Modeling Chlorophyll <formula formulatype="inline"> <tex Notation="TeX">$a$</tex> </formula> Concentration Affected by Artificial Upwelling in Qiandao Lake

Abstract

Artificial upwelling of ocean water has been increasingly used to enhance primary marine productivity. This approach brings nutrient-rich deep water to the surface and alters the physical characteristics of water and thus the distribution of the nutrient over an effective period. Various water physical characteristics may affect chlorophyll $a$ concentration, a key indicator for assessing primary marine productivity. Therefore, it is important to study the correlation between chlorophyll $a$ concentration and various water physical characteristics in the artificial upwelling, which has never been examined in literature. In this paper, several water quality parameters, which have potential correlation with chlorophyll $a$ concentration, are selected and multiple linear regression (MLR) and multivariate quadratic regression (MQR) are employed to test the correlation. Then, two novel neural network approaches, i.e., genetic-algorithm-based neural network (GA-NN) and particle-swarm-optimization-based neural network (PSO-NN) are proposed to model the correlation, which are further compared with regular backpropagation neural network (BP-NN), MLR, and MQR. The experimental data are collected from Qiandao Lake, China. The modeling results show that chlorophyll $a$ concentration has a strong correlation with salinity, temperature, dissolved oxygen (DO), and pH in the process of artificial upwelling, and GA-NN and PSO-NN can provide reliable prediction of the correlation.