Research On The Evolution Driving Factors Of Submerged Plants In Yilong Lake Based On Machine Learning Method And Pclake Model

Submerged plants are the core ecological components that maintain the health of shallow lake ecosystems.Yilong Lake is a typical plateau shallow lake.Studies have shown that in the past 10 years,Yilong Lake has experienced obvious disappearance and recovery of submerged plants.Among them,in 2009 Submerged plants died out in a large area and recovered significantly in 2013.The reasons for the significant changes in Yilong Lake submerged plants in the short term have not yet been clearly concluded.In order to explore the key factors affecting the growth and extinction of submerged plants in Yilong Lake,and to provide a basis for the management of submerged plants in Yilong Lake,this study took Yilong Lake as the research object,and collected the total phosphorus,total nitrogen,water depth,Water quality data such as temperature,transparency,etc.,and the change data of submerged vegetation coverage in the past 25 years were inverted through remote sensing images,and the influence of submerged vegetation in Yilong Lake in recent years was statistically analyzed by using generalized additive model,decision tree,structural equation and other methods.The main driving factors of the change,the Yilong Lake ecosystem model was established to simulate the growth of submerged vegetation under different levels of driving factors.The research results are as follows:(1)Based on Landsat TM/ETM+/OLI remote sensing images,a decision tree classification model based on maximum synthetic year NDVI and minimum synthetic year NDVI was constructed,and remote sensing inversion of the evolution process of aquatic plants in Yilong Lake from 1995 to 2020 was carried out.,the overall classification accuracy reaches 90.09%,and the Kappa coefficient is 0.8392;(2)From 1995 to 2020,the submerged vegetation of Yilong Lake changed significantly.From 1995 to 2004,the submerged vegetation of Yilong Lake showed a significant upward trend.From 2004 to 2012,the submerged vegetation coverage fluctuated greatly,but the interannual The trend of change is not obvious.From 2012 to 2020,the vegetation coverage showed a significant downward trend.Through the analysis of generalized additive model,decision tree and structural equation model,it was found that in Yilong Lake,total phosphorus(TP),underwater light intensity and ammonia nitrogen(NH4)were significantly related to the growth of underwater vegetation,among which the underwater light environment It is measured by the quotient of transparency and water depth,that is,SD/Depth is the main driving factor for the growth of submerged vegetation,and the depth of water(Depth)has the most significant impact on submerged vegetation.(3)In the model calibration,the genetic algorithm is used instead of the usual random value selection method to find the optimal parameters,and the PCLake model is calibrated with sensitive parameters.Through the comparison of the detection value and the fitted value in 2017,19 were obtained through multi-generation screening.The specific value of the sensitive parameter.The model was verified and analyzed by using the measured water quality data and water body characteristic data(DO,SD)in 2018,and the fitting degree between the measured value and the simulated value was determined by the RMSE value.The model has better simulation results.In terms of model prediction,the results show that although the increase of water depth will correspondingly reduce the concentration of nutrients such as total phosphorus and nitrogen in the lake,the change of water depth reflects the change of submerged plant biomass is still very obvious.The biomass of submerged plants in low water level operation shows an increase year by year.Submerged vegetation at high water depths showed a decreasing trend.Maintaining a suitable water depth is conducive to the growth of submerged vegetation in Yilong Lake.