ABSTRACT
The philosophy of scientific data j information management, which posits that data are a valuable resource worthy of preservation, is vindicated only to the extent to which those data are used with confidence. Data of understood quality, free from introduced errors or biases, are critical for analyses that aim to isolate subtle trends or patterns related to disturbance, succession or ecosystem evolution. As data accrue, the demand for data-driven, statistically based quality control techniques increases. A data quality control method for long-term environmental data that emphasizes statistical data visualization is described. Traditional parametric modelling is compared with semiparametric smoothing techniques using the Akaike Information Criterion (A 1C) for model selection. An interface in development fits a series of parametric and semiparametric models, selects the minimum AlCfit, returns graphical output and estimates the potential intervention points. This application is reviewed on the macrobenthos and nutrient chemistry data from North Inlet, South Carolina.
