A multi-site intercomparison of integrated water vapour observations for climate change analysis

Abstract. Water vapour plays a dominant role in the climate change debate. However, observing water vapour over a climatological time period in a consistent and homogeneous manner is challenging. On one hand, networks of ground-based instruments able to retrieve homogeneous integrated water vapour (IWV) data sets are being set up. Typical examples are Global Navigation Satellite System (GNSS) observation networks such as the International GNSS Service (IGS), with continuous GPS (Global Positioning System) observations spanning over the last 15+ years, and the AErosol RObotic NETwork (AERONET), providing long-term observations performed with standardized and well-calibrated sun photometers. On the other hand, satellite-based measurements of IWV already have a time span of over 10 years (e.g. AIRS) or are being merged to create long-term time series (e.g. GOME, SCIAMACHY, and GOME-2). This study performs an intercomparison of IWV measurements from satellite devices (in the visible, GOME/SCIAMACHY/GOME-2, and in the thermal infrared, AIRS), in situ measurements (radiosondes) and ground-based instruments (GPS, sun photometer), to assess their use in water vapour trends analysis. To this end, we selected 28 sites world-wide for which GPS observations can directly be compared with coincident satellite IWV observations, together with sun photometer and/or radiosonde measurements. The mean biases of the different techniques compared to the GPS estimates vary only between −0.3 to 0.5 mm of IWV. Nevertheless these small biases are accompanied by large standard deviations (SD), especially for the satellite instruments. In particular, we analysed the impact of clouds on the IWV agreement. The influence of specific issues for each instrument on the intercomparison is also investigated (e.g. the distance between the satellite ground pixel centre and the co-located ground-based station, the satellite scan angle, daytime/nighttime differences). Furthermore, we checked if the properties of the IWV scatter plots between these different instruments are dependent on the geography and/or altitude of the station. For all considered instruments, the only dependency clearly detected is with latitude: the SD of the IWV observations with respect to the GPS IWV retrievals decreases with increasing latitude and decreasing mean IWV.