Evaluation And Optimization Of The Biome-BGC Model Under Different Meteorological Conditions

The Biome-Biogeochemical Cycles?Biome-BGC?model is based on plant physiological and ecological processes,and has been widely applied to describe the behaviors and characteristics of terrestrial forest ecosystems,and to evaluate and predict regional and global carbon cycle processes as well as their spatio-temporal patterns under global climate change.To date,the model is still an important tool for analyzing and predicting large-scale ecosystem processes.Most previous studies had focused on the model evaluation or the model application at seasonal and annual scales.While the performance of the model under different meteorological conditions have less been considered.In this study,data from four FLUXNET sites?Lavarone,Renon,Loobos,Fyodorovskoye?were used to evaluate the performance of the Biome-BGC model on daily primary productivity?GPP?in 2011-2013.The forest transpiration?Tr?of an evergreen broad-leaved forest site?Xiaoqingshan,Guangzhou?was simulated with the input of the meteorological data in 2011 before it was verified by measured forest transpiration data based on sap flow monitoring.In addition,the accuracy of the daily GPP and daily Tr under different meteorological conditions were also analyzed to evaluate the performances of carbon and water flux simulations.Besides EFAST sensitivity analysis was adopted to optimized the parameter in the original model,before it was used to predict the impact of future climate changes on forest GPP.The detailed results were presented below:?1?EFAST sensitivity analysis is used to quantitatively analyze the individual and reciprocal impacts for the different parameters during the GPP and Tr simulations.The results showed that the annual leaf and fine root turnover?LFRT?made up the most of the contributions to the GPP and Tr simulation?by 29.8%and 28.7%?,which followed by the ratio of fine root C to leaf carbon?FRC:LC??by 20.5%and 23.1%?,as well as the specific leaf area?SLA?.Vapor pressure deficit at complete conductance reduction?VPD_f?,Maximum stomatal conductance(g_s,max),etc.The water interception coefficient(W_int)and the predawn leaf water potential at final reduction of stomatal conductance??_f?have a greater impact on the Tr than the GPP.Besides,the second-order sensitivity index(S_Ti)for each parameter was higher than the first-order sensitivity index?S_i?,implying the stronger reciprocal effects than the individual effects for the simulation of GPP and Tr.?2?Biome-BGC model simulated of daily GPP from 2011 to 2013 shows that although the simulated GPP was highly related to the GPP flux measurements?r=0.873⁰.945,P<0.01?,the former was smaller than the later?4%²5%?.The RU-Fyo site had the best GPP simulation accuracy among the four sites,with the fitting dependency average,maximum and minimum GPP(r=0.90,RMSE=1.55gCm^-2d^-1,aSAE=1.05,P<0.01).Of note,the model simulated Tr in the Xiaoqingshan site in 2011 was weak related to the sap flow based stand transpiration?r=0.33,RMSE=0.94 mm,P<0.01?,and the simulation value is overestimated by 7%compared with the experimental verification data.?3?Through the clearness index Kt,typical sunny days were selected for evaluating the model under sunny days and rainy days.The results showed that the simulation of GPP in the rainy days were better than that without precipitation(r_{no rain}=0.843⁰.936 and r_rain=0.887⁰.952,P<0.01),and the simulation accuracy of typical rainy days is higher than that of typical sunny days(r_sun=0.830⁰.915 and r_rain=0.887⁰.952,P<0.01).Furthermore,found that the daily GPP were overestimates by 7%²5%when the VPD is high?VPD>0.6Kpa?.The simulation of Tr under different meteorological conditions also showed better simulation under the condition of precipitation(r_{no rain}=0.14<r_rain=0.58,P<0.01).Thus,although the simulated results based on the Biome-BGC model has a good correlation with the measured data,bias still existed among the data under different meteorological conditions are not ideal,and the model tended to be more applicable on rainy days.?4?The optimized Biome-BGC model based on the changes of sensitive parameters to different meteorological conditions was used to re-simulate the typical sunny GPP of each station in 2011-2013.The accuracy of simulated GPP and Tr were significantly improved?by 3.2%⁵.7%and 6.5%?when the meteorological data input to the model were classified by the climate conditions.The annual GPP and Tr simulation were also improved?0.6%².2%and 4.3%?.The results indicated that it would be better to do the modelling separately under different meteorological conditions considering the different parameters sensitivity to the climatic conditions.?5?The dynamics of daily mean temperature and precipitation at each site during the period of 2011 to 2013 was analyzed to determine the response of carbon flux to these meteorological factors.The results indicated that the temperature and precipitation were significantly different among the four stations.The annual mean temperature was relatively stable during the period.The highest annual mean temperature was found in NL-Loo station?9.8??.On the contrary,variations of precipitation was more significantly different among different sites and years.The daily GPP was positively correlated with daily mean temperature?r=0.82⁰.93,P<0.01?,but not with precipitation.?6?By modelling the GPP under the scenes of future climate change,we found that the increase of temperature,precipitation and atmospheric CO₂ concentration would promote the accumulation of GPP in the four study areas.Temperature and CO₂concentration were the dominant meteorological factors contributed to the GPP simulations in these regions.In this thesis the simulation accuracy of a typical daily gas exchange model Biome-BGC under different meteorological conditions had been analyzed.The obtained results are helpful to understand the influence of meteorological conditions on gas exchange estimation,and to clarify the applicability of simulated carbon and water flux under different meteorological conditions.They also help to understand the response of terrestrial ecosystems in response to climate change and human activities.