|An algorithm for the attenuation of the photosynthetically available radiation (KPAR): application to MODIS and MERIS imagery and validation with Smart Buoys plateforms|
Nechad, B.; Ruddick, K. (2012). An algorithm for the attenuation of the photosynthetically available radiation (KPAR): application to MODIS and MERIS imagery and validation with Smart Buoys plateforms, in: 44th international Liège colloquium on ocean dynamics "Remote sensing of colour, temperature and salinity – new challenges and opportunities" - May 7-11, 2012. pp. 2
In: (2012). 44th international Liège colloquium on ocean dynamics "Remote sensing of colour, temperature and salinity – new challenges and opportunities" - May 7-11, 2012. GHER, Université de Liège: Liège. 126 pp., more
KPAR, the attenuation of the photosynthetically light radiation (PAR) with depth and the euphotic depth, ZEU, i.e. the depth at which PAR reduces to 1% of its surface value, are amongst the key parameters in ecosystem modeling for estimation of primary production in the water column. A bio-optical model of KPAR has been developed for open and coastal waters that estimates KPAR at ZEU, KPAR1%, and near the surface, KPAR95%, from the inherent optical properties (IOPs): the total absorption, a, and backscattering, bb, coefficients at 490 nm measured at water surface. KPAR is also parameterised as a function of the sun position. The model was built based on 500 synthetic data set of inherent optical properties and the associated light attenuation data generated with Hydrolight software (IOCCG, 2006). In the present study, this KPAR model is applied to MODIS and MERIS imagery and validated using in situ PAR collected by Cefas Smart Buoys and also by comparing with estimations of KPAR1% at ZEU from the in situ profiles of PAR.The absorption and backscattering coefficients are obtained from MODIS and MERIS products via 2 different schemes: the NIR atmospheric correction algorithm (Bailey et al., 2010; Stumpf et al., 2002) is applied to the top of atmosphere (TOA) reflectance of MODIS data yielding the water-leaving reflectance. In MERIS the standard Neural Network algorithm is used to remove the atmospheric contributions from the TOA reflectance. The Quasi-Analytical Algorithm (QAA) of (Lee et al., 2005) inverts the marine signal to obtain a and bb, as is currently implemented in SeaDAS. Further, the QAA is applied to MERIS water-leaving reflectances adapted to the 4 MERIS channels 443, 490, 560 and 650 nm. Time series of KPAR1% and KPAR95% maps are derived from MERIS and MODIS a(490nm) and bb(490nm) products over the Southern North Sea, covering the period 2003-2009. Concurrent Cefas measurements of PAR at 0, 1 and 2 m depth at 2 stations located in the North Sea, namely Warp Anchorage (very turbid) and Oyster Grounds (clearer waters), are used to estimate KPAR0.5 and KPAR1.5 respectively at 0.5 and 1.5 m. The satellite KPAR95% are compared to KPAR0.5 (respectively KPAR1.5) when KPAR95% equals 0.103 m-1 (respectively 0.034 m-1). Next, a Look Up Table generated from Hydrolight simulations assuming a mixed water column with constant IOPs along the depth, is used to retrieve the ranges of absorption and backscattering coefficients that correspond to the two in situ KPAR values at 0.5 and 1.5 m, and the given sun zenith angle. This LUT is also used to retrieve the associated KPAR1%. The LUT-retrieved KPAR1% deviations from the satellite derived KPAR1% are explained in terms of a) the propagation of uncertainties from the input absorption and backscattering as expressed by the LUT-retrieved a and bb deviations from the satellite retrieved a and bb, b) the errors in in situ KPAR measurements and c) the impact of errors in the KPAR model.Despite the fact that different algorithms and data sources were injected in the KPAR model, a generally good agreement is found between the satellite derived KPAR1% and the corresponding in situ measurements and estimations of KPAR1%.