Dynamic monitoring and simulation of Cyanobacteria Bloom based on wireless sensor network and GIS

Abstract: Because of the fast development of Chinese urbanization, the intense growth of mankind activities, and the global warming, the inland lake pollution problems have become more serious and complex, the inland lakes are becoming more eutrophic, suffering from both point and non-point pollutions, and exposing to both endogenous and exogenous pollutions, and the deterioration of water quality has severely hindered sustainable social and economic development of nearby cities. Eutrophication is not just the result of natural process, but the aggregated result of the interaction and mutual influences between natural process and human process. As the "eyes" of water protection and management, water quality monitoring is the premise for the forecast of cyanobacteria outbreak and the assessment of bloom intensity. As technology advances, the monitoring tools and measuring indicators become increasingly diverse. In the meantime, the measurement accuracy is also constantly improving. In the domain of water quality monitoring, the combination of position sensors and wireless sensor networks makes the observations with high density, high precision and continuity possible. In addition, such methods can simultaneously collect both water quality data and surrounding environmental data, which can be used to simulate the interaction between the internal and external factors of water pollution as well as the mechanism of pollution process. The cyanobacteria bloom of inland lake has the characteristics of suddenness, randomness, and regionalism, the paper proposed a cyanobacteria bloom dynamic monitoring and spatial-temporal process simulation method based on wireless sensor networks (WSNs) and geographic information system (GIS). Firstly, multi-parameter sensor array is designed using water quality sensors in order to acquire real-time water quality data, and the monitoring node is composed of 4-layer architecture, including monitoring layer, data storage layer, model layer, and application layer; secondly, the improved grey model and back propagation (BP) artificial neural network are combined to forecast the cyan bacteria bloom in 24 h. This model incorporates both environmental factors (precipitation, wind speed, and wind direction) and the internal water environment factors (nitrogen, phosphorus, chemical oxygen demand, permanganate, chlorophyll, dissolved oxygen, pH value, and temperature). Thirdly, with the advantage of spatial analysis, GIS is used to describe the spatial-temporal processing of cyan bacteria. Finally, the simulation experiment results showed that the correlation coefficient achieved up to 0.995. Moreover, we conducted the empirical experiment in Dianchi Lake, Yunnan, China using the proposed method, and the correlation coefficient was 0.86 and the predicting error was 9.74. The field experiment results show that the method discussed in the paper has certain universality and can provide theoretical basis and data support for protecting and controlling the lake environment.