Rainfall And Water Quality Monitoring In Inland Lakes

In the context of water shortage and serious water pollution problems in China,timely,effective,scientific and reasonable water quality monitoring is of great significance for water pollution control,water environmental protection and public safety.In this paper,aiming at the problem that surface water quality monitoring data is spatially discrete and insufficient in spatial coverage,the water quality classification based on remote sensing images is studied to realize the water quality monitoring covering the entire water space.To deal with the problem that remote sensing data is insufficient in instantaneity,while the surface water quality monitoring cost is creeping up with the increasing number of sampling times,the real-time rainfall prediction based on wireless sensor network(WSN)meteorological monitoring data is studied to provide accurate rainfall prediction information.Then the relationship between rainfall and water quality is studied to analyze the impact of rainfall on variation trends of water quality.Combining the relationship with accurate rainfall prediction information,sampling frequency and monitoring parameters of surface water quality monitoring stations can be adjusted,which not only improves the effectiveness of monitoring data,but also controls the monitoring cost growth.Reasonable and efficient water quality monitoring can be achieved.The main research contents and contributions of this paper are as follows:In view of the limitations of existing surface water quality monitoring,the water quality classification based on remote sensing images,which is easy to obtain,economical and spatially continuous,is studied.Compared with common retrieval of water quality parameters,we establish the classification model that uses remote sensing images as the input and water quality grade as the output.Compared with the custom water quality division rules,the water quality division of this paper follows the national surface water environmental quality standard(GB3838-2002).Taking the Erhai Lake and Chaohu Lake as research areas,Support Vector Machine(SVM),Random Forest(RF)and Convolutional Neural Network(CNN)are conducted as classification models.Experiment results demonstrate that CNN can effectively extract the water quality information based on remote sensing images.Aiming at the inherent random error of Micro Electro Mechanical System(MEMS)sensors used by WSN,which leads to deviation between measured value and actual value,a set of data validity checking process is designed.The STL(Seasonal and Trend decomposition using Loess)algorithm is used to extract the trend items of monitoring data and reference data.Then analyze the similarity between the two groups of trend data.Experiment results prove that the monitoring data of MEMS sensor is effective.On this basis,real-time rainfall prediction based on meteorological monitoring data is studied.Wuhan City and Dali City are used as research areas.Auto-Regressive and Moving Average(ARMA),SVM,RF,Back Propagation Neural Network(BPNN)and Long Short-Term Memory(LSTM)are conducted as regression prediction models.With Mean Absolute Error(MAE)and Root Mean Square Error(RMSE)as the evaluation indicators,experiment results show that LSTM can not only accurately predict real-time rainfall,but also effectively extract seasonal changes in rainfall.Finally,the variation correlation between rainfall and multiple water quality parameters is analyzed.The experiment is carried out with monitoring data in Erhai Lake.Experiment results show that rainfall is significantly correlated with water quality parameter concentration.The runoff generated by rainfall will carry industrial wastewater,domestic sewage,and farmland pollutants such as fertilizers,pesticides to the water body,which gradually leads to the deterioration of water quality.Rainfall and water quality changes have a significant correlation in the time dimension.Through the research of this paper,the water quality monitoring covering the entire water space is realized,and the validity of MEMS sensor monitoring data is proved.Combining the real-time rainfall prediction information and the correlation analysis between rainfall and water quality can provide a new idea and method to reasonable and efficient water quality monitoring.