We present the preliminary results of the first stage of the project AO-594, which comprises the development andrncalibration of algorithms for photosynthetic pigment mapping in Spanish reservoirs. In the years 2001-2002, anrnextensive field campaign was made in 36 reservoirs and lakes in order to obtain a database of Rrs spectra (400-1000rnnm), photosynthetic pigments concentration and phytoplankton composition. The sampled water bodies cover a widernrange of environmental conditions, trophic levels and phytoplankton communities. As a first approach in algorithmrndevelopment, we have explored the relationships between ratios of MERIS bands and pigment concentrations throughrnsimple linear regression analysis. The bands have been selected based on the spectral properties of each pigment and arnpeak analysis of the Rrs spectra. For chlorophyll a, we have found a very good linear relationship (R2=0.919) using thernratio between bands 9 and 7. Similar results are found using band 8 instead of 7. In any case, the model derived for thernwhole range of concentrations (0-500 mg m-3) fails for low values (<15 mg m-3). Possible solutions include the use ofrnnon-linear models or the use of two different models depending on the ratio values. For cyanobacteria detection, thernratio between bands 9 and 6 (the later centred at 620 nm) shows a good correlation (R2=0.723) with phycocyaninrnconcentration measured fluorometrically, and better (R2=0.945) with zeaxanthin measured using HPLC. The correlationrnof other indicator pigments with MERIS band ratios is less strong, but is still possible to develop algorithms accuraternenough for bloom monitoring.rnWe also discuss the problems found with the L2 MERIS reflectance imagery that we have tried to use for modelrncalibration. We present the results of the study carried on six reservoirs in northeastern Spain. In a date coincident withrna MERIS image (June 19th, 2003) we have collected pigment concentration and reflectance data measured from a boatrnand 10 m resolution imagery for the 15 MERIS bands acquired with an airborne CASI-2 sensor. This three-level datarnhas allowed us to calculate a new set of reflectance MERIS bands from the L1B imagery and compare them with the L2rnimagery. The comparison could be useful for modifying the current atmospheric correction algorithms applied for L2rndata generation, which we have observed that frequently fail in inland waters.
展开▼
