A physically based, simultaneous air/water constituent algorithm is used for automatic processing of 5 MERIS level 1B FR datasets of Lake Constance. Its sensitivity to the adjacency correction tool ICOL is tested, and compared to variations in the wavelength dependence of aerosols. Further comparisons are made with MERIS NN algorithms for eutrophic lakes and case II waters. Reference data measured during the MERIS lakes field campaign in April 2007 is used, consisting of both direct concentration and optical in situ measurements. AERONET data of the two closest stations are evaluated. A good representation of a pattern of extending suspended matter is found for adjacency corrected MIP outputs and MERIS NN products, while chlorophyll and gelbstoff are insufficient.