科研之友
微信
新浪微博
Facebook
分享链接
摘要
The main aim of this work is to evaluate the NASA EOS AURA Ozone Monitoring Instrument (OMI) UV irradiance estimates through ground -based measurements performed by a NILU-UV multichannel radiometer (NILU-UV) operating in Thessaloniki, Greece, for the time period between January 2005 and December 2014. NILU-UV multi -filter radiometers can provide measurements at 5 UV wavelength bands with full width at half maximum (FWHM) of 10 nm approximately and a time analysis of 1 min. An additional channel measuring the Photosynthetically Active Radiation (PAR)" is also incorporated to the instrument and is used for the stringent characterization of the cloud free instances. The OMI instrument estimates solar UV irradiances at four wavelengths close to those of the NILU-UV in Thessaloniki. Clear and all-slcy overpass -time, as well as solar local -noon time, UV estimates are provided by the NASA Aura Data Validation Center. Spectra measured from a collocated MKIII Brewer spectrophotometer with serial number 086 (Brewer #086) were utilized for the whole "period (2005-2014) in order to estimate the NILU-UV irradiances at the OMI wavelength irradiances and therefore provide a direct comparison and validation to the NILU UV measurements provided by OMI. For the nominal comparisons, using un-flagged OMI data within a 50 km radius from Thessaloniki, the linear determination coefficient, R2, ranges between 0.91 and 0.97 for the 305 nm and between 0.75 and 0.92 for 380 nm depending on the choice of overpass or local -noon time data and the cloudiness flags. The best agreement is found for the clear-slcy overpass -time comparisons as well as the both PAR- and satellite algorithm -deduced clear -sky overpass and local -noon comparisons for all wavelengths. The OMI irradiances were found to overestimate the NILU-UV observations in Thessaloniki between similar to 4.5% and 13.5% for the 305 nm and between similar to 1.5% and similar to 10.0% for the 310 nm wavelength depending on the choice of time [overpass vs local noon] and cloudiness restrictions [only satellite -deduced or both PAR- and satellite -deduced]. For the 324 nm and 380 nm wavelengths, the satellite -deduced local -noon time clear skies comparisons show a satellite under -estimation of -3.75% and -4.15% respectively whereas the overpass-time comparisons range between -1.55% and -1.90% for the same wavelengths. When imposing the PAR -deduced clear -sky assessment as well, the comparisons show a satellite over-estimation of 2.50% and 2.75% for the local-noon and 3.70% and 4.10% for the overpass-time cases for the 324 nm and 380 nm wavelengths. The effect of a stricter radius of collocation selection (10 km and 25 km vs 50 km) appears to have a negligible impact on the comparisons however results in a smaller statistical sample. When flagging restrictions are applied to the quality indicators of the OMI data, the amount of co-locations decreases appreciatively but the correlations improve for most cases (R-2 > 0.9 in most cases and wavelengths). The effect of the temporal averaging window of the ground-based measurements, of around 10 and 60 min to the overpass time, was also investigated. The differences of the time averaging span result in a more obvious improvement of the agreement for the all-sky cases; the R-2 factor improves to 0.94, 0.90, 0.87 and 0.85 from the original 0.91, 0.87, 0.82 and 0.78 for the local-noon comparisons and to 0.96, 0.94, 0.92 and 0.91 from the original 0.93, 0.90, 0.80 and 0.83 for the overpass-time comparisons for the 305 nm, 310 nm, 324 nm and 380 nm wavelengths respectively. Hence, it is shown that this type of temporal averaging can implicitly compensate for the changes of cloud position and optical properties when comparing UV measurements from ground and space. For the detection of the seasonal characteristics of the satellite irradiances it was shown that the 380 nm wavelength can be safely studied. The standard deviation of the strict cloud free overpass-time comparisons of 12% may form a dependable measure of the seasonality imposed by the satellite retrieval algorithm assumptions. Additionally, the aerosol load at 340 nm and the possible solar zenith angle (SZA) effect on the comparisons was considered. No marked aerosol dependence was found with the ratio of satellite-to-ground irradiances remaining 1.1 +/- 1.3 for the all-sky comparisons and 1 +/- 0.1 for the clear-sky comparisons, while the SZA effect becomes appreciable only for angles higher than 70 degrees with ratios 1.1 +/- 1.0 for all-skies and 1.0 +/- 0.1 for the clear-sky cases below that threshold.
- 出版日期2016-9
