AUTHOR=Tang Yandong , Deng Jiahao , Zang Cuiping , Wu Qihong 

TITLE=Chaotic Modeling of Stream Nitrate Concentration and Transportation via IFPA-ESN and Turning Point Analyses

JOURNAL=Frontiers in Environmental Science

VOLUME=Volume 10 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/environmental-science/articles/10.3389/fenvs.2022.855694

DOI=10.3389/fenvs.2022.855694

ISSN=2296-665X

ABSTRACT=Escalated concentrations of nitrogenous compounds in stream networks is proved to be detrimental for the health of both human and ecosystems. Monitoring, modeling, and forecasting nitrate concentration in the temporal domain is essential for in-depth understanding of the nitrate dynamics and transformations within the stream networks. To investigate the temporal dynamics of stream nitrate concentration, an advanced chaotic modeling and forecasting approach integrated with turning point analysis has been proposed in this research. First, the time-series daily nitrate concentrations in the form of nitrate-nitrite (NOx-N) are reconstructed based on the chaotic characteristics and then input into the forecasting models. Second, an echo state network (ESN) is developed for one-day-ahead nitrate concentration forecasting and the hyperparameters are optimized through an improved flower pollination algorithm (IFPA) to achieve high efficiency. Furthermore, turning point analysis is performed to quantify the relationship between the discharge and peak nitrate concentration. The Ricker Function is constructed and the parameters are estimated for turning points using the forecasted daily nitrate concentration and measured daily discharge. The field data including daily stream nitrate concentration and discharge collected from 8 different monitoring sites in southern Sichuan Basin, China has been utilized for case studies. Comparative analysis is performed under three modeling scenarios including conventional time-series modeling, temporal signal decomposition, and data reconstruction & embedding with chaotic characteristics. Four benchmark time-series forecasting algorithms are also compared against the proposed IFPA-ESN in the above scenarios. For each site, the parameters of the Ricker functions are estimated and turning points are computed based on the forecasted nitrate concentration and discharge volume. Computational results validate the superiority of the proposed approach on improving stream nitrate concentration prediction accuracy. Limitations of supply and transportation of nitrogenous compounds are quantified which is valuable for future pollution mitigation.