A multi-site techniques 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. At one hand, networks of ground-based instruments allowing to retrieve homogeneous Integrated Water Vapour (IWV) datasets 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+ yr, 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 yr (e.g. AIRS) or are being merged in order to create long-term time series (e.g. GOME, SCIAMACHY, and GOME-2). The present study aims at setting up a techniques 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 the applicability of either dataset for water vapour trends analysis. To this end, we selected 28 sites worldwide at which GPS observations can directly be compared with coincident satellite IWV observations, together with sun photometer and/or radiosonde measurements. We found that the mean biases of the different techniques w.r.t. the GPS estimates vary only between −0.3 to 0.5 mm of IWV, but the small bias is accompanied by large Root Mean Square (RMS) values, especially for the satellite instruments. In particular, we analysed the impact of the presence of clouds on the techniques IWV agreement. Also, the influence of specific issues for each instrument on the intercomparison is 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, etc. 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. We could only detect a clear dependency of the RMS, for all considered instruments, on latitude or mean IWV: the RMS of the IWV observations w.r.t. the GPS IWV retrievals decreases with increasing latitude and decreasing mean IWV.