A method for estimating high-resolution atmospheric carbon dioxide concentrations over the land area of ​​China

Abstract

The present invention relates to a method for estimating high-resolution atmospheric carbon dioxide concentration in China's land area. The method comprises the following steps: (1) using satellite data to calculate vegetation index, surface moisture index, and vegetation phenology index to generate vegetation data files; (2) using the CAMS global greenhouse gas concentration dataset and the fifth-generation atmospheric reanalysis dataset of the European Centre for Medium-Range Weather Forecasts to generate carbon dioxide background field and meteorological background field files; (3) using the multi-scale emission inventory of Tsinghua University to generate anthropogenic carbon dioxide emission data files; (4) constructing an optimized atmospheric carbon dioxide simulation system; and (5) inputting the vegetation data files, carbon dioxide background field and meteorological background field files, and anthropogenic carbon dioxide emission data files into the optimized atmospheric carbon dioxide simulation system and running the system to obtain high-resolution atmospheric carbon dioxide concentration estimates for China's land area. The present invention can significantly improve the ability to estimate atmospheric carbon dioxide concentration in China's land area.

Description

Technical Field

[0001] The present invention relates to a method for producing data in the fields of ecological environment and atmospheric science, and in particular to a method for estimating high-resolution atmospheric carbon dioxide concentration in land areas of China. Background Art

[0002] The range of issues caused by climate change has had a significant impact on human life and development, garnering significant attention from governments and the scientific community worldwide (Fang et al., 2011). Atmospheric carbon dioxide, the primary greenhouse gas, has a significant impact on global climate change (Zimov et al., 2006). Atmospheric carbon dioxide concentrations have increased by over 50% compared to pre-industrial levels (Friedlingstein et al., 2023). Countries around the world have been making various efforts over the years to control carbon dioxide emissions and mitigate global warming. Therefore, developing appropriate strategies to control carbon dioxide emissions and accurately estimating atmospheric carbon dioxide concentrations and their spatial and temporal distribution are crucial (Deng et al., 2013).

[0003] Currently, atmospheric carbon dioxide (CO2) estimates are primarily based on site observations, satellite remote sensing, and numerical model simulations (Piao Shilong et al., 2022). Ground-based observations can provide relatively accurate atmospheric CO2 concentration values ​​with high temporal resolution and good continuity. However, due to their limited spatial distribution, ground-based observations are unable to reflect large-scale regional distribution characteristics (Wang et al., 2020). Compared to site observations, satellite remote sensing has a higher coverage density and can provide regional CO2 distribution information. However, due to its specific orbital pattern and observation period, it is unable to provide continuous high-temporal resolution observations over the same region (Zhao et al., 2023). Numerical models can simulate regional and even global atmospheric CO2 concentrations in a continuous manner, making them an effective tool for studying the distribution and spatiotemporal variations of atmospheric CO2 concentrations.

[0004] Global numerical models often have large uncertainties in their estimates of atmospheric CO2 concentrations in specific regions because they cannot simulate at high spatial resolution and have difficulty in accounting for the influence of small- and medium-scale weather processes (Geels et al., 2007). In contrast, regional models can simulate at higher spatial and temporal resolutions and better account for small- and medium-scale weather systems, enabling more accurate descriptions of CO2 emissions and transport in the atmosphere over specific regions. Currently, a number of regional models are capable of estimating atmospheric CO2 concentrations, including RAMS-SIB2 (Denning et al., 2003), RAMS-SIB3 (Corbin et al., 2010), RAMS-CMAQ (Xing et al., 2013), WRF-CMAQ (Li et al., 2017), and WRF-Chem (Ahmadov et al., 2009).

[0005] Among regional models, the WRF-Chem model calculates biogenic CO2 concentrations based on the VPRM model. Because it does not require complex parameters and its simulation results are accurate and reliable, it is widely used in atmospheric CO2 research. Even in areas with complex topography, WRF-Chem has been shown to estimate atmospheric CO2 concentrations that are consistent with observations (Pillai, 2011). However, the model uses a simple linear formula to calculate ecosystem respiration, which leads to certain errors in the CO2 uptake and release by terrestrial ecosystems (Hu et al., 2020). Furthermore, using appropriate parameters across different study areas is crucial for the accuracy of model estimates of atmospheric CO2 concentrations (Mahadevan et al., 2008). Previous studies of atmospheric CO2 over China have mostly used parameters derived from international sites (Dong et al., 2021; Li et al., 2020) or fitted parameters for specific local underlying surface types (Dayalu et al., 2018). The WRF-Chem model used a simple linear formula to calculate ecosystem respiration. Therefore, to accurately estimate the atmospheric carbon dioxide concentration in China's land area, a more complete ecosystem respiration calculation formula and calculation parameters applicable to the region are needed.

[0006] The WRF-Chem model's estimation of atmospheric carbon dioxide relies on WRF-simulated meteorological fields (temperature, relative humidity, wind, etc.) and downward shortwave radiation. Studies have found that direct model simulations of meteorological fields and photosynthetically active radiation exhibit significant errors (Ahmadov et al., 2007; Dayalu et al., 2018). Nudging is a method for improving model meteorological simulations. Its core idea is to bring model results closer to observations by adding a false bias term proportional to the difference between the forecast and observations to the prediction equation. Nudging has been shown to effectively improve the WRF model's meteorological simulations (Ma et al., 2016; Mai et al., 2017). However, few studies have used nudging, particularly observational nudging, to improve meteorological simulations in atmospheric carbon dioxide estimation. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to provide a method for estimating the high-resolution atmospheric carbon dioxide concentration in the land area of ​​China with higher accuracy.

[0008] To solve the above problems, the present invention provides a method for estimating high-resolution atmospheric carbon dioxide concentration in China's land area, comprising the following steps:

[0009] (1) Use satellite data to calculate vegetation index, surface moisture index, and vegetation phenology index, and distribute the calculated data together with land type data to the numerical model grid to generate the vegetation data file required for the numerical model operation;

[0010] (2) The CAMS global greenhouse gas concentration dataset from the European Centre for Medium-Range Weather Forecasts (ECMWF) was used as the original CO2 background field, and the fifth-generation atmospheric reanalysis dataset (ERA5) from the European Centre for Medium-Range Weather Forecasts (ECMWF) was used as the original meteorological background field. The data of the original CO2 background field and the original meteorological background field were interpolated onto the numerical model grid using the inverse distance weighted method to generate the CO2 background field and meteorological background field files.

[0011] (3) The Multi-scale Emission Inventory (MEIC) from Tsinghua University was used as the anthropogenic emission inventory data. The monthly resolution anthropogenic emission inventory data was first converted to hourly resolution by linear interpolation. The anthropogenic emission inventory data was then interpolated onto the numerical model grid using the inverse distance weighting method to generate anthropogenic carbon dioxide emission data files.

[0012] (4) By improving the calculation formulas for photosynthesis and respiration of terrestrial ecosystems and improving the three-dimensional temperature field calculated by numerical models, an optimized atmospheric carbon dioxide simulation system is constructed; among which:

[0013] ① In the WRF-Chem model, the calculation formula for terrestrial ecosystem photosynthesis is:

[0014]

[0015] PAR new =0.77×PAR model +7.36

[0016] Where: GEE is the photosynthesis efficiency of terrestrial ecosystems, unit: μmol m -2 s -1 ;λ is the maximum light energy utilization rate, unit is μmol CO2 m -2 s -1 / μmol PAR mm -2 s -1 PAR stands for photosynthetically active radiation; PAR0 is the photosynthetically active radiation when photosynthesis reaches half-saturation, in μmol mm -2 s -1 ;T scale is the temperature impact index; P scale represents vegetation phenology index; W scale1 LSWI is the photosynthesis water stress index; max The maximum value of the multi-year surface moisture index; PAR newis the corrected light and effective radiation, unit is μmolm -2 s -1 ;;PAR model Light and effective radiation calculated for the original numerical model;

[0017] ②The calculation formula for terrestrial ecosystem respiration is:

[0018] R=β+α1×T′+α2×T′ 2 +γ×EVI+k1×W scale2 +k2×W scale2 ×T′+k3×W scale2 ×T′ 2

[0019]

[0020] When T a ≥T c Time: T′=T a

[0021] When T a <T c Time: T′=T c -T m ×(T c -T a )

[0022] Where: T' is the temperature of the improved pair, unit: ℃; W scale2 is the respiratory water stress index; β, α1, α2, γ, k1, k2, and k3 are calculation parameters applicable to different underlying surface vegetation types in China's land area obtained by fitting the productivity observation data of different vegetation types in China over many years; LSWI min is the minimum value of the multi-year surface moisture index; T a T is the atmospheric temperature at 2 m, in °C; c is the low temperature threshold, unit is ℃; T m The atmospheric temperature is lower than T c The empirical coefficient when

[0023] ③ Improve the three-dimensional temperature field calculated by the numerical model:

[0024] The model-estimated 2-meter temperature was improved by assimilating the observed 2-meter temperature data at the stations using the observation nudging method. The model-estimated upper-level temperature was improved by bringing the model-estimated upper-level temperature closer to the upper-level temperature in the European Centre for Medium-Range Weather Forecasts (ERA5) reanalysis data using the gridded nudging method. Ultimately, the three-dimensional temperature field calculated by the improved numerical model was obtained.

[0025] (5) Input the vegetation data files, carbon dioxide background field and meteorological background field files, and anthropogenic carbon dioxide emission data files into the optimized atmospheric carbon dioxide simulation system and run it to obtain the high-resolution atmospheric carbon dioxide concentration estimation results for the Chinese land area.

[0026] In step (1), the vegetation index, surface moisture index, and vegetation phenology index are calculated using MODIS satellite data according to the following formula:

[0027]

[0028] Where: EVI represents the enhanced vegetation index, ranging from -1 to 1; ρ mir represents the reflectivity in the near-infrared band; ρ red Represents the reflectivity of red light band; ρ blue represents the reflectance of the blue light band; G is the gain factor, which is 2.5; C1 and C2 are the correction parameters for atmospheric correction of red light and blue light, which are 6 and 7.5 respectively; L is the soil correction parameter, which is 1; LSWI represents the surface moisture index, which ranges from -1 to 1; ρ swir Indicates the reflectivity of the short-wave infrared band; P scale Represents the vegetation phenology index.

[0029] The numerical model grid in step (1) refers to the Lambert projection, with a resolution of 12×12 km, a grid number of 354×415, and a center point of 36.461°N, 102.53°E.

[0030] The value ranges of λ and PAR0 in step ① are shown in Table 1:

[0031] Table 1 Photosynthesis calculation parameters

[0032]

[0033] The value ranges of β, α1, α2, γ, k1, k2, and k3 in step ② are shown in Table 2:

[0034] Table 2 Respiration calculation parameters

[0035]

[0036] Compared with the prior art, the present invention has the following advantages:

[0037] 1. Based on the mesoscale numerical model WRF-Chem, this paper constructs an optimized atmospheric carbon dioxide simulation system by improving the vegetation respiration calculation formula, fitting calculation parameters suitable for the Chinese land area, providing temperature fields and photosynthetically active radiation that are closer to actual observations for model calculations, and adopting a more refined atmospheric carbon dioxide background field.

[0038] 2. The present invention combines observational nudging and grid nudging methods to improve the meteorological field simulated by the model. It further uses multi-year station observation data to correct the photosynthetically active radiation in the model, thereby improving the model's ability to simulate atmospheric carbon dioxide in China's land area from multiple perspectives. Compared with other methods, the present invention can obtain three-dimensional atmospheric carbon dioxide concentration estimates with high temporal and spatial resolution, effectively improving the ability to estimate atmospheric carbon dioxide concentration in China's land area. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0040] Figure 1 Flowchart of the present invention.

[0041] Figure 2 Hourly 2-meter temperature, light, and active radiation time series estimated by this method for different stations in July 2016 compared with observations. In the figure, OBS represents observations, CTL represents the control experiment, which indicates the model results before optimization, and VPRM_C represents the estimated results of this method. (a) 2-meter temperature time series estimated from observations at Haibei Station and the VPRM_C experiment in July 2016; (b) 2-meter temperature time series estimated from observations at Huzhong Station and the VPRM_C experiment in July 2016; (c) 2-meter photosynthetically active radiation time series estimated from observations at Haibei Station, the control experiment, and the VPRM_C experiment in July 2016; (d) 2-meter photosynthetically active radiation time series estimated from observations at Huzhong Station, the control experiment, and the VPRM_C experiment in July 2016.

[0042] Figure 3 Comparison of the estimated daily two-meter temperature, light, and active radiation at different stations in July 2016 with observations. (a) Daily two-meter temperature estimated from observations at Haibei Station and the VPRM_C experiment in July 2016; (b) Daily two-meter temperature estimated from observations at Huzhong Station and the VPRM_C experiment in July 2016; (c) Daily average photosynthetically active radiation estimated from observations at Haibei Station, the control experiment, and the VPRM_C experiment in July 2016; (d) Daily average photosynthetically active radiation estimated from observations at Huzhong Station, the control experiment, and the VPRM_C experiment in July 2016.

[0043] Figure 4Comparison of hourly surface net ecosystem exchange time series estimated by this method and observations at different stations in July 2016. (a) Time series of net ecosystem exchange estimated from observations, controlled experiments, and the VPRM_C experiment at Haibei Station in July 2016; (b) Time series of net ecosystem exchange estimated from observations, controlled experiments, and the VPRM_C experiment at Huzhong Station in July 2016; (c) Daily average net ecosystem exchange estimated from observations, controlled experiments, and the VPRM_C experiment at Haibei Station in July 2016; (d) Daily average net ecosystem exchange estimated from observations, controlled experiments, and the VPRM_C experiment at Huzhong Station in July 2016.

[0044] Figure 5 Time series of daily atmospheric carbon dioxide concentrations estimated from observations, control experiments, and the VPRM_C experiment at the Waliguan station in July 2016.

[0045] Figure 6 Comparison of the July 2016 average atmospheric CO2 column concentration estimated by this method with observations and reanalysis data. (a) Satellite-observed July 2016 average atmospheric CO2 column concentration, (b) CarbonTracker reanalysis dataset-based July 2016 average atmospheric CO2 column concentration, (c) estimated from the control experiment, and (d) estimated from the VPRM_C experiment. DETAILED DESCRIPTION

[0046] like Figure 1 As shown in FIG, a method for estimating high-resolution atmospheric carbon dioxide concentration over the land area of ​​China includes the following steps:

[0047] (1) Use satellite data to calculate vegetation index, surface moisture index, and vegetation phenology index, and distribute the calculated data together with land type data to the numerical model grid to generate the vegetation data file required for the numerical model operation.

[0048] Among them: Vegetation index, surface moisture index, and vegetation phenology index are calculated using MODIS satellite data according to the following formula:

[0049]

[0050]

[0051] Where: EVI represents the enhanced vegetation index, ranging from -1 to 1; ρ nir represents the reflectivity in the near-infrared band; ρ red Represents the reflectivity of red light band; ρ bluerepresents the reflectance of the blue light band; G is the gain factor, which is 2.5; C1 and C2 are the correction parameters for atmospheric correction of red light and blue light, which are 6 and 7.5 respectively; L is the soil correction parameter, which is 1; LSWI represents the surface moisture index, which ranges from -1 to 1; ρ swir Indicates the reflectivity of the short-wave infrared band; P scale Represents the vegetation phenology index.

[0052] Land type data uses MODIS satellite observation data.

[0053] The numerical model grid refers to the Lambert projection with a resolution of 12 × 12 km, a grid number of 354 × 415, and a center point of 36.461°N, 102.53°E.

[0054] The vegetation score is obtained by calculating the proportion of underlying surface types in each numerical model grid. Finally, the vegetation index, surface moisture index, vegetation phenology index and vegetation score are integrated to generate the vegetation data file required for the operation of the numerical model.

[0055] ⑵ The CAMS global greenhouse gas concentration dataset from the European Centre for Medium-Range Weather Forecasts is used as the original carbon dioxide background field, and the fifth-generation atmospheric reanalysis dataset (ERA5) from the European Centre for Medium-Range Weather Forecasts is used as the original meteorological background field. The data of the original carbon dioxide background field and the original meteorological background field are interpolated onto the numerical model grid using the inverse distance weighted method to generate carbon dioxide background field and meteorological background field files.

[0056] (3) The multi-scale emission inventory (MEIC) from Tsinghua University is used as the anthropogenic emission inventory data. The monthly resolution anthropogenic emission inventory data is first converted to hourly resolution through linear interpolation, and then the anthropogenic emission inventory data is interpolated onto the numerical model grid through the inverse distance weighted method to generate anthropogenic carbon dioxide emission data file.

[0057] (4) By improving the calculation formulas for photosynthesis and respiration of terrestrial ecosystems and improving the three-dimensional temperature field calculated by numerical models, an optimized atmospheric carbon dioxide simulation system is constructed; among which:

[0058] ① In the WRF-Chem model, the calculation formula for terrestrial ecosystem photosynthesis is:

[0059]

[0060] PAR new =0.77×PAR model +7.36 Where: GEE is the photosynthesis of terrestrial ecosystems, unit: μmol m -2 S -1;λ is the maximum light energy utilization rate, unit is μmol CO2 m -2 s -1 / μmol PAR mm -2 s -1 PAR stands for photosynthetically active radiation; PAR0 is the photosynthetically active radiation when photosynthesis reaches half-saturation, in μmol mm -2 s -1 ;T scale is the temperature impact index; P scale represents vegetation phenology index; W scale1 LSWI is the photosynthesis water stress index; max The maximum value of the multi-year surface moisture index; PAR new is the corrected light and effective radiation, unit is μmolm -2 s -1 ;;PAR model Light and effective radiation calculated for the original numerical model.

[0061] In the present invention, the light and effective radiation observed at the site over many years are used to correct the light and effective radiation calculated by the numerical model, with the aim of providing light and effective radiation closer to the actual observations for the numerical model calculation of carbon dioxide concentration.

[0062] In the present invention, λ and PAR0 are obtained by fitting the observation data of productivity sites of different vegetation types in China for many years, as shown in Table 1:

[0063] Table 1 Photosynthesis calculation parameters

[0064]

[0065] ② In the original WRF-Chem model, the calculation formula for terrestrial ecosystem respiration is:

[0066] R=α×T+β

[0067] Where: R is the respiration of terrestrial ecosystems, unit: μmol m -2 s -1 ; α and β are parameters obtained by fitting the station observation data; T is the air temperature at 2m, unit ℃.

[0068] After improving the original formula, the present invention obtains the following calculation formula for terrestrial ecosystem respiration:

[0069] R=β+α1×T′+α2×T′ 2 +γ×EVI+k1×W scale2 +k2×W scale2 ×T′+k3×W scale2 ×T′ 2

[0070]

[0071] When T a ≥T c Time: T′=T a

[0072] When T a <T c Time: T′=T c -T m ×(T c -T a )

[0073] Where: T' is the improved temperature, unit: ℃; W scale2 is the respiratory water stress index; β, α1, α2, γ, k1, k2, and k3 are calculated parameters applicable to different underlying vegetation types in China's land area obtained by fitting the productivity observation data of different vegetation types in China over many years, as shown in Table 2; LSWI min is the minimum value of the multi-year surface moisture index; T a T is the atmospheric temperature at 2 m, in °C; c is the low temperature threshold, unit is ℃; T m The atmospheric temperature is lower than T c The empirical coefficient of .

[0074] Table 2 Respiration calculation parameters

[0075]

[0076] T of different underlying vegetation types c and T m As shown in Table 3:

[0077] Table 3 Temperature calculation parameters

[0078]

[0079] ③ Improve the three-dimensional temperature field calculated by the numerical model:

[0080] In the WRF-Chem model, the three-dimensional temperature field is used to calculate the carbon dioxide concentration absorbed and released by terrestrial ecosystems, as well as the diffusion and transport of total atmospheric carbon dioxide concentration. Therefore, the model-estimated 2-meter temperature is improved by assimilating observed 2-meter temperature data from the stations using the observational nudging method. The model-estimated upper-level temperature is then brought closer to the upper-level temperature in the European Centre for Medium-Range Weather Forecasts (ERA5) reanalysis data using the gridded nudging method. Ultimately, the model-estimated upper-level temperature is obtained, resulting in an improved three-dimensional temperature field calculated by the numerical model.

[0081] (5) Input the vegetation data files, carbon dioxide background field and meteorological background field files, and anthropogenic carbon dioxide emission data files into the optimized atmospheric carbon dioxide simulation system and run it to obtain the high-resolution atmospheric carbon dioxide concentration estimation results for the Chinese land area.

[0082] In order to verify the optimization effect of the present invention on estimating atmospheric carbon dioxide concentration in China's land area, a set of control experiments (CTL, simulated using the unoptimized model) and a set of optimization experiments (VPRM_C, simulated using the optimized model of the present invention) were designed to estimate the atmospheric carbon dioxide concentration in China's land area in July 2016. The detailed experimental design is shown in Table 4:

[0083] Table 4 Experimental design table

[0084]

[0085] Both simulations ran from June 20 to July 31, 2016. The first 10 days served as a spin-up period to allow the model to reach equilibrium. Results from July 1 to 31 were used for analysis. This paper is based on WRF-Chemv4.1.3. The carbon dioxide background field is derived from the 1.4° × 0.7° CAMS global greenhouse gas concentration dataset from the European Centre for Medium-Range Weather Forecasts, and the meteorological background field is derived from the 0.25° × 0.25° Fifth Generation Atmospheric Reanalysis (ERA5) dataset from the European Centre for Medium-Range Weather Forecasts. The model uses a single-layer nesting scheme with a horizontal resolution of 12 km, a 354 × 415 grid, 48 vertical layers, and an integration step of 40 seconds. The following model parameterization schemes were used: Morrison2 microphysics scheme, RRTM longwave radiation scheme, Dudhia shortwave radiation scheme, Noah land surface scheme, YSU planetary boundary layer scheme, and New Grell cumulus parameterization scheme.

[0086] Accurate 2m temperature, light and effective radiation are crucial for estimating carbon dioxide concentrations. The hourly 2m temperature, light and effective radiation time series for July 2016 estimated from site observations and experiments are shown in the table below. Figure 2 As shown in Figure 2, the optimized experiment simulated the 2m temperature at different stations, which was very close to the observation ( Figure 2 ab), the correlation coefficient (R, the higher the correlation between the data) of Haibei station reached 0.95, and the root mean square error (RMSE, the lower the better, indicating that the data difference is smaller) reached 1.82℃. The R of Huzhong station was 0.92, and the RMSE was 3.12℃. For photosynthetic active radiation, compared with the control experiment, the optimization experiment effectively improved the simulation ability of light and active radiation at different stations ( Figure 2cd), the RMSE at Haibei Station decreased by 15%, and the RMSE at Huzhong Station decreased by 28%. Subsequently, the simulated daily average 2m temperature, light and effective radiation of the present invention were further tested. The results showed that both groups of experiments could simulate the diurnal variation characteristics of 2m temperature, light and effective radiation close to the observed ones at different stations ( Figure 3 ), and the optimization experiment showed better simulation ability.

[0087] my country is rich in ecological resources, and net ecosystem exchange (NEE) can significantly affect atmospheric carbon dioxide concentrations. The net ecosystem exchange estimated by this method was tested. For the hourly NEE in July 2016, the optimization experiment significantly improved the model's ability to estimate NEE at different stations. The improvement in nighttime respiration at Haibei Station was particularly significant ( Figure 4 The NNE value in a is greater than 0), and the R with the observation is 0.85 and the RMSE is 4.74 μmol m -2 s -1 , effectively improving the estimated NEE value during the day at Huzhong Station ( Figure 4 (See the portion of the NNE value less than 0 in b) where the R value is 0.74, and the RMSE decreases by 45%. Similarly, for the daily mean NEE in July 2016, the optimization experiment produces more realistic estimates of diurnal variations, reaching R values ​​of 0.98 and a 46% reduction in RMSE at Haibei Station and 0.98 and a 74% reduction in RMSE at Huzhong Station.

[0088] To verify the atmospheric carbon dioxide concentration estimated by this invention, the hourly carbon dioxide concentration time series of ground stations in July 2016 ( Figure 5 ) and the distribution of carbon dioxide column concentration ( Figure 6 At the Waliguan station, the ground CO2 concentration estimated by the optimization experiment is more consistent with the observed value ( Figure 5 )R is 0.7, and RMSE is 4.21ppm. The distribution of carbon dioxide column concentration can reflect the distribution of carbon dioxide in the entire atmosphere. Compared with the distribution of carbon dioxide column concentration observed by satellite ( Figure 6 a), both sets of tests can simulate the distribution of carbon dioxide column concentration, but the results of the optimization test are closer to the observed values ​​overall ( Figure 6 d), while the control experiment overestimated CO2 concentrations in most regions ( Figure 6 c), especially in the Yangtze River Delta and southeastern Tibet. Compared with the CarbonTracker reanalysis data ( Figure 6 b), the present invention provides a more detailed distribution of carbon dioxide concentration.

[0089] In summary, the present invention effectively improves the model-calculated results of 2m air temperature, photosynthetically active radiation, and net ecosystem exchange, and significantly enhances the ability to estimate atmospheric carbon dioxide concentration in China's land areas in all aspects.

Claims

1. A method for estimating high-resolution atmospheric carbon dioxide concentration over the land area of ​​China, comprising the following steps: (1) Use satellite data to calculate vegetation index, surface moisture index, and vegetation phenology index, and distribute the calculated data together with land type data to the numerical model grid to generate the vegetation data file required for the numerical model operation; (2) The CAMS global greenhouse gas concentration dataset from the European Centre for Medium-Range Weather Forecasts (ECMWF) was used as the original CO2 background field, and the fifth-generation atmospheric reanalysis dataset from the European Centre for Medium-Range Weather Forecasts (ECMWF) was used as the original meteorological background field. The data of the original CO2 background field and the original meteorological background field were interpolated onto the numerical model grid using the inverse distance weighted method to generate the CO2 background field and meteorological background field files. (3) The multi-scale emission inventory from Tsinghua University was used as the anthropogenic emission inventory data. The monthly resolution anthropogenic emission inventory data was first converted to hourly resolution through linear interpolation. The anthropogenic emission inventory data was then interpolated onto the numerical model grid using the inverse distance weighting method to generate anthropogenic carbon dioxide emission data files. (4) By improving the calculation formulas for photosynthesis and respiration of terrestrial ecosystems and improving the three-dimensional temperature field calculated by numerical models, an optimized atmospheric carbon dioxide simulation system is constructed; among which: ① In the WRF-Chem model, the calculation formula for terrestrial ecosystem photosynthesis is: ABOUT new =0.77×PAR model +7.36 Where: GEE is the photosynthesis efficiency of terrestrial ecosystems, unit: μmol m -2 s -1 ;λ is the maximum light energy utilization rate, unit is μmolCO2 m -2 s -1 / μmol PAR mm -2 s -1 PAR stands for photosynthetically active radiation; PAR0 is the photosynthetically active radiation when photosynthesis reaches half-saturation, in μmol mm -2 s -1 ;T scale is the temperature impact index; P scale represents vegetation phenology index; W scale1 LSWI is the photosynthesis water stress index; max The maximum value of the multi-year surface moisture index; PAR new is the corrected light and effective radiation, unit is μmolm -2 s -1 PAR model Light and effective radiation calculated for the original numerical model; ②The calculation formula for terrestrial ecosystem respiration is: R=β+α1×T′+α2×T′ 2 +γ×EVI+k1×W scale +k2×W scale2 ×T′+k3×W scale2 ×T′ 2 When T a ≥T c Time: T′=T a When T a <T c Time: T′=T c -T m ×(T c -T a ) Where: T' is the improved temperature, unit: ℃; W scale2 is the respiratory water stress index; β, α1, α2, γ, k1, k2, and k3 are calculation parameters applicable to different underlying surface vegetation types in China's land area obtained by fitting the productivity observation data of different vegetation types in China over many years; LSWI min is the minimum value of the multi-year surface moisture index; T a T is the atmospheric temperature at 2 m, in °C; c is the low temperature threshold, unit is ℃; T m The atmospheric temperature is lower than T c The empirical coefficient when ③ Improve the three-dimensional temperature field calculated by the numerical model: The model-estimated 2-meter temperature was improved by assimilating the observed 2-meter temperature data at the stations using the observation nudging method. The model-estimated upper-level temperature was improved by bringing the model-estimated upper-level temperature closer to the upper-level temperature in the European Centre for Medium-Range Weather Forecasts (ERA5) reanalysis data using the gridded nudging method. Ultimately, the three-dimensional temperature field calculated by the improved numerical model was obtained. (5) The vegetation data files, CO2 background field and meteorological background field files, and anthropogenic CO2 emission data files are input into the optimized atmospheric CO2 simulation system and run, thus obtaining the high-resolution atmospheric CO2 concentration estimation results for the land area of ​​China.

2. The method for estimating high-resolution atmospheric carbon dioxide concentration in China's land area according to claim 1, characterized in that: In step (1), the vegetation index, surface moisture index, and vegetation phenology index are calculated using MODIS satellite data according to the following formula: Where: EVI represents the enhanced vegetation index, ranging from -1 to 1; ρ nir represents the reflectivity in the near-infrared band; ρ red Represents the reflectivity of red light band; ρ blue represents the reflectance of the blue light band; G is the gain factor, which is 2.5; C1 and C2 are the correction parameters for atmospheric correction of red light and blue light, which are 6 and 7.5 respectively; L is the soil correction parameter, which is 1; LSWI represents the surface moisture index, which ranges from -1 to 1; ρ swir Indicates the reflectivity of the short-wave infrared band; P scale Represents the vegetation phenology index.

3. The method for estimating high-resolution atmospheric carbon dioxide concentration in China's land area according to claim 1, characterized in that: The numerical model grid in step (1) refers to the Lambert projection, with a resolution of 12×12 km, a grid number of 354×415, and a center point of 36.461°N, 102.53°E.

4. The method for estimating high-resolution atmospheric carbon dioxide concentration in China's land area according to claim 1, characterized in that: The value ranges of λ and PAR0 in step ① are shown in Table 1: Table 1 Photosynthesis calculation parameters 5. The method for estimating high-resolution atmospheric carbon dioxide concentration in China's land area according to claim 1, characterized in that: The value ranges of β, α1, α2, γ, k1, k2, and k3 in step ② are shown in Table 2: Table 2 Respiration calculation parameters

Images