Constraining Regional Terrestrial Carbon Flux Estimates Through Assimilating Atmospheric Carbonyl Sulfide And Carbon Dioxide Data

The increase of atmospheric CO2 concentration is the main cause of global warming.Terrestrial ecosystems sequestrate about one fourth of the emitted CO2 from global anthropogenic activities every year,playing a crucial role in slowing down the increase of atmospheric CO2 concentration.Influenced by anthropogenic activities and climate changes,terrestrial carbon fluxes exhibit strong interannual variations.Accurate estimation of terrestrial carbon fluxes is the prerequisite for understanding the global carbon cycle and predicting future climate change.This study aims to:(1)improve North American terrestrial carbon flux estimates through developing an approach to employ atmospheric concentration observations(including atmospheric carbonyl sulfide and CO2)to simultaneously constrain photosynthetic and respiratory carbon fluxes,namely gross primary production(GPP)and ecosystem respiration(Reco),respectively;(2)characterize the interannual variations of North American net carbon uptake and the impact of droughts on it with the aid of emerging remote sensing data.The main contents and conclusions are summarized as follows:(1)Developed a regional carbon assimilation system-CTDAS-LagrangeBased on the Carbon Tracker Data Assimilation Shell(CTDAS),the Kalman filter smoothing algorithm and a Lagrangian particle diffusion model WRF-STILT,I developed a high-precision and high-efficiency regional carbon assimilation system.This system simultaneously optimizes surface carbon fluxes and lateral boundary conditions that affect the carbon flux optimization.With observing system simulation experiments(OSSE)and inversions using the real data,it is proved that the additive inflation factor is more flexible and effective than the multiplicative factor in optimizing the net fluxes of CO2.The OSSEs and an ensemble of sensitivity and uncertainty tests indicate that this system is able to effectively reduce the error in lateral boundary conditions and refine the uncertainties in prior carbon fluxes.The ensemble inversion tests place the North American net carbon fluxes as a sink with the magnitude from-0.92 to-1.26 PgC/yr,which is in good agreement with the estimates from other atmospheric inversions.(2)Constrained terrestrial photosynthetic and respiratory carbon flux estimates for North America over 2007-2011 using atmospheric OCS and CO2 data and assessed the impact of uncertainties in input OCS fluxes on the estimatesWith CTDAS-Lagrange,I performed the inversions with OCS as the only constraint and analyzed the impact of uncertainties from a variety of prior and other input fluxes on the inversions.The results indicate that the optimized GPP shows a weak dependence on the prior plant OCS fluxes;the difference of the GPP estimates with four different soil fluxes could come to 67.3%with an average of 42.9%;the difference between the GPP estimates with two anthropogenic emission datasets reaches 19.1%;and the difference between the GPP estimates with two ocean fluxes is about 7.5%.On this basis,we developed a joint data assimilation method that simultanously ingested OCS and CO2 and analyzed the effects of these two signals in the joint assimilation.The results show that OCS plays a dominant role in the joint assimilation;the optimized GPP is in good agreement on the magnitude of seasonal variations with other independent estimates,i.e.the upscaled dataset FLUXCOM/MTE-GPP based on in-situ flux measurements and remote sensing data,GOME-2 solar-induced fluorescence(SIF)and the optimized GPP(GOPT)with GOME-2 SIF as the constraint;the spatial distribution of GPP during the peak growing season in the U.S.Corn-Belt area is well coincided with the simulations of the VPM model and the CASA-GFED3 model,which were reported to be reasonable,and the GOME-2 SIF and the GOPT GPP.Further investigations indicate the systematic underestimation of GPP during the peak growing season in the U.S.Corn-Belt area in the BEPS model can be effectively corrected after employing the OCS data as constraints;the optimized GPP with OCS shows more significant seasonal and interannual variations than those in the FLUXCOM/MTE-GPP dataset and the GOPT dataset;the optimized Reco shows good agreements with that in FLUXCOM on the amplitude of seasonal variations while a large difference on the spatial distribution.(3)Characterized the interannual variations of North American terrestrial carbon fluxes and the impacts of recent severe drought events on these changesBy integrating multiple carbon flux estimates,including outputs from atmospheric inversion models,terrestrial biosphere models and FLUXCOM,and remotely sensed data of hydrology and vegetation,I analyzed the anomalies of terrestrial carbon fluxes over North America in the recent years and the effects of droughts on the terrestrial carbon cycle of North America with a focus on the impacts of the two drought extreme events in 2011 and 2012.The model ensemble suggests an annual average increase of GPP(0.09±0.26 PgC/yr)and Reco(0.12±0.19 PgC/yr),and as a result a slight decrease of NEP(-0.03±0.08 PgC/yr)relative to the baseline 2008-2010 over North America during 2007-2014.The processes leading up to the anomalies are contrasting for the two years:in 2011,losses are attributed to decreased GPP(-0.17 PgC/yr)with increased Reco(+0.07 PgC/yr)mainly caused by a summer drought;in 2012,it is due to the fact that Reco increased(+0.48 PgC/yr)more than GPP(+0.31 PgC/yr)triggered by wanner spring.The warm spring compensated largely the negative carbon anomaly but also enhanced the summer drought in 2012.We further identified that Crops/Agriculture and Grass/Shrubs in the Temperate area contributed the largest proportion to total annual carbon uptake reductions and also dominated the interannual variations of net carbon uptake during the study period.Finally,the dominance of these ecoregions in driving carbon cycle anomalies highlights that spring-summer compensatory dynamics can only be seen as an ecoregion-specific phenomenon,rather than a continental-scale event.