Publications

Ri, TC; Jo, JS (2023). A Genetic Algorithm-Optimized Neural Network for Chlorophyll a Estimation Using MODIS Satellite Data in Coastal Water: Application to the Sinpho Bay of DPR Korea. JOURNAL OF THE INDIAN SOCIETY OF REMOTE SENSING, 51(7), 1541-1551.

Abstract
In the present study, the hybrid model based on artificial neural network (ANN) and genetic algorithm (GA) is proposed to investigate the seasonal and spatial variability of chlorophyll a (Chla) concentration using MODIS data in the coastal water of the Sinpho Bay, DPR Korea. The GA has been used to optimize the ANN parameters such as the initial weights and biases. The four spectral bands (488, 547, 667 and 678 nm) of MODIS data were used for estimating Chla concentration in this study. The study results have shown that the GA-optimized ANN (GA-ANN) model has higher accuracy (RMSE of 2.98 mg m(-3), R-2 of 0.85, MB of 2.67 mg m(-3) and MRD of 14.24%) in retrieving Chla concentrations compared to the OCI algorithm and the ANN without GA optimization. In the Sinpho Bay, the highest Chla concentrations (> 30 mg m(-3)) occurred near the Namdae and Tonggol River estuaries and along the coast. Furthermore, the Chla concentrations were highest (> 27 mg m(-3)) during the spring season (May and early June) and were lowest (< 5 mg m(-3)) during the winter season (January and February). The present results have shown that the proposed GA-ANN model can significantly improve the remote estimation of the Chla concentration in optically complex case 2 waters, which will be of great value to natural resource managers and scientists involved in managing the inland and coastal aquatic environments.

DOI:
10.1007/s12524-023-01719-8

ISSN:
0974-3006