The physically based Modular Inversion & Processing System (MIP) is used in a processing chain for inland water constituent retrieval from MERIS level 1B full resolution data. Reformatting, data import, water/land masking, atmospheric correction and water constituent retrieval are fully automized, in order to allow the efficient analysis of large data quantities in a time series of water quality in Swiss Lakes. To account for the temporal variation of atmosphere and water properties, a set of input parameters is optimized specifically for Lake Constance. The algorithm’s sensitivity to its input parameters is studied, allowing the derivation of a first guess parameterization. Results of successfully processed data are presented and reasons for processing errors are discussed.