Study On The Water Quality Of Qinghe Reservoir Inversion Based On Landsat8

The study aims to establish an optimum model for the total nitrogen and total phosphorus in Qinghe reservoir.Single band,band combination and least squares support vector machine were built,and proved.Examines the three models’ relative error between predicted and measured values.The most suitable model is deduced,which is the least squares support vector machine model.There are three research contents including laboratory analysis of measured data,band sensitivity analysis and building single band,band combination and least squares support vector machine.The research contents are as follows.A total of 80 sets of data were collected in June,July,September and October 2015.The highest average nitrogen content in June is 1.73mg/L and the average value of total nitrogen content in the year of September is 1.11mg/L.The highest average value of total phosphorus in June is 0.07mg/L and total phosphorus content of the other three months are 0.04mg/L.Analyze the sensitivity of single band and band combination.B4 band has the highest correlation with total phosphorus and B5 band has the highest correlation with total nitrogen.In the band combination,the best correlation with TP in summer was B4/B5,and the best correlation with total nitrogen was B3/B4.The best correlation with total phosphorus in autumn was B4/B5,and the best correlation with total nitrogen was B3+B4.Analyze the relative errors between the measured values and the predicted values of the three models.The maximum error of total nitrogen is 71.49%,and the average value is up to 27.21%in summer.The maximum error value of single band model inversion is 92.25%,and the average value is up to 36.12%in autumn.For band combination the maximum error of total nitrogen was 42.73%,and the average value was up to 20.19%in summer.The maximum error value of band combination model inversion is 67.43%,and the average value is up to 24.23%in autumn.The maximum error value of the total nitrogen was 17.73%and the average value was 7.10%by the least square support vector machine model in summer.The maximum error of the inversion of the least squares support vector machine model was 17.91%,and the average value was the highest value of 6.70%in autumn.Therefore,the inversion result is better than the band combination model,and the band combination model is superior to the single band model.The single band model and band combination model can not be used in the inversion of total nitrogen and total phosphorus in Qinghe reservoir.The inversion accuracy is too low to reach the requirement of inversion precision.The least squares support vector machine model can be applied to the inversion of total nitrogen and total phosphorus in Qinghe reservoir.It provides an effective monitoring method and theoretical basis for the water quality monitoring of Qinghe Reservoir in the future.