitude conditions: vertically constant relative humidity (RH), temperature and pressure biases (%), (K) and (hPa), produce a percent PWV bias ; sonde errors (%) and (K), representing the dispersion of temperature and RH measurements between identical radiosonde units, produce a percent PWV uncertainty , whereas sonde errors in pressure have a negligible effect. The uncertainty of PWV due to random errors in soundings is at least one order of magnitude smaller than the uncertainty due to sonde errors, decreasing as the vertical resolution of data increases. The detailed error‐propagation formulas developed here provide a framework for quantifying the bias and uncertainty of PWV associated with radiosonde errors. Regarding algorithmic accuracy, the study highlights the benefits of using altitude rather than pressure in PWV calculations, in line with recent radiosonde systems and meteorological data‐exchange practices.