2026/05/27 by K. L. Priya, Megha R. Raj, Hisana Nahas +5 · 1 voice
Environmental Science · Earth and Planetary Sciences · #Hydrological Forecasting Using AI #Water Quality Monitoring and Analysis #Marine and coastal ecosystems
paper · pdf · doi:10.3897/emt.3.183902
openalex publication_date 2026/05/27 · openalex created_date 2026/05/28 · openalex updated_date 2026/07/02
Eutrophication in aquatic bodies is a growing concern to the ecosystem as well as public health. Chlorophyll-a is an indicator of nutrient levels in the water body and can be considered for identifying the trophic level. Continuous monitoring of chlorophyll-a is thus a necessity and developing predictive models for chlorophyll-a can reduce the cost of analysis, time and resources. The present study firstly investigated the major controlling parameters of chlorophyll-a in the Ashtamudi wetland, India using a statistical approach. Secondly, a predictive model using Artificial Neural Network has been developed for chlorophyll-a. The study revealed that chlorophyll-a peaks during the pre-monsoon season, followed by post-monsoon and monsoon seasons. The major driving parameters of chlorophyll-a (Chl-a) were identified to be phosphates (P), nitrates (N), sulphates (S), dissolved oxygen (DO) and salinity (Sal). Further, ANN models were developed for predicting chlorophyll-a, wherein five scenarios were considered: (i) Chl-a = f(P,N,S,DO,Sal); (ii) Chl-a = f(P,N,S,Sal); (iii) Chl-a = f(P,N,Sal); (iv) Chl-a = f(P,N); (v) Chl-a = f(P). The first four scenarios yielded a similar predictability with a coefficient of determination of 0.99 with a data set collected across 41 sampling stations over 3 years, thereby revealing that phosphates and nitrates can be used for predicting chlorophyll-a and has a higher control than salinity, DO and sulphates. Further, in order to predict the chlorophyll-a on-site, the in-situ parameters namely temperature, dissolved oxygen, pH and salinity were considered for developing the model. The results revealed that salinity and pH can predict chlorophyll-a with an accuracy of 88.7% and can be used for random assessment of chlorophyll-a at site, thereby enabling the sample collection process to be streamlined.