Precipitable water vapour content from ESR/SKYNET sun–sky radiometers: validation against GNSS/GPS and AERONET over three different sites in Europe

摘要

The estimation of the precipitable water vapour content () with hightemporal and spatial resolution is of great interest to both meteorologicaland climatological studies. Several methodologies based on remote sensingtechniques have been recently developed in order to obtain accurate andfrequent measurements of this atmospheric parameter. Among them, the relativelow cost and easy deployment of sun–sky radiometers, or sun photometers,operating in several international networks, allowed the development ofautomatic estimations of  from these instruments with high temporalresolution. However, the great problem of this methodology is the estimationof the sun-photometric calibration parameters. The objective of this paper isto validate a new methodology based on the hypothesis that the calibrationparameters characterizing the atmospheric transmittance at 940 nm aredependent on vertical profiles of temperature, air pressure and moisturetypical of each measurement site. To obtain the calibration parameters somesimultaneously seasonal measurements of , from independent sources, takenover a large range of solar zenith angle and covering a wide range of ,are needed. In this work yearly GNSS/GPS datasets were used for obtaining atable of photometric calibration constants and the methodology was appliedand validated in three European ESR-SKYNET network sites, characterized bydifferent atmospheric and climatic conditions: Rome, Valencia and Aosta.Results were validated against the GNSS/GPS and AErosol RObotic NETwork (AERONET)  estimations. In both the validations the agreement was very high, with a percentage RMSD of about 6, 13 and 8 % in the case ofGPS intercomparison at Rome, Aosta and Valencia, respectively, and of 8 %in the case of AERONET comparison in Valencia.Analysing the results by  classes, the present methodology was found toclearly improve  estimation at low  content when compared against AERONET in terms of % bias, bringing the agreement with the GPS (considered the reference one) from a % bias of 5.76 to 0.52.