Evaluation method for probability distribution and prediction models for chlorophyll-a based on statistical models and artificial intelligence-based algorithms

Abstract

In South Korea, sites that are not under the algae alert system yet frequently experience algal blooms are managed as designated algal observation sites. Chlorophyll-a is one of the key water quality parameters monitored at these sites. To investigate trends in chlorophyll-a, this study used water quality monitoring data from two representative algal observation sites in the Yeongsan River basin from January 2016 to December 2002. Based on the data, an exploratory data analysis was conducted to examine the distributional characteristics of each variable, after which an appropriate probability distribution was inferred to explain fluctuations in chlorophyll-a. Building on these data, three statistical models and four artificial intelligence-based algorithms for predicting chlorophyll-a were developed, and their levels of predictive performance were quantitatively compared. The more precise methods proposed in this study for predicting chlorophyll-a levels are expected to significantly aid in water quality management at various monitoring sites.