Simulation Study On The Temporal Dynamics Of Methane Flux In The Submerged Macrophyte Area Of Taihu Lak

Lake is one of the main natural sources of CH₄ in the atmosphere.Simulating the temporal dynamic of CH₄ flux and CH₄ emission in lake is important for clarifying the processes of CH₄ production,transport and emission,and estimating the changes of CH₄emission and carbon balance in the lake ecosystem under current climate and climate change in future.However,there are still great challenges in simulating CH₄ dynamic by using models in lake,especially in the lakes where macrophytes grow and ebullition is main emission pathway of CH₄.In this study,the submerged aquatic vegetation zone of Lake Taihu was selected.We identified the main influential factors of CH₄ flux at the half hour scale by using the CH₄ flux data measured by eddy covariance and environmental factors data at the Bifenggang（BFG）site of Taihu eddy flux network during the observation period from 2014to 2017.The empirical model of CH₄ flux was constructed,and the LAKE 3.0 process model was used to simulate the CH₄ flux in this region.Based on the comprehensive evaluation of the results simulated by the empirical model and the process model LAKE 3.0 in the submerged aquatic vegetation zone of Lake Taihu,the temporal dynamics of the CH₄ flux in this region were simulated,and the annual accumulated flux was estimated.The main results of this study were as follows:（1）The influential factors of CH4 flux in the submerged aquatic vegetation zone of Lake Taihu.At the half hour scale,the dynamic variation of total CH₄ flux in the growing season was explained by sediment temperature（T_s）,friction velocity（u_*）,air temperature（T_a）,relative humidity（RH）,latent heat flux（LE）and water temperature at 20 cm(T_W20),while CH₄ bubble flux was controlled by T_s,u_*,T_a and T_W20,and CH₄ diffusion flux was controlled by wind speed（WS）,T_s,sensible heat flux（H）,LE and RH.In the non-growing season,the total CH₄flux was affected by u_*,RH,LE,WS,H and T_s,while the main controlling factors of CH₄bubble flux were u_*,RH,LE,WS,T_s,T_a and air pressure（P）,the CH₄ diffusion flux was affected by u_*,H,P,WS,RH,T_s and T_W20.（2）The construction and validation of CH4 flux empirical model in the submerged aquatic vegetation zone of Lake Taihu.At half hour scale,the total CH₄ flux continually multiplying model LCMT2 and the CH₄bubble flux continually multiplying model LCMB2 constructed with T_s and u_*,and the CH₄diffusion flux continually multiplying model LCMD2 constructed with T_s and WS met the requirements of optimal model effect and the least drive variables.The LCMT2 can be selected as the optimal continually multiplying model of total CH₄ flux,and LCMB2+LCMD2 can be selected as the optimal continually multiplying model of total CH₄flux considering bubble and diffusion pathways at the half hour scale.Based on BP artificial neural network（BPNN）,in the growing season,the total CH₄ flux model BPGT4(T_s,u_*,T_a,T_W20,RH,P,WS,H,LE,DOY,h)can be selected as the optimal total CH₄ flux BPNN model in submerged aquatic vegetation zone of Lake Taihu during the growing season,while the bubble flux model BPGB1(T_s,u_*,T_a,T_W20,DOY,h)and the diffusion flux model BPGD4(T_s,u_*,T_a,T_W20,RH,P,WS,H,LE,DOY,h)can be selected to construct the total CH₄ flux BPNN model（BPGB1+BPGD4）considering bubble and diffusion pathways during the growing season.In the non-growing season,the total CH₄ flux model BPNT3(T_s,u_*,T_a,T_W20,RH,P,WS,DOY,h)can be selected as the optimal total CH₄ flux BPNN model during the non-growing season,while the bubble flux model BPNB1（u_*,RH,T_s,T_a,WS,DOY,h）and the diffusion flux model BPND2(u_*,H,P,WS,RH,T_s,T_W20,DOY,h)can be selected to construct the total CH₄ flux BPNN model（BPNB1+BPND2）considering bubble and diffusion pathways during the non-growing season.At the half-hour scale,based on random forest（RF）,in the growing season,the total CH₄ flux model RFGT3(T_s,u_*,T_a,T_W20,RH,P,WS,DOY,h)can be selected as the optimal total CH₄ flux RF model in submerged aquatic vegetation zone of Lake Taihu during the growing season,while the bubble flux model RFGB3(T_s,u_*,T_a,T_W20,RH,P,WS,DOY,h)and the diffusion flux model RFGD1(WS,T_s,RH,T_a,T_W20,DOY,h)can be selected to construct the total CH₄ flux RF model（RFGB3+RFGD1）considering bubble and diffusion pathways during the growing season.In the non-growing season,the total CH₄ flux model RFNT3(T_s,u_*,T_a,T_W20,RH,P,WS,DOY,h)can be selected as the optimal total CH₄ flux RF model during the non-growing season,while the bubble flux model RFNB3(T_s,u_*,T_a,T_W20,RH,P,WS,DOY,h)and the diffusion flux model RFND3(T_s,u_*,T_a,T_W20,RH,P,WS,DOY,h)can be selected to construct the total CH₄ flux RF model（RFNB3+RFND3）considering bubble and diffusion pathways during the non-growing season.（3）Simulation and verification of CH4 flux in submerged aquatic vegetation zone of Lake Taihu using LAKE 3.0 process model.Simulated by LAKE 3.0 model,the CH₄ concentration of lake surface ranged from 0.66to 599.42 nmol/L,the total CH₄ flux ranged from 0.0002 to 3.88μg/（m²·s）,the CH₄ diffusion flux ranged from 0.0002 to 3.42μg/（m²·s）,while the CH₄ bubble flux ranged from 0 to 1.20μg/（m²·s）.The simulation results showed that,bubbling was the main emission pathway of CH₄ in the submerged aquatic vegetation zone of Lake Taihu,and 73.07%of CH₄ flux was emitted from bubbles.Compared with the measured values,the simulated CH₄ diffusion flux was better than the simulated CH₄ bubble flux.Under the conditions with high temperature,strong upward long-wave radiation and low air pressure,the bubble flux was underestimated when the higher total CH₄ flux was observed,and the diffusion flux was overestimated when the lower total CH₄ flux was observed.The CH₄ flux simulated by LAKE 3.0 model in the submerged aquatic vegetation zone of Lake Taihu was more sensitive to the change of parameter that characterized the mass and quantity of organic matter in sediment as an active substrate for CH₄ production P₀.When P₀ was changed by 10%,the total CH₄ flux,bubble flux and diffusion flux at the water-air interface varied by 11.4%,6.6～6.7%and 10.1%,respectively.（4）Comprehensive evaluation and discussion of CH4 flux model in the submerged aquatic vegetation zone of Lake Taihu.Based on the comprehensive evaluating the CH₄ flux simulated by empirical model and process model in submerged vegetative region in Lake Taihu,the simulation performed by the machine learning model was better than that simulated by nonlinear continually multiplying model.RF model had the highest simulation accuracy among all empirical model methods.By analyzing the influential factors of CH₄ flux from different emission paths in the submerged aquatic vegetation zone of Lake Taihu,the empirical model of CH₄ bubble flux and diffusion flux was constructed respectively.The total CH₄ flux model was set up by the CH₄ bubble flux model and diffusion flux model can improve the simulation of the empirical model in the submerged vegetation zone of Lake Taihu.In addition,the CH₄ flux,bubble flux and diffusion flux simulated by empirical model and process model showed the"Ω"type temporal dynamic characteristics in the submerged aquatic vegetation zone of Lake Taihu.Moreover,based on empirical model 1（Emission paths were not differentiated）,empirical model 2（emission path were differentiated）and LAKE 3.0 process model,the annual accumulative CH₄ flux of submerged aquatic vegetation zone of Lake Taihu in 2015 was 7.03±7.81,6.07±7.00 and 10.10±9.33 g CH₄/（m²·a）,respectively.Through comprehensive comparison,it is found that the annual total CH₄ estimation of LAKE 3.0 model had a high value.Thereby,it was necessary to further improve the construction of the CH₄ flux process model by clarifying the transport capacity of CH₄flux in the shallow lakes and the flow resistance of submerged plants to the lake.