In marine research, as in many other areas, models are a fundamental concept. Models are essential for predictions and control purposes. They are obtained, improved and verified by measurements of the modelled entity.These measurement results are usually obtained within one or more experiments. As an experiment we identify any process which is at least partially controlled by the researcher and results in a collection of data. Controllable conditions are usually where and when measurements are performed. Sometimes initial conditions and the temporal control of the experiment can also be influenced. Examples for that are the quantity of substances and initial temperature at chemical experiments and there temporal changes. All the controllable conditions of an experiment are summarized as the corresponding experimental design. The experimental design crucially affects the information content of the measurement results. Depending on the purpose of the measurements the experimental design can be optimized to gain the maximum of information. As a result fewer measurements and experiments are conducted to obtain sufficient information. Thus time, money and effort can be saved. There are mainly two types of purposes in respect to which the experimental design is optimized. One major purpose of measurement results is to determine unknown parameters in models. The experimental design for these measurements can be optimized to determine the parameters as good as possible. Another purpose is to choose the most realistic model of a set of potential models. This process is called model discrimination. For that the experimental design can be optimized in such a way that the most realistic model can be determined as clearly as possible.
