Reconstructing missing data by comparing interpolation techniques: Applications for long?term water quality data

Missing data are typical yet must be addressed for proper inferences or expanding datasets to guide our limnological understanding and management of aquatic systems. Interpolation methods (i.e., estimating missing values using known within the dataset) can alleviate gaps common problems. We compared seven popular interpolation predicting substantial missingness in a long-term water quality dataset from Upper Mississippi River, U.S.A. The included 80,000 sampling sites collected over 30 yr that had total nitrogen (TN), phosphorus (TP), velocity. For all three interpolated variables, random forests very high prediction accuracy outperformed ordinary kriging, polynomial regressions, regression trees, inverse distance weighting. TP mean absolute error (MAE) 0.03 mg (L-TP)?1, TN MAE 0.39 (L-TN)?1, velocity 0.10 m s?1. forests' rates were mapped showed low spatiotemporal variability across riverscape, indicating model performance many habitat types large spatial scales. In current era “big data,” becomes an imperative step prior ecological analyses remains unfamiliar underutilized. Our research briefly describes importance addressing provides roadmap conduct intercomparisons other big datasets. also share adaptable analysis scripts, which allows others readily comparisons limnology applications contexts.