A new algorithm for correcting irregular patterns associated with timestamp error in spatial correlated sensor data

Timestamp error commonly occurs in the capture of sensor-dependent data, and the impact can be profound. Yet, literature remains scarce on addressing this big-data challenge, particularly within the domain of environmental monitoring. In this paper, we propose an algorithm for precise realignment of displaced sequences of data when the occurrence of timestamp error in data captured by certain air monitoring stations within a spatial network manifests as an irregular diurnal pattern. The algorithm may be generalizable to other domains too, given the simple premise of gradually shifting and comparing the displaced measurements of one select variable from the multivariate set, to data positioned at corresponding time points in multiple imputed representations. A pooled measure of mean absolute scaled error is used to identify the required shift for data repair. The algorithm was stress-tested in a univariate simulation study against baseline methods, and was found to be superior and highly reliable in the exact correction of contrived sequences of displaced air temperature data, particularly under conditions of strong cross-correlation between series recorded at different locations. The algorithm was further applied to an empirical case of timestamp error in multivariate data, with sensitivity analysis yielding robust and consistent results.