An approach to solve the problem of operative anomalies recognition in scalar fields data for monitoring objects and processes using the gradient method and the system of specific information metrics is proposed. Metrics of module and direction standard deviation of the gradient fields difference are used. A binary classifier system method and a mathematical model and a recognition procedure that works in an interactive mode are constructed. The process of proposed approach operation is illustrated using the example of water temperature vertical spatial sounding data sample of Kruglaya Bay area of the city of Sevastopol.