CN117010560A - Remote sensing carbon dioxide concentration downscaling method based on extreme random tree - Google Patents
Remote sensing carbon dioxide concentration downscaling method based on extreme random tree Download PDFInfo
- Publication number
- CN117010560A CN117010560A CN202310974262.7A CN202310974262A CN117010560A CN 117010560 A CN117010560 A CN 117010560A CN 202310974262 A CN202310974262 A CN 202310974262A CN 117010560 A CN117010560 A CN 117010560A
- Authority
- CN
- China
- Prior art keywords
- data
- carbon dioxide
- remote sensing
- dioxide concentration
- variable data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- CURLTUGMZLYLDI-UHFFFAOYSA-N Carbon dioxide Chemical compound O=C=O CURLTUGMZLYLDI-UHFFFAOYSA-N 0.000 title claims abstract description 160
- 229910002092 carbon dioxide Inorganic materials 0.000 title claims abstract description 80
- 239000001569 carbon dioxide Substances 0.000 title claims abstract description 79
- 238000000034 method Methods 0.000 title claims abstract description 26
- 230000007613 environmental effect Effects 0.000 claims abstract description 21
- 238000001556 precipitation Methods 0.000 claims description 7
- 238000001308 synthesis method Methods 0.000 claims description 3
- 230000015572 biosynthetic process Effects 0.000 claims description 2
- 238000003786 synthesis reaction Methods 0.000 claims description 2
- 238000009826 distribution Methods 0.000 abstract description 12
- 230000009467 reduction Effects 0.000 abstract description 6
- 238000010276 construction Methods 0.000 abstract description 2
- GVVPGTZRZFNKDS-JXMROGBWSA-N geranyl diphosphate Chemical compound CC(C)=CCC\C(C)=C\CO[P@](O)(=O)OP(O)(O)=O GVVPGTZRZFNKDS-JXMROGBWSA-N 0.000 description 17
- 239000005431 greenhouse gas Substances 0.000 description 9
- 230000000694 effects Effects 0.000 description 8
- 230000008859 change Effects 0.000 description 7
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 6
- 229910052799 carbon Inorganic materials 0.000 description 6
- 238000003066 decision tree Methods 0.000 description 6
- 238000012544 monitoring process Methods 0.000 description 6
- 238000012549 training Methods 0.000 description 6
- 238000012795 verification Methods 0.000 description 6
- 239000002803 fossil fuel Substances 0.000 description 4
- 230000005855 radiation Effects 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 3
- 239000007789 gas Substances 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000000638 solvent extraction Methods 0.000 description 3
- 241000282414 Homo sapiens Species 0.000 description 2
- 238000010521 absorption reaction Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 239000000443 aerosol Substances 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008033 biological extinction Effects 0.000 description 1
- 238000002485 combustion reaction Methods 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 238000013524 data verification Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000007717 exclusion Effects 0.000 description 1
- 239000003337 fertilizer Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000002427 irreversible effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000002844 melting Methods 0.000 description 1
- 230000008018 melting Effects 0.000 description 1
- 230000000813 microbial effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 230000029058 respiratory gaseous exchange Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000002689 soil Substances 0.000 description 1
- 241000894007 species Species 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000005436 troposphere Substances 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 238000010792 warming Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Chemical compound O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Abstract
The invention aims to provide a remote sensing carbon dioxide concentration downscaling method based on an extreme random tree to obtain carbon dioxide concentration remote sensing data; acquiring environment variable data, and unifying the environment variable data to the space-time resolution of the carbon dioxide concentration remote sensing data; the environment variable data consists of vegetation factors, meteorological factors and geographic factors; taking the carbon dioxide concentration remote sensing data as input, taking environment variable data after unifying space-time resolution as independent variables, and establishing a relation model of the two by using an extreme random tree model; and (3) reducing the spatial resolution of the environmental variable data by a scale, inputting the environmental variable data into the relation model, and predicting the concentration of carbon dioxide. The method improves the spatial resolution of the carbon dioxide remote sensing data, can better analyze the spatial-temporal distribution of the atmospheric carbon dioxide concentration, and has important significance for supporting government emission reduction work, formulating a proper emission reduction scheme and establishing national economic construction.
Description
Technical Field
The invention belongs to the technical field of remote sensing, and relates to a carbon dioxide concentration downscaling method based on an Extreme Random Tree (ERT).
Background
Carbon dioxide (CO) in the atmosphere 2 ) Is the most threatening greenhouse gas, and is the second largest greenhouse gas next to water vapor, accounting for about 0.04% of the total volume of the atmosphere. Since the third industrial revolution, the carbon dioxide concentration in the atmosphere has increased by 40% and is now higher than at any time in the past 80 ten thousand years. Atmospheric carbon dioxide has a strong greenhouse effect, a high content and a long residence time, and is therefore considered as the primary greenhouse gas for global warming, and the contribution rate to the greenhouse effect is as high as 65%. From NOAA observations, the global near-surface air temperature has risen on average about 0.74 ℃ for the 20 th century. If the temperature is continuously increased, a series of irreversible natural disasters such as accelerated melting of two-pole glaciers, sea level rising, desertification in subtropical areas, frequent extreme weather, species extinction and the like are caused, and finally the living environment of human beings is threatened. In this situation, the paris climate change in 2015 established a national autonomous contribution INDC (Intended Nationally Determined Contributions) mechanism, requiring contractors to set forth emissions reduction targets for the climate change in a "bottom-up" manner with reference to the national conditions of each country.
Currently, there are less than 300 ground-based observation sites for global monitoring of greenhouse gases, and the regional distribution is very uneven, mostly in developed countries and densely populated areas. While the number of observation sites is still expanding, their limited three-dimensional space is representative, leading to a great problem in quantitatively understanding the source and sink distribution of atmospheric chamber gases. The satellite observation can realize global observation on higher spatial resolution, and provides important scientific observation data for carbon monitoring research, global carbon circulation, climate change and greenhouse gas emission reduction. The satellite remote sensing data can obtain global continuous spatial distribution and change of greenhouse gases, has the advantages of stability, long time sequence, wide space area and three-dimensional space monitoring, can make up the deficiency of foundation sites, and is beneficial to improving the understanding of carbon circulation and climate change. The European Union, japan, united states and China have emitted a CO-bearing sequence 2 ENVISAT, GOSAT/GOSAT-2 and OCO-2 with concentration observation capability and a satellite platform special for TanSat gas observation. Remote sensing monitoring data based on the satellite platform is obtainedThe stable high space-time resolution and long time sequence observation data in the global scope has become the mainstream technical means for monitoring global greenhouse gas variation. In addition, some countries have transmitted satellites for observing carbon dioxide and still perform well.
However, current satellites are often affected by confined areas. For example, OCO-2 and tanSat have observation band widths of 10.6 km and 20 km, respectively, and thus the obtained observations have a large number of spatial gaps. Furthermore, cloud and aerosols make it difficult for these satellite instruments to retrieve carbon dioxide. Thus, satellite-derived carbon dioxide products, although observed over a wider range than ground stations, still present a number of gaps that prevent subsequent analysis and application. Therefore, how to obtain high spatial resolution satellite carbon dioxide data is of great importance. XCO of GOSAT is reconstructed by Zeng et al by modeling space-time correlation structures 2 Data, the predicted results were found to have better correlation and less error than the ground measurements. Considering that different satellite platforms cover different areas, it is intuitive to fuse the multi-sensor estimates to fill the carbon dioxide gap. For example, jing et al CO based on GOSAT and SCIAMACHY 2 Uncertainty of estimation, a weighted fusion strategy was developed to generate XCO with 1 ° spatial resolution 2 Concentration. Likewise, wang et al determined the fusion criteria by considering the multi-platform imaging time differences and obtained a CO with a spatial resolution of 0.5 2 The product is obtained.
In recent years, there has been an increasing search for atmospheric CO observed by analog satellites 2 Reconstruction and modeling of regional or global atmospheric CO from relationships between concentration and various environmental data 2 Concentration. For example, regression models are used on five continents of atmospheric CO 2 The spatial distribution of concentration was modeled, including the CO observed from GOSAT data 2 Data, temperature, vegetation coverage (FVC), and productivity in MODIS products. The air atmosphere CO above the illion in the 2015 growing season (4 months to 9 months) was evaluated using an artificial neural network (Ann) 2 Spatial distribution of concentration, modeling data including XCO2 data extracted from OCO-2 data toAnd environmental variables such as normalized vegetation index (NDVI), net Primary Productivity (NPP), surface temperature (LST), leaf Area Index (LAI), air temperature, wind speed, and wind direction. In CO 2 In the simulation, the air temperature was found to be more relevant than the surface temperature. Other studies estimate spatial atmospheric CO using MODIS products and GOSAT data based on LST, vegetation Index (VIs), LAI/photosynthetically active radiation (standard rod count), and total primary productivity (GPP) predictors 2 Concentration. The results indicate that the surface parameters can be used to simulate CO in the atmosphere 2 Concentration.
However, while ground stations may provide accurate carbon dioxide measurements, the sparse distribution of global ground monitoring network sites is insufficient to derive large-scale spatiotemporal distributions of atmospheric carbon dioxide.
Disclosure of Invention
The invention aims to provide a method for reducing the atmospheric carbon dioxide concentration scale based on a ERT (Extremely Randomized Trees) model.
In order to achieve the above purpose, the technical scheme adopted by the invention is as follows:
the remote sensing carbon dioxide concentration downscaling method based on the extreme random tree comprises the following steps:
acquiring remote sensing data of carbon dioxide concentration;
acquiring environment variable data, and unifying the environment variable data to the space-time resolution of the carbon dioxide concentration remote sensing data; the environment variable data consists of vegetation factors, meteorological factors and geographic factors;
taking the carbon dioxide concentration remote sensing data as input, taking environment variable data after unifying space-time resolution as independent variables, and establishing a relation model of the two by using an extreme random tree model;
and (3) reducing the spatial resolution of the environmental variable data by a scale, inputting the environmental variable data into the relation model, and predicting the concentration of carbon dioxide.
As a preferred embodiment, the remote sensing data of the carbon dioxide concentration is derived from Mapping-XCO 2 Product data.
As a preferred embodiment, in the environmental variable data, the vegetation factor is selected from the group consisting of normalized vegetation index, leaf area index, enhanced vegetation index, total primary productivity; the meteorological factors are selected from vapor emission, precipitation, relative humidity and wind speed; the geographic factor selects elevation, surface temperature.
As a preferred embodiment, the normalized vegetation index, leaf area index, enhanced vegetation index, total primary productivity, vapor emission, relative humidity data are derived from MODIS; precipitation, wind speed and relative humidity data are derived from ERA5; elevation data is derived from SRTM.
As a preferred embodiment, the vegetation factor is normalized vegetation index, total primary productivity; the meteorological factors are vapor emission, relative humidity and wind speed; the geographic factor is elevation, surface temperature.
As a preferred embodiment, the environmental variable data is unified to a month scale time resolution, a 1 ° spatial resolution; after the model is built, inputting environment variable data of which the scale is reduced to 0.1 degree of spatial resolution.
As a preferred embodiment, the vegetation factor is normalized vegetation index, total primary productivity; the meteorological factors are vapor emission, relative humidity and wind speed; the geographic factors are elevation and surface temperature;
calculating month data by the normalized vegetation index through a maximum synthesis method;
the total primary productivity synthesizes month data through the accumulated value;
the evapotranspiration synthesizes month data by calculating a mean value;
the relative humidity is obtained based on the dew point temperature and the surface temperature data of ERA5 in a calculating mode;
the wind speed is obtained based on the vertical wind speed and horizontal wind speed data vector synthesis of ERA 5.
As a preferred embodiment, the method further comprises verifying the accuracy of the prediction result of the relational model using the atmospheric background station data provided by TCCON and WDCGG.
As a preferred embodiment, the decision coefficient R is used 2 And root mean square errorThe RMSE index examines the prediction result accuracy.
ERT model is a new tree-based machine learning method that shows better robustness and regression accuracy in many studies. The random forest adopts a random substitution mode to obtain a training set of each decision tree, so that repeated samples exist in the training set, the full utilization of all the samples cannot be ensured, and the similarity possibly exists among the decision trees. Based on the above considerations, geurns et al (2006) propose an extremely random tree ERT model. In the ERT model, each decision tree is obtained based on the whole data set training, so that the utilization rate of training samples is ensured, and the final prediction Bias (Bias) can be reduced to a certain extent; in order to ensure the structural difference between each decision tree, the ERT model introduces greater randomness in node partitioning: the division threshold value of each feature is randomly selected from the sub-data set, and the feature with the best division effect is selected as the optimal division attribute according to the designated threshold value. After comparing the applicability and operability of the various methods, the present invention selects ERT models to reconstruct XCO 2 Distribution. ERT models can take into account both spatial variability and atmospheric CO 2 Nonlinear relationship between concentration and environmental factors.
The invention has the following beneficial effects:
(1) The invention utilizes the method of the extreme random tree to downscale the carbon dioxide concentration data acquired by the satellite, and acquires the CO in the China region of 2015-2019 after the algorithm performance is fully evaluated by the measured data 2 Concentration space-time distribution. In the invention, the input parameters are screened, and the high space-time resolution CO is established on the basis of less input parameters and easy acquisition 2 The concentration is distributed in time and space, and the prediction result is verified by site measured data, so that the method has higher prediction precision.
(2) The method improves the spatial resolution of the carbon dioxide remote sensing data, can better analyze the spatial-temporal distribution of the atmospheric carbon dioxide concentration, and has important significance for supporting government emission reduction work, formulating a proper emission reduction scheme and establishing national economic construction. In addition, the principles, processes and results of the method can provide an important basis for analyzing the temporal and spatial variation of carbon dioxide concentration and driving factors thereof.
It should be understood that all combinations of the foregoing concepts, as well as additional concepts described in more detail below, may be considered a part of the inventive subject matter of the present disclosure as long as such concepts are not mutually inconsistent. In addition, all combinations of claimed subject matter are considered part of the disclosed inventive subject matter.
The foregoing and other aspects, embodiments, and features of the present teachings will be more fully understood from the following description, taken together with the accompanying drawings. Other additional aspects of the invention, such as features and/or advantages of the exemplary embodiments, will be apparent from the description which follows, or may be learned by practice of the embodiments according to the teachings of the invention.
Drawings
Fig. 1 is a flow chart of the method of the present invention.
Figure 2 is a graph of the characteristic importance of the present invention for each environmental factor during different seasons.
Fig. 3 is a validation of the downscaling results of the present invention with TCCON and WDCGG sites in 2015-2019.
FIG. 4 shows the decision coefficient R of the site verification after the original data and the downscaling of the present invention 2 And (5) comparing.
Fig. 5 is a graph of CO2 concentration trend between 2015-2019 for downscaled data and site data according to the present invention.
Detailed Description
For a better understanding of the technical content of the present invention, specific examples are set forth below, along with the accompanying drawings.
Aspects of the invention are described in this disclosure with reference to the drawings, in which are shown a number of illustrative embodiments. The embodiments of the present disclosure are not necessarily intended to include all aspects of the invention. It should be understood that the various concepts and embodiments described above, as well as those described in more detail below, may be implemented in any of a number of ways, and that the concepts and embodiments disclosed herein are not limited to any implementation. Additionally, some aspects of the disclosure may be used alone or in any suitable combination with other aspects of the disclosure.
The specific data sources in the examples are shown in table 1 below.
TABLE 1 data sources
Remote sensing data of carbon dioxide concentration is from Mapping-XCO 2 And (5) a product. Mapping-XCO 2 XCO for integrating GOSAT and OCO-2 satellite observation inversion of products 2 Global land XCO for data generation from 4 months in 2009 to 12 months in 2020 2 A spatio-temporal continuous data set. Mapping-XCO 2 The data set is aimed at different satellite observation times and priori CO 2 Deviation adjustment is carried out on the profile, the space-time resolution and the like, different satellite data are normalized by taking the Carbon Tracker model profile as a reference, and XCO with uniform space-time scale is generated 2 Data; then XCO is performed on different areas based on a space-time statistics method 2 Carrying out space-time correlation analysis and modeling on the data to generate XCO of 1-degree grid central point 2 Data.
Example 1
In this embodiment, a 2015-2019 chinese area is taken as an example, and the technical scheme of the present invention is further described.
Step 1, mapping-XCO with spatial resolution of 1 DEG and temporal resolution of Monthly is adopted 2 Product data. The environmental variables participating in model training are selected in groups according to vegetation factors, meteorological factors, geographic factors and emission factors. Vegetation factors include normalized vegetation index (NDVI), total primary productivity (GPP), enhanced Vegetation Index (EVI), leaf Area Index (LAI), meteorological factors include Evapotranspiration (ET), precipitation (TP), relative Humidity (RH), wind speed (U, V), geographic factors include elevation (DEM), surface temperature (LST), and emissions factors include fossil fuel carbon dioxide emissions product data (ODIAC).
And 2, for the processing of the environment variable data, NDVI calculates month data through a maximum synthesis method, GPP synthesizes month data through calculating an accumulated value, and ET synthesizes month data through calculating a mean value, so that the time resolution is unified. The wind speed data is vector addition synthesized according to U and V. RH is expressed by the formulaAnd (5) performing calculation. Wherein T is a Representing the surface temperature, T d Representing the dew point temperature.
In addition, the environment variable data with uniform time resolution are resampled to the spatial resolution of 0.1 DEG and 1 DEG by the nearest neighbor method, and cut according to the scope of the Chinese research area.
And 3, constructing and experimental a carbon dioxide concentration downscaling model based on an extreme random tree. According to experimental requirements, building a Mapping-XCO 2 The concentration product is used as input data, a plurality of environmental factors are used as independent variables, a relation between the two is established, and a high-spatial-resolution carbon dioxide concentration downscaling model ERT-CO is predicted 2 And (5) a model. In ERT, each decision tree is trained based on the whole data set, so that the utilization rate of training samples is guaranteed, and the final prediction deviation can be reduced to a certain extent. To ensure structural differences between each decision tree, ERT introduces greater randomness in node partitioning: the division threshold value of each feature is randomly selected from the sub-data set, and the feature with the best division effect is selected as the optimal division attribute according to the designated threshold value.
In this example, the relevant environment variables were selected and finally divided into 6 combinations, as shown in Table 2 below.
TABLE 2 environmental variable combination partitioning
| Group of | Environmental variable |
| 1 | NDVI、LAI、EVI、GPP |
| 2 | NDVI、LAI、EVI、GPP、DEM、LST |
| 3 | NDVI、LAI、EVI、GPP、DEM、LST、RH、WS、TP、ET |
| 4 | NDVI、LAI、EVI、GPP、DEM、LST、RH、WS、TP、ET、ODIAC |
| 5 | NDVI、GPP、DEM、LST、RH、WS、TP、ET、ODIAC |
| 6 | NDVI、GPP、DEM、LST、RH、WS、TP、ET、ODIAC |
And (3) performing a test by using the constructed ERT algorithm, and judging the combination effect according to the precision of the station. For the verification of the ground station, the atmospheric background station data provided by the TCCON and the WDCGG are selected. The global carbon column total observation network (TCCON) is a ground network for recording solar spectrum information in the near infrared region. Reliable CO calculation using radiometric observation 2 、CH 4 And H 2 Total carbon column of O. The world greenhouse gas data center (WDCGG) is one data center under the GAW program. It is used for collecting, archiving and providing information about greenhouse gases (e.g. CO 2 Etc.) and related gases.
The invention selects four sites of WLG, LLN, fertilizer combination and Shanghai for data verification, which are respectively positioned in Qinghai, taiwan, anhui and Hebei in China.
The invention uses the decision coefficient R 2 And the index of the Root Mean Square Error (RMSE) to comprehensively investigate the accuracy of the downscaling result.
Wherein N is the number of site samples, f i As predicted value, y i To be a true value of the value,is the average value.
The accuracy verification results are shown in table 3 and fig. 3 (sixth group).
Table 3 results of accuracy verification of combinations of environmental variables (R 2 )
As known from the literature, vegetation factors affect CO 2 Important factors of concentration. Research has shown that carbon dioxide induced climate change can have an impact on the earth's vegetation coverage. Significant regional and global climate feedback may occur, whether due to climate change or changes in vegetation patterns caused by human activity. Thus, a combination of vegetation was first selected for testing, including NDVI, LAI, EVI, GPP, i.e., group 1.
In addition, elevation data is used as a geographic factor in connection with CO 2 The concentration profile has a certain correlation. Surface temperature is achieved by accelerating microbial decomposition and respiration in soil and land ecologyPlant development in the system indirectly affects CO in the atmosphere 2 The concentration, therefore, is chosen for this part of the factors, including NDVI, LAI, EVI, GPP, DEM, LST, i.e., group 2.
Meteorological conditions can directly or indirectly influence the CO in the atmosphere 2 Concentration. Relative humidity and atmospheric CO 2 There is a link between concentrations when atmospheric CO 2 As the concentration increases, the skin layer is expected to warm up, resulting in the troposphere becoming more moist. Precipitation is also indirectly subjected to CO to a certain extent 2 Concentration effects. Slightly increased global wind speed affects ocean CO 2 Absorption of concentration, thereby affecting CO in the atmosphere 2 Concentration. CO 2 When the concentration is increased, the solar short wave radiation can be absorbed, and the long wave radiation emitted by the heating surface is CO 2 Absorption, resulting in net ground radiation and increased ET. The portion of content is selected to include NDVI, LAI, EVI, GPP, DEM, LST, RH, WS, TP, ET, i.e., group 3.
Considering that artificially discharged fossil fuel has a certain effect on concentration, fossil fuel combustion data ODIAC is added. ODIAC (artificial CO) 2 Open source data list) is a global high resolution emissions data product for fossil fuel carbon dioxide emissions, this part comprising NDVI, LAI, EVI, GPP, DEM, LST, RH, WS, TP, ET, ODIAC, group 4.
In the previous experiments, great similarity exists between factors of vegetation factors, and after the factors are adjusted, experiments show that after partial vegetation factors, namely LAI and EVI are removed, the data size is reduced, the verification accuracy is not reduced, even improved to a certain extent, and the part comprises NDVI, GPP, DEM, LST, RH, WS, TP, ET, ODIAC, namely group 5.
Because the addition of ODIAC data does not improve accuracy, the consideration is because of atmospheric CO 2 The concentration distribution values are relatively uniform, while the data values of ODIAC are unevenly distributed and greatly float, which interferes with experimental accuracy, so that the exclusion adjustment is made, which includes NDVI, GPP, DEM, LST, RH, WS, TP, ET, i.e., group 6.
In addition, other combinations than six were also made in this experiment, but the results were lower than those of the group6, finally, determining group 6 as ERT-CO 2 The most desirable environmental factors in the model are vegetation factors, meteorological factors and geographic factors. Vegetation factors include normalized vegetation index (NDVI), total primary productivity (GPP), meteorological factors include Evapotranspiration (ET), precipitation (TP), relative Humidity (RH), wind speed (U, V), geographic factors include elevation (DEM), surface temperature (LST).
Thus, the general structure of the model can be described as shown in the equation.
CO 2j ＝f (xj,yj) (NDVI j ,GPP j ,ET j ,LST j ,TP j ,WS j ,RH j ,DEM j )
Wherein CO 2j Representative is the predicted month average atmospheric carbon dioxide concentration for grid cell j, which is the average of the carbon dioxide concentration for grid cell in (x j ,y j ) The center coordinates of the location. f (f) (xj,yj) Representing atmospheric CO 2 A specific location estimation function of concentration.
ERT-CO is adopted 2 Accuracy R of model 2 And RMSE of 0.77 and 0.863ppm, respectively. The verification index shows that the model has better anti-interference performance, and the model has better generalization capability. The 0.1℃spatial resolution environmental factor data is then brought into ERT-CO 2 The scale is reduced to obtain XCO of 0.1 degree in China area 2015-2019 2 Concentration data.
FIG. 3 provides a scatter plot of site data versus downscaled data, embodying the reliability of downscaled data. In FIG. 4, by comparing the original Mapping-XCO 2 Determination coefficients R of product data at four site positions 2 And the decision coefficients R of the downscaled data at four sites 2 It can be found that the determination coefficient R of each position after the downscaling 2 Are all higher than the original data Mapping-XCO 2 Determination coefficient R of product data at site 2 . This indicates downscaled XCO 2 The concentration is more nearly the true value.
While the invention has been described with reference to preferred embodiments, it is not intended to be limiting. Those skilled in the art will appreciate that various modifications and adaptations can be made without departing from the spirit and scope of the present invention.
Claims (9)
1. The remote sensing carbon dioxide concentration downscaling method based on the extreme random tree is characterized by comprising the following steps of:
acquiring remote sensing data of carbon dioxide concentration;
acquiring environment variable data, and unifying the environment variable data to the space-time resolution of the carbon dioxide concentration remote sensing data; the environment variable data consists of vegetation factors, meteorological factors and geographic factors;
taking the carbon dioxide concentration remote sensing data as input, taking environment variable data after unifying space-time resolution as independent variables, and establishing a relation model of the two by using an extreme random tree model;
and (3) reducing the spatial resolution of the environmental variable data by a scale, inputting the environmental variable data into the relation model, and predicting the concentration of carbon dioxide.
2. The method of claim 1, wherein the remote sensing data of carbon dioxide concentration is derived from Mapping-XCO 2 Product data.
3. The method of claim 1, wherein in the environmental variable data, the vegetation factor is selected from the group consisting of normalized vegetation index, leaf area index, enhanced vegetation index, total primary productivity; the meteorological factors are selected from vapor emission, precipitation, relative humidity and wind speed; the geographic factor selects elevation, surface temperature.
4. The method of claim 3, wherein the normalized vegetation index, leaf area index, enhanced vegetation index, total primary productivity, evapotranspiration, relative humidity data are derived from MODIS; precipitation, wind speed and relative humidity data are derived from ERA5; elevation data is derived from SRTM.
5. The method of claim 3, wherein the vegetation factor is a normalized vegetation index, total primary productivity; the meteorological factors are vapor emission, relative humidity and wind speed; the geographic factor is elevation, surface temperature.
6. The method of claim 2, wherein the environmental variable data is unified to a month scale time resolution, a 1 ° spatial resolution; after the model is built, inputting environment variable data of which the scale is reduced to 0.1 degree of spatial resolution.
7. The method of claim 4 or 6, wherein the vegetation factor is normalized vegetation index, total primary productivity; the meteorological factors are vapor emission, relative humidity and wind speed; the geographic factors are elevation and surface temperature;
calculating month data by the normalized vegetation index through a maximum synthesis method;
the total primary productivity synthesizes month data through the accumulated value;
the evapotranspiration synthesizes month data by calculating a mean value;
the relative humidity is obtained based on the dew point temperature and the surface temperature data of ERA5 in a calculating mode;
the wind speed is obtained based on the vertical wind speed and horizontal wind speed data vector synthesis of ERA 5.
8. The method of claim 1, further comprising verifying accuracy of a predicted outcome of the relational model using atmospheric background station data provided by TCCON and WDCGG.
9. The method of claim 8, wherein the decision coefficient R and the root mean square error are usedRMSEAnd (5) inspecting the precision of the prediction result by the index.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310974262.7A CN117010560A (en) | 2023-08-04 | 2023-08-04 | Remote sensing carbon dioxide concentration downscaling method based on extreme random tree |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310974262.7A CN117010560A (en) | 2023-08-04 | 2023-08-04 | Remote sensing carbon dioxide concentration downscaling method based on extreme random tree |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117010560A true CN117010560A (en) | 2023-11-07 |
Family
ID=88570581
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310974262.7A Pending CN117010560A (en) | 2023-08-04 | 2023-08-04 | Remote sensing carbon dioxide concentration downscaling method based on extreme random tree |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117010560A (en) |
-
2023
- 2023-08-04 CN CN202310974262.7A patent/CN117010560A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106372730B (en) | Utilize the vegetation net primary productivity remote sensing estimation method of machine learning | |
| Chen et al. | Effects of foliage clumping on the estimation of global terrestrial gross primary productivity | |
| Li et al. | Accuracy assessment of global historical cropland datasets based on regional reconstructed historical data—A case study in Northeast China | |
| CN105403664A (en) | WRF-CHEM-based large point pollution source atmosphere environment influence evaluating method | |
| Hudiburg et al. | Evaluation and improvement of the Community Land Model (CLM4) in Oregon forests | |
| Yang et al. | Sensitivity of surface air temperature change to land use/cover types in China | |
| Jiménez et al. | Exploring the merging of the global land evaporation WACMOS-ET products based on local tower measurements | |
| CN111487216A (en) | Carbon dioxide flux inversion method and system | |
| Zhou et al. | Assessing the impact of urban sprawl on net primary productivity of terrestrial ecosystems using a process-based model—a case study in Nanjing, China | |
| CN114460013B (en) | Coastal wetland vegetation overground biomass GAN model self-learning remote sensing inversion method | |
| Nguyen et al. | Downscaling over Vietnam using the stretched-grid CCAM: verification of the mean and interannual variability of rainfall | |
| Jia et al. | Uncertainty in simulating regional gross primary productivity from satellite-based models over northern China grassland | |
| Bolun et al. | Estimating rice paddy areas in China using multi-temporal cloud-free normalized difference vegetation index (NDVI) imagery based on change detection | |
| Liu et al. | Estimating surface carbon fluxes based on a local ensemble transform Kalman filter with a short assimilation window and a long observation window: an observing system simulation experiment test in GEOS-Chem 10.1 | |
| CN116223395A (en) | Near-surface trace gas concentration inversion model and inversion method | |
| CN109657988B (en) | Tobacco leaf quality partitioning method based on HASM and Euclidean distance algorithm | |
| Wu et al. | On the origin of surface ozone episode in Shanghai over Yangtze River Delta during a prolonged heat wave | |
| CN111879915A (en) | High-resolution monthly soil salinity monitoring method and system for coastal wetland | |
| CN116312849A (en) | Full-coverage methane concentration measuring and calculating method based on random forest model and satellite remote sensing | |
| CN114974453B (en) | High-coverage reconstruction method for atmospheric carbon dioxide column concentration | |
| Koo et al. | Projection of red spruce (Picea rubens Sargent) habitat suitability and distribution in the Southern Appalachian Mountains, USA | |
| CN117010560A (en) | Remote sensing carbon dioxide concentration downscaling method based on extreme random tree | |
| CN113128871B (en) | Cooperative estimation method for larch distribution change and productivity under climate change condition | |
| AU2021100706A4 (en) | An Integrated Assessment Method of Climate Change Based on Multi-Source Carbon Data | |
| Maselli et al. | Modelling and analyzing the water and carbon dynamics of Mediterranean macchia by the use of ground and remote sensing data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |