A general zero-inflated left-censored testing approach with a single detection limit for environmental exposure

The exponential growth of modern industry has led to environmental pollution globally. The inherent limitations of measurement instruments have resulted in the prevalent occurrence of left-censored data in environmental pollution monitoring. To effectively evaluate and control the risk of contamination, it is crucial to determine the probability distribution of pollutant concentrations. In this study, we developed a general zero-inflated left-censored mixture probability distribution model with a single detection limit and derived an explicit solution for the zero-exposure proportion. Furthermore, we applied an EM algorithm to estimate the parameters of the probability distribution based on the explicit solution, and it showed good performance. The proposed method was used to characterize 22 pesticide residues from 241 water plants in China between 2008 and 2012. The results revealed a widespread occurrence of zero exposure in pesticide residues, particularly in the case of prohibited pesticides. Given the frequent existence of zero exposure, this method can improve the precision of environmental pollution risk quantification and serve as a valuable tool for parameter estimation of zero-inflated, left-censored data in environmental management.