The evaluation of the integrated water vapour annual cycle over the Iberian Peninsula from EOS‐MODIS against different ground‐based techniques. (8th January 2013)

• Record Type: Journal Article
• Authors: Bennouna, Y. S.; Torres, B.; Cachorro, V. E.; Ortiz de Galisteo, J. P.; Toledano, C.
• Is Part Of: Quarterly journal of the Royal Meteorological Society. Volume 139:Number 676(2013:Oct.)
• Page Start: 1935
• Page End: 1956
• Publication Date: 2013-01-08
• Subjects: remote sensing -- water vapour -- annual cycle -- MODIS -- GPS; Meteorology -- Periodicals; 551.5
• DOI: 10.1002/qj.2080
• Languages: English
• ISSNs: 0035-9009

Abstract

Abstract: The annual cycle of precipitable water vapour is inferred from the MODIS thermal infrared (IR) and near‐infrared (NIR) satellite products under clear‐sky conditions for the period 2002–2008 at 18 sites in the Iberian Peninsula, with the aim to evaluate the capabilities of both algorithms. The article presents these results in relation to ground observations using different techniques (GPS, sunphotometer, radiosounding), with a special emphasis on GPS. At all sites the annual cycle presents the typical shape with low values in winter (minimum ∼1 cm) and high values in summer (maximum ∼3 cm), and a smaller cycle amplitude at continental sites than at coastal sites. The satellite results clearly depict the north–south gradient, as well as singular patterns such as the July minimum characterizing the most southern stations. The differences in the monthly means with respect to ground observations are mostly underestimations, ranging between a few percent to 40%, being generally larger in winter than in summer. Overall, NIR performs better than IR, especially in the wintertime. However, NIR presents overestimations in summer that can reach up to 30%. The comparison results of the annual cycle appear much more heterogeneous within coastal than continental areas, in particular for the Mediterranean region. It was found that the performance of IR algorithm is strongly influenced by the seasonal variations, a dependence which is much weaker in NIR retrievals. The time‐coincident comparison shows that best MODIS/GPS agreement is found for continental regions (r.m.s.∼0.3) for both methods. The largest regression biases and rms differences are found at the Mediterranean sites. The influence of sampling in the evaluation of the MODIS annual cycle was shown to affect mainly winter values. Errors induced by the retrieval accuracy and those induced by sampling deficiencies sometimes add up or compensate each other partially or totally.