Accuracy of current Arctic springtime water vapour estimates, assessed by Raman lidar

Abstract

Current estimates of Arctic water vapour, obtained using reanalysis, numerical weather predictions and satellite products may be biased due to the concentration of moisture in the lowest levels, below 2–3 km. Here we investigate the water vapour content of the atmosphere during 2 weeks in May 2016, as estimated with (a) the ECMWF/Integrated Forecast System reanalyses; (b) Weather Research and Forecasting (WRF) simulations at high resolution; and (c) satellite retrievals from AIRS/Aqua and IASI/MetOp. We compare such estimates to a unique dataset of uninterrupted Raman lidar observations obtained at Hammerfest, Norway (70°N, 24°E), within the framework of the Pollution in the ARCtic System (PARCS) project. Ground‐based lidar profiles were acquired between 150 and 4,000 m altitude, with a vertical and temporal resolution of 15–200 m and 30 min, respectively. A strong variability of the water vapour field was observed in the lower troposphere in this period, ranging 0.43–6.35 g/kg. Dry air intrusions from the polar vortex, alternating with wet air intrusions associated with lows over the ocean have been revealed, creating strong gradients of water vapour, of the order of 2 g/kg per hour and per km. The satellite retrievals showed a good correlation with the measurements, despite a relatively large dispersion and a slight dry bias of the IASI level 2 product (standard deviation up to 1.5 g/kg). The ECMWF/IFS reanalysis and WRF both showed a wet bias below 1 km altitude (+0.4 g/kg), coherent with previous comparisons that made use of radiosondes.