Temporal correction of irregular observed intervals of groundwater level series using interpolation techniques.

Groundwater level observations due to the lack of synoptic water-level measurements are often available with irregular time intervals. The results lead to an inaccurate understanding of the parameters affecting groundwater level changes and their time delays. This paper presents a simple and low-cost approach to interpolate the irregular, periodic intervals of recorded groundwater level data for the Najafabad aquifer located in central Iran. Forty-six monitoring wells with an average of 20 years of data were used for this study. Piecewise cubic Hermite interpolating polynomial (PCHIP), cubic spline, linear, and nearest neighbor are the selected interpolation methods. Cross-validation with leave-one-out technique was also applied to evaluate the accuracy of the methods under three conditions, including low, average, and high groundwater level decline. The analysis proved that (1) PCHIP interpolation showed a high performance in all of the groundwater level decline conditions (0.22 < RMSE < 0.59 and 0.72 < R2 < 0.98). (2) Under slight and without abrupt fluctuations of groundwater level conditions, cubic spline showed a better performance than the linear interpolation (R2 = 0.98, RMSE = 0.49 for cubic spline and R2 = 0.98, RMSE = 0.51 for linear). The selected approach reduces the uncertainty in the study of issues such as changes in groundwater recharge and storage, groundwater flow directions, surface–groundwater interaction, and groundwater modeling. [ABSTRACT FROM AUTHOR]