Abstract Water vapour is an important greenhouse gas that plays a key role in the thermal‐infrared (long‐wave) radiation balance at the surface. In the Arctic, previous studies have identified uncertainties regarding water vapour trends and integrated water vapour (IWV) estimates in certain seasons and regions, mostly due to missing high‐quality observations from ground stations and uncertainties in satellite remote sensing. Therefore, accurate field measurements of water vapour are crucial for providing reference observations in the data‐sparse central Arctic to evaluate satellite products, reanalyses, and weather forecast models. In this study, we use observations from ground‐based microwave radiometers and radiosondes from the year‐long Multidisciplinary Drifting Observatory for the Study of Arctic Climate expedition to analyse the quality of two global reanalyses, two weather forecast models, and two satellite products. Specifically, we evaluate the IWV and specific‐humidity profiles. Our analyses revealed dry biases regarding specific‐humidity profiles and IWV in dry (moist) conditions for the reanalyses and one of the weather forecast models (satellite products). We found a strong correlation between specific‐humidity profile deviations to the reference (radiosondes) and the representation of specific humidity inversions. We therefore also statistically analysed the representation of humidity inversions in each dataset and determined which inversions can be detected. The presence of surface‐based inversions (found in of the radiosonde profiles) is generally well captured by all datasets except the satellite products, whereas elevated and weak inversions are often missed. To assess the importance of humidity inversions, we quantified their impact on downward long‐wave radiation (DLR) using radiative transfer simulations of original and modified humidity profiles in clear‐sky conditions. The presence of humidity inversions increases the DLR in clear‐sky conditions by up to 16 Wm, whereas misrepresentations thereof can cause DLR differences of 5 Wm.
Walbröl et al. (Fri,) studied this question.