Soil carbon estimation and dynamic monitoring method for seasonal flooded wetland

Abstract

The invention relates to the technical field of wetland ecosystem soil carbon sequestration capability, in particular to a seasonal waterlogged wetland soil carbon estimation and dynamic monitoring method, which comprises the following steps of: selecting a sampling zone and a quadrat according to an equal-elevation water level gradient and a vegetation type; an optimal remote sensing estimation carbon model is determined through remote sensing image interpretation and screening of characteristic factors such as image waveband reflectivity, vegetation indexes, principal component characteristics and single-waveband textures, and soil organic carbon spatial distribution characteristics are analyzed through remote sensing inversion and GIS interpolation; constructing a Wetland-DNDC wetland carbon dynamic simulation model based on a DNDC model to predict wetland long-term carbon dynamic change and greenhouse gas emission; the invention provides a method for simulating carbon reserves and carbon dynamic monitoring based on field investigation and coupling a 3S technology with a Wetland-DNDC model. Compared with a traditional field observation method, the method has obvious technical and cost advantages, and can realize the technical advantages of spatial analysis and long-time sequence monitoring. The research method is suitable for a wetland ecosystem with seasonal hydrological characteristic changes.

Description

Technical Field

[0001] The present invention relates to the technical field of soil carbon sequestration capacity of wetland ecosystems, and in particular to a method for estimating and dynamically monitoring soil organic carbon (SOC) in seasonally flooded wetlands. Background Art

[0002] Improving the carbon sequestration capacity of terrestrial ecosystems is an important measure to achieve the "dual carbon" goals, and it is also a direct ecological response to global climate change. Wetlands are one of the components of the terrestrial carbon pool in the global carbon cycle, and carbon sequestration is one of its important ecosystem service functions, which plays a key role in global carbon balance and climate change. At the same time, wetlands are also an important source of greenhouse gas methane emissions. Therefore, the dynamic changes between carbon accumulation and carbon emissions in wetland ecosystems determine the carbon source and carbon sink functions of wetlands. In recent years, human activities have caused the development and destruction of wetlands to reduce their carbon sequestration capacity, weaken their carbon "sink" function, and even show a trend of transforming from carbon sinks to carbon sources, exacerbating global climate change, which will inevitably affect wetland ecosystems. Therefore, the study of wetland carbon sequestration capacity has become one of the hot spots for wetland ecological protection and restoration.

[0003] For a long time, the research on the carbon sink function of wetland ecosystems has been very active, and a large amount of scientific data and conclusions have been obtained. However, wetland soil data is difficult to obtain. The traditional field survey method is time-consuming and labor-intensive, and even the research area is too large to be feasible; and the soil sampling survey data of a single time scale cannot reflect the evolution of the regional soil organic carbon pool. In addition, due to the complexity and diversity of wetland ecosystems, there is no recognized classification method so far. In addition, the difficulty of profile surveys and the uncertainty of areas, the parameters such as depth used in statistics vary greatly, resulting in different research methods for estimating wetland carbon reserves, and there are certain differences in the results, which makes it difficult to correctly assess the importance of wetlands in the global carbon cycle.

[0004] Since the 1990s, the "3S" technology has been used in the estimation of global or regional soil organic carbon pool reserves and the spatial distribution differences of organic carbon density. At the same time, the use of computer-aided information to establish models to estimate and predict SOC has become a dominant processing method, which helps to predict the long-term changes of SOC and provides a powerful tool for integrating a large amount of observational data, analyzing and predicting large-scale ecosystem processes. The combination of "3S" technology and field data can solve the problem of estimating soil carbon reserves from point to region. The high temporal and spatial resolution characteristics of remote sensing, the process model reflecting the dynamic changes of ecosystem carbon cycle, and the actual measured SOC are combined to ensure the accuracy of the total SOC estimation. The process-based model combined with GIS software provides a good platform for regional-scale soil organic carbon simulation and has gradually become a new research direction. However, in the study of wetland ecosystem carbon sequestration capacity, the method based on 3S technology coupled with biogeochemical models has not yet been explored. In addition, there are still few discussions on the measurement methods and technologies for wetland ecosystem carbon sequestration monitoring, which to a certain extent limits the in-depth development of wetland carbon sequestration research.

[0005] Therefore, a method for estimating and dynamically monitoring soil carbon in seasonally flooded wetlands was proposed. Summary of the invention

[0006] The purpose of the present invention is to provide a method for solving the above problems by combining point data obtained by sample plot survey with surface data of remote sensing images using 3S technology (GPS, GIS, RS), and at the same time introducing the DNDC biogeochemical model to construct a Wetland-DNDC model suitable for simulating the dynamic carbon of seasonally flooded wetland soils. By optimizing the coupling, a new method for estimating and dynamically monitoring wetland soil carbon is formed, which provides theoretical and technical innovations for integrating a large amount of observation data, analyzing and predicting the carbon fixation capacity of large-scale wetland ecosystems and studying the characteristics of long-term carbon dynamic changes.

[0007] The technical problem to be solved by the present invention is to screen the remote sensing characteristic factors of the optimal remote sensing estimation model of SOC and the model parameters that can accurately predict the spatiotemporal dynamic changes of SOC in seasonally flooded wetlands, develop a method for coupling 3S technology and biogeochemical models, and improve the accuracy of carbon estimation and simulation.

[0008] The present invention achieves the above-mentioned purpose through the following technical scheme, a method for estimating and dynamically monitoring soil carbon in seasonally flooded wetlands, comprising the following steps:

[0009] Select sample strips and plots according to the water level gradient and dominant vegetation community type, perform GPS positioning, set up stakes, investigate the wetland area, wetland type, and distribution characteristics in the sample strips and plots, and summarize the imaging rules and image characteristics (hue, brightness, shape, and structural texture) of different types of wetlands in remote sensing images;

[0010] Compare the remote sensing image features with the actual conditions recorded by the wetland remote sensing survey and interpretation signs one by one to form interpretation signs and establish a standard table for wetland remote sensing interpretation in the study area.

[0011] At the same time, in each sample strip, three small sample plots (1mx1m) were randomly selected for GPS positioning, the longitude and latitude of the sampling points in the sample plots were recorded, the location and range of the sampling points were determined, and investigations were carried out on vegetation types, soil physical and chemical properties, and influencing environmental factors.

[0012] The above-ground parts of plants in small sample plots (1mx1m) were collected by mowing at the same level as the ground. The underground parts of plants (root samples) in small sample plots (1mx1m) were collected by soil drilling. The sampling depth was 0-20cm. After the soil was washed away with water using a 0.5-mesh sand net bag, the clean root samples were placed in a paper envelope.

[0013] Five sampling points were set up in each research transect and arranged randomly to collect the topsoil (0-20 cm), and the geographical information of the sampling points was recorded.

[0014] Collect basic information such as meteorology (temperature), hydrology, and geology of the research site, extract and interpret the data, and produce relevant maps.

[0015] Extract remote sensing characteristic factors. Use remote sensing digital image processing software to calculate band reflectance, wetland vegetation index, texture characteristics and principal component characteristics. After the analysis is completed, import the image into ArcGIS software and use GIS spatial analysis to extract remote sensing parameter values.

[0016] Construct the optimal SOC remote sensing estimation model and remote sensing inversion. Using remote sensing characteristic parameters extracted from remote sensing images, analyze the correlation between each remote sensing parameter and the SOC content and carbon density of the sample point, apply linear, curvilinear and multivariate stepwise regression methods to establish a multivariate stepwise linear regression remote sensing estimation model for SOC content and carbon density, and verify the accuracy of the regression model. Select the optimal estimation model for SOC content and carbon density to calculate and invert the spatial distribution characteristics of SOC content and carbon storage.

[0017] The GIS Kriging spatial interpolation method was used to interpolate the sampling points and predict the spatial distribution characteristics of SOC content, carbon density and carbon storage in the study area.

[0018] The DNDC biogeochemical model developed by the Land, Ocean and Space Research Center of the University of New Hampshire was introduced for digestion and absorption, and the model parameters such as the meteorological data, vegetation physiological data and basic soil physical and chemical properties collected and measured were used to prepare the model parameter database. The model was repeatedly debugged and verified, the model parameters were screened, and sensitivity analysis was performed to construct a Wetland-DNDC model suitable for wetland soil carbon dynamics simulation, simulate the dynamic changes of wetland SOC, and predict the long-term changes in SOC and greenhouse gas emissions.

[0019] Compare and analyze the results of field sample survey, the optimal remote sensing carbon estimation model and remote sensing inversion, GIS spatial interpolation analysis and the results of Wetland-DNDC model simulation of soil carbon density and estimation of soil carbon storage, optimize the coupling of field survey, 3S technology and Wetland-DNDC model simulation technology, form a seasonal flooded wetland soil carbon estimation and dynamic monitoring method, evaluate the wetland soil carbon sink capacity, and predict the response of wetland soil carbon dynamics to future climate. Among them, the calculation formulas of experimentally measured soil organic carbon density (SOCD) and soil organic carbon storage (SOCS) are:

[0020] SOCD=SOC×BD×D×0.1

[0021] SOCS=A×SOCD

[0022] Where, SOCD: soil organic carbon density (t / hm 2 ); SOC: soil organic carbon content (g / kg); BD: soil bulk density (g / cm 3 ); D: soil thickness (cm); SOCS: soil organic carbon storage (t); A: study area (hm 2 ).

[0023] Remote sensing inversion, GIS interpolation analysis, and Wetland-DNDC model simulation of SOCD automatically present results after the model is run, and SOCS estimation is calculated based on the experimentally determined carbon storage formula.

[0024] The beneficial effects of the present invention are:

[0025] The present invention provides a method for estimating and dynamically monitoring soil carbon in seasonally flooded wetlands. The method is based on 3S (GPS, GIS, RS) technology coupled with the Wetland-DNDC biogeochemical model to estimate the SOC content, carbon density and carbon storage of seasonally flooded wetlands, analyze their spatial distribution characteristics, monitor long-term carbon dynamic changes and greenhouse gas emissions, and evaluate the carbon sink capacity of wetland soils. Compared with traditional field observation methods, it has obvious technical and cost advantages, and can achieve the technical advantages of spatial analysis and long-term series monitoring. This research method is suitable for wetland ecosystems with seasonal hydrological characteristic changes, and can estimate soil organic carbon storage at a regional scale and evaluate the soil carbon source-sink effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a schematic diagram of the method flow in the present invention;

[0027] Figure 2 It is a schematic diagram of sample strips, sample plots and sample quadrats;

[0028] Figure 3 It is the original Landsat 8OLT remote sensing image in the present invention;

[0029] Figure 4 The radiation calibration parameter settings and the spectrum curve after calibration in the present invention;

[0030] Figure 5 This is a schematic diagram of the atmospheric correction parameter settings in the present invention;

[0031] Figure 6 Spectral curves before and after atmospheric correction in the present invention: (a) before correction;

[0032] (b) after correction;

[0033] Figure 7 A schematic diagram of the study area boundary and image cropping parameter settings in the present invention;

[0034] Figure 8 This is the image map of the study area in the present invention;

[0035] Fig. 9 This is a schematic diagram of calculating the band1 parameter settings in the present invention;

[0036] Fig.10 It is a schematic diagram of the single-band reflectivity of b1 to b7 of the band reflection in the present invention;

[0037] Fig.11 This is a schematic diagram of setting the vegetation index parameters in the present invention;

[0038] Fig.12 It is a schematic diagram of vegetation index in the present invention;

[0039] Fig.13 It is a schematic diagram of the principal component analysis parameter settings and results in the present invention;

[0040] Fig.14 It is a characteristic diagram of the first three components of the principal component in the present invention;

[0041] Fig.15 A schematic diagram of setting parameters for texture feature analysis in the present invention;

[0042] Fig.16 are the mean (MEAN), entropy (ENT), correlation (COR) and variance (VAR) of the texture features of the first band b1 in the present invention;

[0043] Fig.17 The spatial distribution characteristics of soil organic carbon content, carbon density and carbon storage in seasonally flooded wetlands inverted by remote sensing in the present invention;

[0044] Fig.18 It is a spatial distribution characteristic map of soil organic carbon content, carbon density and carbon storage in seasonally flooded wetlands using the GIS Kriging interpolation method in the present invention;

[0045] Fig.19 It is a simulation diagram of the Wetland-DNDC model for predicting annual greenhouse gas emissions from seasonally flooded wetland soils in the present invention;

[0046] Fig. 20 This is a simulation diagram of the Wetland-DNDC model for the dynamic prediction of long-term changes in soil organic carbon content in seasonally flooded wetlands in the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] A method for estimating and dynamically monitoring soil carbon in seasonally flooded wetlands. The schematic diagram of the method is shown in Figure 1 , including the following:

[0049] Sample strip, plot, quadrat setting and sample collection ( Figure 2). In the study area, a relatively flat and open area with no traces of human activities was selected. Several sampling strips were appropriately selected from the shore to the center of the lake according to the terrain (equal elevation water level gradient), GPS positioning was performed, and stakes were fixed. Several sampling points were set up on each sampling strip according to the dominant vegetation community type, and several sample plots were arranged within the sampling point range. At the sampling point, a stainless steel soil sampler was used to obtain surface (0-20 cm) soil samples according to the five-point sampling method. After removing foreign matter from the surface of the soil, the soil samples were placed in self-sealing bags, and the soil samples were numbered, and their vegetation cover type and surrounding environmental factors were recorded. At the same time, the latitude and longitude information of the sampling points were recorded with GPS.

[0050] Sample pretreatment. After the collected soil samples are brought back to the laboratory, each soil sample is first divided into two parts. One soil sample is kept in its original state for determining the soil bulk density, and the other is naturally air-dried. After air-drying, roots, sand and other smaller debris are further removed. Then, it is ground and passed through a 100-mesh sieve before being sealed and packaged again for subsequent sample determination.

[0051] Experimental determination of various indicators. Soil organic carbon (SOC) content was determined using potassium dichromate oxidation-external heating method. Calculation formula:

[0052]

[0053] Where: 0.8000 is the concentration of potassium dichromate standard solution in mol / L; 5.0 is the volume of potassium dichromate standard solution in ml; V 0 represents the volume of ammonium ferrous sulfate standard solution consumed by the blank sample (ml); V represents the volume of ammonium ferrous sulfate standard solution consumed by the soil sample (ml); 0.003 represents the millimolar mass of 1 / 4 carbon atom (g); 1.10 represents the oxidation correction coefficient; and m represents the sample mass (g).

[0054] The soil bulk density is determined by the ring knife method. Take out a reserved soil sample and cut the original soil with a ring knife (i.e., a circular iron container) to fill the entire container. Then weigh it and dry it to a constant weight. After weighing, calculate the bulk density (unit: g / cm 3 ), the calculation formula of bulk density is:

[0055]

[0056] Where, ρ: soil bulk density (g / cm 3 ); V: ring knife volume (cm 3 ) ; m: weight of wet soil in the cutter ring (g); W: moisture content of soil in the cutter ring (%).

[0057] Calculate soil organic carbon density (SOCD) and soil organic carbon stock (SOCS). The calculation formula is:

[0058] SOCD=SOC×BD×D×0.1

[0059] SOCS=A×SOCD

[0060] Where, SOCD: soil organic carbon density (t / hm 2 ); SOC: soil organic carbon content (g / kg); BD: soil bulk density (g / cm 3 ); D: soil thickness (cm); SOCS: soil organic carbon storage (t); A: study area (hm 2 ).

[0061] Remote sensing characteristic factor extraction. Obtain Landsat 8 multispectral remote sensing images, perform remote sensing interpretation on them in ENVI 5.3, and then use GIS spatial analysis functions to extract remote sensing characteristic factors of sampling points, mainly including: image band reflectance, vegetation index, principal component characteristics, and single-band texture characteristics. The specific steps are as follows:

[0062] (1) Obtain remote sensing images. Select a Landsat 8OLT image with row and column number 122 / 40. The imaging time is synchronized with the field sampling time. The image covers 100% of the study area. The image is clear with a resolution of 30 m. There is no cloud cover in the target area. The data is good. The specific information of the original image is shown in Table 1 and Figure 3 shown.

[0063] Table 1 Image information

[0064]

[0065] (2) Remote sensing image preprocessing. Image preprocessing steps mainly include radiometric calibration, atmospheric correction and image cropping.

[0066] 1) Radiometric calibration. Open the Radiometric Calibration panel in ENVI 5.3, select the original image file, and set the parameters: first, select the Calibration Type, select the Radiance type, then click the Apply FLAASH Settings button, select the BIL mode for the output image storage order (Output Interleave), and set the output data type (Output Data Type) to Float. At this time, the radiometric data automatically adjusts the unit correction factor (Scale Factor) to 0.1. Finally, open the spectral curve of the image file after radiation calibration. The relevant settings and spectral curve are shown as follows: Figure 4 .

[0067] 2) Atmospheric correction: Atmospheric correction is to eliminate the influence of the atmosphere or light reflection on the ground, and to eliminate the influence of oxygen, nitrogen dioxide, methane and aerosol in the atmosphere, so as to achieve the process of inverting the true reflectivity.

[0068] Open the FLAASH Atmospheric Correction module in the Toolbox toolbox and set the parameters required for atmospheric correction in the FLAASHAtmospheric Correction Module Input Parameters panel ( Figure 5 ). The steps for setting the atmospheric correction parameters are as follows:

[0069] ① File input and output information items. Select the image after radiation calibration in [Input Radiance Image], and select Use single scale factor for allbands (Single scale factor: 1.000000) in the Radiance Scale Factors dialog box. Set the output file name and path in [Output Reflectance File].

[0070] ②Sensor and image information. Lat and LON are automatically acquired. Select Landsat-8OLT as the sensor type. The average elevation of the image area is acquired from the data column of the original image. The remaining imaging time (Flight Date) and pixel size (Pixel Size) are automatically acquired.

[0071] ③Multispectral Settings: In the Defaults drop-down box under Multispectral Settings, select Over-Land Retrieval Standard (660:2100).

[0072] ④Advanced Settings: Set Tile Size in Advanced Settings to 200. Other settings follow the default settings.

[0073] After performing the atmospheric correction, select the same pixel point and open the spectrum curves of the original image and the image after atmospheric correction at the same time ( Figure 6 ), it can be found that the corrected ground object spectrum curve is closer to the real ground object spectrum.

[0074] 3) Image cropping

[0075] Import the vector data of the study area boundary into ENVI, open the Subset Data from ROIs option in the Toolbox toolbox, select the image file after atmospheric correction, and set the parameters in the pop-up Spatial Subset via ROI Parameters panel ( Figure 7 ): Select the study area boundary vector file in the Select Input ROTs box, set the output file name and path, and follow the default settings for other settings. After cropping, you will get the image map of the study area ( Figure 8 ).

[0076] (3) Calculation of remote sensing characteristic factors

[0077] 1) Band reflectance. Each band of remote sensing images has rich spectral information. By obtaining the reflectance value of the band, we can analyze various ground features such as vegetation and soil. Use the band math tool (BandMath) in the Toolbox of ENVI, enter the band 1 calculation formula, select the image band corresponding to the calculation formula in the pop-up Variables to Bands Pairings panel, and set the relevant parameters as follows: Fig. 9 The calculation steps of the reflectivity of band 2 to band 9 are the same as that of band 1.

[0078] Band 8 (panchromatic band) and band 9 (water vapor absorption band) of Landsat 8OLT satellite digital images are used for enhanced resolution and cloud detection respectively, which have no significant relationship with the soil organic carbon and carbon density of the research object in this study, so they are removed. Extract the reflectance values ​​of band 1 to band 7 of the preprocessed remote sensing image ( Fig.10 ) were used to analyze its relationship with soil organic carbon content and soil organic carbon density.

[0079] The band operation tool in the Toolbox of ENVI is used to extract the single-band reflectance value of the sampling point, and the attribute table of the sampling point is constructed so that each sampling point has a corresponding value of each remote sensing characteristic factor.

[0080] 2) Vegetation index. Several commonly used vegetation indices were selected: Normalized Difference Vegetation Index (NDVI), Simple Ratio Index (SR), Soil Adjusted Vegetation Index (SAVI), and Enhanced Vegetation Index (EVI). The calculation formula of vegetation index is shown in Table 2.

[0081] Table 2 Calculation formula of vegetation index

[0082]

[0083] Note: ρ λ It represents the spectral reflectance at wavelength λ. NIR, Red, and Blu represent the near-infrared band, red band, and blue band of the Landsat8 OLT satellite, respectively.

[0084] Open the Band Math panel in the Toolbox, enter the calculation formula for NDVI, and select the image band corresponding to the calculation formula in the pop-up Variables to Bands Pairings panel to calculate NDVI. The relevant parameter settings are as follows: Fig.11 Referring to the calculation steps of NDVI and the calculation formulas of SR, SAVI and EVI (Table 2), the vegetation indices of SR, SAVI and EVI were determined ( Fig.12 ).

[0085] The remote sensing characteristic factors calculated in ENVI were exported to ArcGIS, and the vegetation index values ​​of the sampling points were extracted using the Multi-Value Extraction to Point tool in its spatial analysis.

[0086] Principal component characteristics. The first three components of the principal component contain most of the ground feature information of the remote sensing image. Open the Forward PCA Rotation New Statistics and Rotate panel in the Toolbox, select the preprocessed image file of the study area, and set the parameters in the pop-up Forward PC Parameters dialog box: [Stats X / YResize Factor] selects the default value 1, and sets an output statistics path and file name. [Calculate using] Select the covariance matrix (Covariance Matrix), select the output path and file name, and the output data type is FloatingPoint. [Number of Output PC Bands] Select the default value. After the execution is completed, the PC Eigenvalues ​​drawing window appears ( Fig.13 ), it can be seen from the figure that the first, second and third components (PCA1, PCA2, PCA3) have large eigenvalues, so the first three components of the principal component features are calculated and extracted, and the results are as follows Fig.14 shown.

[0087] The principal component features calculated in ENVI were exported to ArcGIS, and the multi-value extraction to point tool in its spatial analysis was used to extract the first three component features of the principal components of the sampling points.

[0088] Texture features. Select the Gray-Level Co-occurrence Matrix (GLCM) method to extract image texture features. Open the Co-occurrence Measures tool in the Toolbox, select the b1 image file in the pop-up Texture Input File dialog box, and set the parameters in the Co-occurrence Texture Parameters dialog box ( Fig.15 ): In [Texture to Compute], check the mean, entropy, correlation, and variance. Set the rest of the options to the default values. Finally, set the output path and file name. The relevant parameter settings and results are as follows: Fig.15 and Fig.16 The steps for texture feature analysis of the remaining band images b2 to b7 are the same.

[0089] The single-band texture features calculated in ENVI were exported to ArcGIS, and the single-band texture features of the sampling points were extracted using the Multi-Value Extraction to Point tool in ArcGIS's spatial analysis.

[0090] Model construction and accuracy verification. A regression algorithm was used to construct a remote sensing estimation model for soil organic carbon content, carbon density and remote sensing characteristic factors. 70% of the soil samples were selected for correlation analysis of organic carbon content, carbon density and remote sensing characteristic factors extracted from images. The remote sensing factors with strong correlation were used as independent variables, and soil organic carbon content and carbon density were used as dependent variables. Univariate linear, univariate curve and multivariate stepwise linear regression models were established, respectively. The equations are shown in Table 3.

[0091] Table 3 Model types and expressions

[0092]

[0093]

[0094] The accuracy of the remote sensing estimation model constructed was verified using 30% sample data. 2 , Mean Relative Error (MRE) and Root Mean Square Error (RMSE) were used as the model accuracy test factors in this study. 2The larger the value, the closer the model regression correlation is. MRE is used to measure the deviation between the measured value and the estimated value. RMSE is very sensitive to very large or very small errors in a set of measured values. These two error factors are often used to verify the accuracy of the model. The calculation formulas for each test factor are:

[0095]

[0096]

[0097]

[0098] Where: y i is the measured value, y i ' is the estimated value, i is the sample point number, n is the number of samples, cov is the covariance, and var is the variance.

[0099] Remote sensing inversion. The optimal remote sensing estimation model of SOC content and carbon density based on remote sensing characteristic factors was imported and run using the raster calculation tool in the spatial analysis of ArcGIS software to perform spatial distribution inversion of carbon content, carbon density and carbon storage in the surface soil of the study area ( Fig.17 ). In the process of remote sensing inversion, it was found that some areas in the image had negative values, and this part of the area was basically consistent with the water area under visual interpretation in the image. It is believed that the method based on remote sensing inversion of SOC is not suitable for areas covered by water bodies. In the result image, this area is divided into water areas and does not participate in the analysis of SOC spatial distribution characteristics.

[0100] GIS spatial interpolation analysis. The ordinary Kriging interpolation method of GIS spatial interpolation analysis was used to interpolate and draw the spatial distribution map of soil organic carbon content, carbon density and carbon storage ( Fig.18 ). The accuracy evaluation of ordinary Kriging interpolation is selected by the determination coefficient R 2 To measure, the calculation formula is as follows:

[0101]

[0102] Where R 2 is the determination coefficient; n is the number of sampling points; is the original value of the ith sampling point; y- is the mean of the original value; yi is the predicted value of ordinary Kriging interpolation.

[0103] The DNDC biogeochemical model developed by the Land, Ocean and Space Research Center of the University of New Hampshire was introduced and absorbed, and a Wetland-DNDC model suitable for simulating the dynamic carbon of seasonally flooded wetland soils was constructed to simulate the organic carbon density of wetland soils and estimate the carbon storage of wetland soils.

[0104] Obtain daily values ​​of meteorological factors such as temperature, precipitation, solar radiation, etc. at the research site, and prepare meteorological data (including the latitude and longitude, name, and simulation scale of the sampling point).

[0105] The soil data obtained from the experimental measurements were classified and sorted before being input into the model. Other parameters such as wilting point, field water holding capacity and porosity saturated hydraulic conductivity were set using the model default values.

[0106] Obtain the plant name, planting and harvesting time, total biomass of the plant, the proportion of seeds, leaves, stems and roots in the total biomass, as well as their respective carbon-nitrogen ratios, total water requirements, carbon fixation coefficients and other plant growth habits and basic ecological data.

[0107] Since the model itself does not have the wetland vegetation type of this study, a new crop naming method is adopted. Due to the particularity of the wetland vegetation itself (no grains), the proportion of grains in biomass is set to a smaller value of 0.01 in the data ratio setting.

[0108] The wetland-DNDC model (Table 4) was constructed using the meteorological data, soil data, and basic vegetation ecological data prepared above, and the model simulation time was determined at the same time.

[0109] Table 4 Required input parameters of DNDC model

[0110]

[0111] If the results obtained are significantly different from the measured results, the specific parameters of the model and the management conditions should be adjusted, and the parameters should be re-specified - the parameters of new sample points should be constructed or new sample points should be constructed - the previous operation results should be summarized, and a new management plan should be established based on the actual situation of the event.

[0112] Through continuous trials and adjustments, the implementation management plan of the site was determined, the model simulation was debugged, and then the model was run. The model output parameters are shown in Table 5.

[0113] Table 5 Output parameters of Wetland-DNDC model

[0114]

[0115] The root mean square error (RMSE), model efficiency (ME), mean absolute error (MAE) and correlation coefficient (R 2 ) and other model evaluation indicators to verify and evaluate the applicability of the constructed model.

[0116] The sensitivity index and the Monte Carlo method (MC) built into the DNDC model are selected to analyze the sensitivity index of the model, optimize the model parameters, minimize the errors of highly sensitive parameters, and effectively reduce the uncertainty range of subsequent model estimation.

[0117] After the simulation of the above points, the soil organic carbon density of different points can be generated, and the average soil carbon density under different vegetation types can be calculated. Combined with the selected areas of the three seasonally flooded wetland beaches, the soil carbon storage under different vegetation types at the three locations can be calculated (Table 6). The soil organic carbon storage under different vegetation types is obtained by multiplying the soil density under the vegetation type by the area obtained from the vegetation soil. The beach areas selected this time all use data from January 2021.

[0118] Table 6 Average soil carbon density and soil carbon storage under different vegetation types simulated based on the Wetland-DNDC model

[0119]

[0120] The spatial distribution characteristics of carbon density and carbon storage of SOC remote sensing estimation model, GIS spatial interpolation and Wetland-DNDC model were compared and analyzed (Table 7).

[0121] Table 7 Comparison of carbon density and carbon storage estimated by three models in the study area

[0122]

[0123] The Wetland-DNDC model was used to simulate soil CO2 changes under different vegetation types in seasonally flooded wetlands. 2 The annual emissions of greenhouse gases such as methane are estimated ( Fig.19 ).

[0124] Multi-year predictions of soil organic carbon under different vegetation types were conducted. The meteorological and soil data of the same region remained unchanged, and only the vegetation ecological data was changed. A 100-year simulation was conducted ( Fig. 20 ).

[0125] The present invention provides a comprehensive and scientific method to estimate carbon storage and carbon dynamics monitoring method of seasonally flooded wetlands. This method is not only applicable to lake wetlands, but can also be extended to other types of wetland systems after slight modification and deformation, providing important data support for ecological research, environmental management and global carbon cycle research.

[0126] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A method for estimating and dynamically monitoring soil carbon in seasonally flooded wetlands, characterized in that The following steps are involved: S1. Sample collection: From the center of the lake to the lakeshore, sample strips and sample plots were selected according to the water level gradient and dominant vegetation community type, GPS positioning was performed, and stakes were set up to fix the wetland area, wetland type, and distribution characteristics in the sample strips and sample plots. The imaging laws and image characteristics of different types of wetlands on remote sensing images were summarized. Several sampling points were set on each sample strip, and several sample plots were arranged within the sampling point range. At the sampling points, several soil samples of the surface soil (0-20 cm) were obtained using a soil sampler according to the five-point sampling method. After removing foreign matter from the surface of the soil, they were placed in self-sealing bags, and the soil samples were numbered, and their vegetation cover type and surrounding environmental factors were recorded. At the same time, the latitude and longitude information of the sample points were recorded using GPS; S2. Sample pretreatment: After the collected soil samples are brought back to the laboratory, each soil sample is first divided into two parts. One soil sample is kept in its original state for measuring soil bulk density, and the other is naturally air-dried. After air-drying, roots, sand and other small debris are further removed. Then, the soil is ground and sieved and sealed again for subsequent sample measurement. S3. Experimental determination of various indicators: Potassium dichromate oxidation-external heating method was used to determine soil organic carbon (SOC) content, and the ring knife method was used to determine soil bulk density, and carbon density and carbon storage were calculated; S4. Remote sensing feature factor extraction: Obtain Landsat 8 multispectral remote sensing images, interpret the remote sensing images, and then calculate the remote sensing feature factors. The remote sensing feature factors mainly include: image band reflectance, vegetation index, principal component characteristics, and single-band texture characteristics; S5. Construction of SOC remote sensing estimation model and accuracy verification: The correlation analysis method was used to select a part of remote sensing characteristic factors as independent variables to participate in the construction of the model. The regression algorithm was used to construct the remote sensing estimation model of SOC content, carbon density and remote sensing characteristic factors. The constructed remote sensing estimation model was verified for accuracy using a part of sample data. The optimal SOC content and carbon density remote sensing estimation model was selected after accuracy verification. The GIS raster statistics method was used to import the constructed optimal remote sensing estimation model to realize the remote sensing inversion of SOC content, carbon density and carbon storage spatial distribution characteristics. S6. GIS spatial interpolation analysis: The ordinary Kriging interpolation method of GIS spatial interpolation analysis was used to interpolate and draw the spatial distribution map of SOC content, carbon density and carbon storage, and predict the spatial distribution characteristics of SOC content, carbon density and carbon storage in the study area; S7. Introduction of DNDC biogeochemical model and construction of input parameter database: The DNDC biogeochemical model developed by the Land, Ocean and Space Research Center of the University of New Hampshire was introduced for digestion and absorption. Based on the input parameters required by the DNDC model, the daily scale values ​​of meteorological factors such as temperature, precipitation, and solar radiation at the research site were obtained, meteorological data were prepared, and the DNDC model input parameter database was prepared in combination with vegetation data and basic soil physical and chemical properties data; S8. Construction of a wetland soil carbon dynamics simulation model (Wetland-DNDC): Based on the constructed DNDC model input parameter database, the DNDC model was repeatedly debugged and verified, model parameters suitable for seasonally flooded wetland soil carbon dynamics simulation were screened, data simulation was performed and model sensitivity was calculated to verify the feasibility of the model, thereby constructing a Wetland-DNDC model suitable for seasonally flooded wetland soil carbon dynamics simulation; S9. Wetland-DNDC model carbon dynamic simulation prediction: The constructed Wetland-DNDC model is used to simulate and obtain the soil organic carbon density value of the study area, and the soil carbon storage in the study area is estimated based on the wetland area. The spatial distribution characteristics of carbon density and carbon storage are analyzed, and the long-term dynamic changes of SOC in seasonally flooded wetlands and greenhouse gas emissions such as CO2 and CH4 are predicted. S10. Field survey, 3S technology, and Wetland-DNDC model coupling to estimate carbon storage and carbon dynamics: Compare and analyze the results of field survey, remote sensing inversion, GIS interpolation analysis, and Wetland-DNDC model simulation. Based on the field survey, optimize the integrated coupling of 3S technology and the Wetland-DNDC model to evaluate the carbon sequestration capacity of wetland soils and predict the response of wetland soil carbon dynamics to future climate.

2. The method for estimating and dynamically monitoring soil carbon in seasonally flooded wetlands according to claim 1, characterized in that: In step S3, the SOC content is determined by the calculation formula when the potassium dichromate oxidation-external heating method is: In the formula: 0.8000 is the concentration of potassium dichromate standard solution in mol / L; 5.0 is the volume of potassium dichromate standard solution in ml; V0 represents the volume of ammonium ferrous sulfate standard solution consumed by the blank sample in ml; V represents the volume of ammonium ferrous sulfate standard solution consumed by the soil sample in ml; 0.003 is the millimolar mass of 1 / 4 carbon atom in g; 1.10 is the oxidation correction coefficient; m is the sample mass in g.

3. The method for estimating and dynamically monitoring soil carbon in seasonally flooded wetlands according to claim 1, characterized in that: In step S3, the soil bulk density is determined by the ring knife method: a portion of the reserved soil sample is taken out, and the original soil is cut with a ring knife (i.e., a circular iron container) so that the original soil fills the entire container, and then weighed and dried to a constant weight, and the bulk density (unit: g / cm 3 ), the calculation formula of bulk density is: Where, ρ: soil bulk density (g / cm 3 ); V: ring knife volume (cm 3 ); m: weight of wet soil in the ring cutter (g); W: moisture content of soil in the ring cutter (%).

4. The method for estimating and dynamically monitoring soil carbon in seasonally flooded wetlands according to claim 1, characterized in that: In step S3, the calculation formulas for soil organic carbon density (SOCD) and soil organic carbon stock (SOCS) are: SOCD=SOC×BD×D×0.1 SOCS=A×SOCD Where, SOCD: soil organic carbon density (t / hm 2 ); SOC: soil organic carbon content (g / kg); BD: soil bulk density (g / cm 3 ); D: soil thickness (cm); SOCS: soil organic carbon storage (t); A: study area (hm 2 ).

5. The method for estimating and dynamically monitoring soil carbon in seasonally flooded wetlands according to claim 1, characterized in that: In step S4, the remote sensing image preprocessing steps mainly include radiation calibration, atmospheric correction and image cropping.

6. The method for estimating and dynamically monitoring soil carbon in seasonally flooded wetlands according to claim 1, characterized in that: In step S5, 70% of the soil samples were selected for correlation analysis of organic carbon content, carbon density and remote sensing characteristic factors extracted from images. The remote sensing factors with strong correlation were used as independent variables, and SOC content and carbon density were used as dependent variables to establish univariate linear, univariate curve and multivariate stepwise linear regression models, respectively. The equations are as follows: A: The model is a univariate linear regression, its equation type is a linear form, and its expression is Y=b0+b1x; B: The model is a univariate curvilinear regression, and its equation type is: B1: The equation type is quadratic, and its expression is Y=b0+b1x+b2x 2 ; B2: The equation type is cubic, and its expression is Y=b0+b1x+b2x 2 +b3x 3 ; B3: The equation type is power type, and its expression is B4: The equation type is growth type, and its expression is B5: The equation type is exponential, and its expression is C: The model is a multivariate stepwise linear regression, and its equation type is a stepwise linear type. Its expression is Y=b0+b1x1+...+b i x i +...+b n x n .

7. The method for estimating and dynamically monitoring soil carbon in seasonally flooded wetlands according to claim 1, characterized in that: In step S5, a portion of sample data is used to verify the accuracy of the constructed remote sensing estimation model: the model determination coefficient R 2 , Mean Relative Error (MRE) and Root Mean Square Error (RMSE) are used as model accuracy test factors. 2 The larger the value, the closer the model regression correlation is. MRE is used to measure the deviation between the measured value and the estimated value. RMSE is very sensitive to very large or very small errors in a set of measured values. These two error factors are used to verify the accuracy of the model. The calculation formulas for each test factor are: Where: y i is the measured value, y i ' is the estimated value, i is the sample point number, n is the number of samples, cov is the covariance, and var is the variance.

8. The method for estimating and dynamically monitoring soil carbon in seasonally flooded wetlands according to claim 1, characterized in that: In step S6, the accuracy evaluation of ordinary Kriging interpolation selects the determination coefficient R 2 To measure, the calculation formula is as follows: Where R 2 is the determination coefficient; n is the number of sampling points; is the original value of the i-th sampling point; is the mean of the original values; yi is the predicted value of ordinary Kriging interpolation.

9. The method for estimating and dynamically monitoring soil carbon in seasonally flooded wetlands according to claim 1, characterized in that: In step S8, the root mean square error (RMSE), the coefficient of determination (R 2 ) to verify the applicability of the model, the calculation formula is as follows: In the formula, Pi represents the i-th simulated value, Qi represents the i-th observed value, Q and P are the average values ​​of the observed value and the measured value, and n is the number of observations or simulations. The sensitivity index is used to analyze the model sensitivity, and the calculation formula is as follows: SLi=(△Oi÷FOi) / (△Fi÷Fi) Among them: SLi is the sensitivity index of the ith parameter; △Oi is the difference between the maximum and minimum simulation values ​​of the ith input parameter; FOi is the mean of the simulation results corresponding to the input parameter; △Fi is the difference between the maximum and minimum values ​​of the ith input parameter; Fi is the mean of the ith input parameter. The larger the absolute value of SLi, the greater the impact of the input parameter on the simulation value.

10. The method for estimating and dynamically monitoring soil carbon in seasonally flooded wetlands according to claim 1, characterized in that: In step S9, the soil organic carbon density value is automatically presented after the Wetland-DNDC model is run. The calculation formula for estimating the wetland soil carbon storage can be found in the calculation formula for the soil organic carbon storage in step S3. In step S10, by comparing and analyzing the results of steps S3, S5, S6, and S9, coupled integrated optimization is performed to propose an optimal seasonally flooded wetland soil carbon estimation and carbon dynamic monitoring method to evaluate the wetland soil carbon sequestration capacity.

Images