Savoirs en partage

Le portail des ressources scientifiques
et pédagogiques de l'AUF

Comparison of several databases of downward solar radiation data at ocean surface with PIRATA measurements

Sujet: [SDU.OCEAN] Sciences of the Universe [physics]/Ocean, Atmosphere, [SDU.STU.CL] Sciences of the Universe [physics]/Earth Sciences/Climatology
Auteur: Trolliet, Mélodie, Lefèvre, Mireille, Boilley, Alexandre, Wald, Lucien
Résumé: Measurements of the downwelling shortwave irradiance at the surface (SSI) at the Atlantic ocean surface are verylimited in space and time. Re-analyses offers high quality and high resolution data of SSI. They can be usedin addition to in-situ measurements in order to increase the knowledge of the relationship between the SSI andatmospheric circulation. However, the accuracy of such re-analyses over the Atlantic ocean remains to assess. Toanswer this question, measurements of the SSI were collected from the Prediction and Research Moored Array inthe Tropical Atlantic (PIRATA) network of buoys located in the equatorial Atlantic ocean. The SSI from severalre-analyses were compared to these measurements. Quality-control was performed onto the measurements fromthe 17 buoys using recognized procedures. SSI from the ERA-Interim and MERRA-2 re-analyses were collectedfor the coincident period: 2011-2016 for the locations of the buoys. In addition, SSI were collected from threesatellite-derived databases: CAMS Radiation, HelioClim-3v4 and v5.Work is underway and exact results are unknown at the moment of submission. From preliminary results, thefollowing is expected. The re-analyses exhibit a tendency to underestimate the SSI in the equatorial Atlantic oceanwith a noticeable influence of the sky conditions on the bias. The re-analyses tend to predict cloud-free conditionswhile actual conditions are cloudy. The correlation coefficients are weak, showing that a large part of the temporalvariability is not captured.In contrast, the three satellite-derived databases offer a fair agreement with PIRATA measurements. Though thebias may be large at times, the standard deviations of the errors are small, meaning a limited scattering of errors. Thecorrelation coefficients are great, meaning that the satellite-derived databases capture a great part of the variabilityin time.It is concluded that efforts must be made on the re-analyses for a better modelling of the clouds. Meanwhile,satellite-derived data sets offer less uncertainties and should be preferred.
Editeur: HAL CCSD