Study On The Development For A Regional High-resolution Assimilation System And Anthropogenic Carbon Emission Estimation

The atmospheric flux inversion method has proved to be one of the effective ways to obtain surface carbon sources/sinks.However,current researches on the carbon source/sink estimation are mainly focused on global or continental scales,whose inversions generally have coarse spatial resolutions that cannot meet the requirements in regional carbon source/sink researches.There are mainly two reasons for the current situation:?1?the sparse and uneven spatial-temporal distribution of observations;?2?massive resource consumption for high-dimension and high-resolution computing.Therefore,it is urgent to develop high-resolution assimilation inversion systems for deeper researches on carbon source/sink estimation.In this thesis,according to the principle of order-reduction in the proper orthogonal decomposition?POD?,an efficient assimilation inversion algorithm,POD4DVar,was developed.Then a regional high-resolution carbon assimilation inversion system,Tracerstracker,was constructed by coupling the POD4DVar and the CMAQ model.According to the characteristics of CO in transportation and variation,the flux inversion method was given and regional CO fluxes were optimized.Besides,in order to overcome the dilemma of insufficient CO₂ observations,two CO-based joint assimilation methods,i.e.,atm-based method and emis-based method,were proposed for CO₂ flux inversion.Two study regions with different geographical and climatic conditions,i.e.,Shangdianzi and Waliguan,were chosen for verifying the efficiency and accuracy of flux inversions by the Tracerstracker system.CO and CO₂fluxes inversions were obtained by the Tracerstracker system,based on which the anthropogenic and natural fluxes of CO₂ were quantitatively separated in the two study areas.Detail contents and results in this thesis are as follows:?1?A regional high-resolution carbon assimilation inversion system was constructed based on POD4DVar.It is very difficult to develop and maintain the tangent linear adjoint model for traditional four-dimensional variational assimilation method,besides,the high-dimensional computing makes the assimilation inversion procedure inefficient,these problems hinder greatly the spread of the method in realistic applications.In this thesis,the efficient inversion algorithm,POD4DVar,was developed by combing the POD and the 4DVar,then massive computing problem can be solved in a relatively low dimension.Afterwards,a regional high-resolution assimilation inversion system,Tracers Tracker,was conducted by coupling the POD4DVar algorithm and the CMAQ model.Validation experiments were conducted for the Tracers Tracker system and results show that the system can effectively absorb the information from observations for eliminating"errors"in the prior fluxes,and then improve significantly the accuracy of the posterior simulations.?2?CO flux inversions were obtained with a high-resolution of 3 km by the Tracers Tracker system in the two study areas.In this thesis,31 groups of CO ground-based observations were assimilated by the Tracers Tracker system,and CO fluxes with a high-resolution of 3 km were obtained in the two study areas.The inversion results show that the CO posterior fluxes are significantly higher than the priors,increase proportions are 32.7%and 62.3%in Shangdianzi and Waliguan respectively.The CO fluxes are more fit with realistic variations of CO concentrations,the MAEs of the posterior simulations are reduced by 23%and 27%in Shangdianzi and Waliguan respectively.The correlation Rs increase from 0.6 to 0.79 in Shangdianzi and from 0.66 to 0.76 in Waliguan respectively.?3?A CO₂ flux inversion method based on CO₂/CO emission ratios was proposed,high-resolution CO₂ flux inversions were obtained by this method and uncertainties analyses were conducted for the inversion results.In order to improve the resolution of CO₂ fluxes effectively,a CO₂ flux inversion method based on CO₂/CO emission ratios was proposed.In this method,the problem due to insufficient CO₂ observations was avoided,the CO₂/CO emission ratios extracted from the inventory were used as a variable in the assimilation procedure,CO₂ flux inversions with a high resolution of 3km were obtained in Shangdianzi and Walguan respectively.Validation results show that the CO₂ fluxes obtained by this method greatly correct the"errors"in the prior fluxes,correlations are also improved significantly in the CO₂posterior simulations.The MAEs of the posterior simulations decrease by 65%and 62%in Shangdianzi and Waliguan respectively,the correlation Rs increase from 0.67 to 0.88 in Shangdianzi and from 0.69 to 0.89 in Waliguan respectively.?4?A CO₂ flux inversion method based on the CO₂/CO concentration ratios was proposed,high-resolution CO₂ fluxes were obtained by this method and uncertainties analyses were conducted for the inversion results.Due to the sparse and uneven distribution of CO₂ observations,researches on regional CO₂ flux inversions cannot be carried out in realistic applications.In this method,monthly CO₂/CO concentration ratios were extracted from current limited ground-based and satellite observations,"CO₂ observations"were reckoned based on above rations and abundant CO observations.Then these"CO₂ observations"were assimilated by the Tracers Tracker system,CO₂ fluxes with a high resolution of 3 km were obtained in Shangdianzi and Waliguan respectively.Validation results show that the CO₂ flux obtained by this method can optimize the prior fluxes in a few periods,but there are still significant instabilities in the whole assimilation procedure.Errors of the posterior simulations are not reduced obviously,only 2%and 3%in Shangdianzi and Waliguan respectively.The correlations Rs change from 0.67 to 0.66 in Shangdianzi,from 0.69 to 0.68 in Waliguan,lower correlations appear in both two study areas.?5?Anthropogenic and natural CO₂ fluxes were quantitatively separated.The anthropogenic and natural CO₂ fluxes were quantitatively separated based on the CO₂posteriors obtained by the two inversion methods.Results show that it is an obvious carbon source with an increasing trend in Shangdianzi,the total CO₂ fluxes obtained by the emis-based method and atm-based method are 38.1 mole/s and 36.5 mole/s respectively,the rations of anthropogenic to natural are 5:1 and 4.9:1 respectively.It is a small carbon sink with a significantly decreasing trend in Waliguan,the total CO₂fluxes obtained by the emis-based method and atm-based method are-4.0 mole/s and-4.4 mole/s respectively,the rations of anthropogenic to natural are 1:1.8 and 1:2respectively.There are 79 diagrams,36 charts and 204 references in this thesis.