A method and system for assessing wetland net ecosystem productivity

Abstract

This invention discloses a method and system for assessing net ecosystem productivity (NEP) of wetlands. The method includes: firstly, collecting initial data of the wetland ecosystem to be assessed; constructing a wetland NEP inversion model through three-dimensional spectral feature extraction and multi-model weighted fusion, generating an NEP spatial distribution map; obtaining a simulation chamber for simulating the wetland ecosystem to be assessed, and constructing a corresponding digital twin system based on the simulation chamber; collecting target data of the wetland ecosystem to be assessed through the simulation chamber device layer; transmitting the target data to the virtual model layer using a data interaction layer, driving the update of the RH inversion model based on a hybrid algorithm to generate RH spatiotemporal distribution data; and generating a spatial distribution map of wetland NEP, a sensitivity analysis map, a community impact map, and an annual variation map of the wetland ecosystem to be assessed based on the NEP spatial distribution map and the RH spatiotemporal distribution data. This method achieves high-precision estimation of wetland ecosystem NEP.

Description

Technical Field

[0001] This invention relates to the fields of wetland ecosystem carbon cycle monitoring, ecological remote sensing analysis, and carbon sink assessment, specifically to a method and system for assessing wetland net ecosystem productivity. Background Technology

[0002] As an important terrestrial blue carbon reservoir, wetland ecosystems rely on net ecosystem productivity (NEP) as a core indicator of carbon sequestration capacity, which is jointly determined by net primary productivity (NPP) and soil microbial respiration (RH). However, the complex and diverse vegetation structure of wetlands, their high proportion of underground biomass, and their strong influence from environmental factors such as hydrology and salinity have led to numerous technical bottlenecks in traditional NEP assessment methods, which have become urgent technical problems to be solved. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for assessing net ecosystem productivity (NEP) in wetlands, addressing the shortcomings of traditional techniques. It achieves high-precision estimation of NEP in wetland ecosystems, providing technical support for research on carbon cycling and ecological protection in wetland ecosystems.

[0004] One embodiment of this application provides a method for assessing the net ecosystem productivity of wetlands, comprising: Initial data on the wetland ecosystem to be evaluated are collected; wherein, the initial data includes hyperspectral image data, basic environmental data, and historical biomass dynamic data; Based on the initial data, a wetland net primary productivity (NPP) inversion model is constructed by three-dimensional spectral feature extraction and multi-model weighted fusion, and an NPP spatial distribution map is generated. A simulation chamber for simulating the wetland ecosystem to be evaluated is obtained, and a corresponding digital twin system is constructed based on the simulation chamber; wherein, the digital twin system includes a simulation chamber device layer, a virtual model layer, and a data interaction layer; The target data of the wetland ecosystem to be evaluated are collected through the simulation chamber device layer; wherein, the target data includes Emission flux data and environmental factor data; The target data is transmitted to the virtual model layer using the data interaction layer, driving the RH inversion model based on the hybrid algorithm to iteratively update in order to generate RH spatiotemporal distribution data; Based on the NPP spatial distribution map and the RH spatiotemporal distribution data, a wetland net ecosystem productivity spatial distribution map, sensitivity analysis map, community impact map, and annual change map are generated for the wetland ecosystem to be evaluated.

[0005] Optionally, based on the initial data, a wetland net primary productivity (NPP) inversion model is constructed through three-dimensional spectral feature extraction and multi-model weighted fusion to generate an NPP spatial distribution map, including: Radiometric correction, illumination correction, geometric correction, and image stitching are performed sequentially on the hyperspectral image data to determine the pixel spectral data; Extract three-dimensional spectral feature indicators from the pixel spectral data; wherein, the three-dimensional spectral feature indicators include vegetation index, sensitive narrowband features, and continuous spectral features; Based on the validation set composed of the three-dimensional spectral feature index, the environmental basic data, and the historical biomass dynamic data, a random forest model, an XGBoost model, and a PLSR model were constructed respectively. The root mean square error and the coefficient of determination of the validation set were used as the basis for weighted fusion to obtain the optimal inversion model of aboveground biomass AGB and the optimal inversion model of belowground biomass BGB. By combining data on plant growth stages, hydrological conditions, and typical plant structural parameters of the wetland ecosystem to be evaluated, the optimal AGB and BGB inversion models are optimized to construct a wetland net primary productivity (NPP) inversion model and generate an NPP spatial distribution map.

[0006] Optionally, obtaining a simulation chamber for simulating the wetland ecosystem to be evaluated, and constructing a corresponding digital twin system based on the simulation chamber, includes: Simulation boxes were evenly distributed within the wetland ecosystem to be evaluated, based on the distribution characteristics of wetland hydrological zones, salinity gradients, and plant types. Each simulation box included an infrared sensor. Analyzer and / or Analyzer, multi-parameter sensing module, and wireless communication module; A digital twin parameter mapping model was constructed using the respiratory mechanism of wetland soil microorganisms. Based on the digital twin parameter mapping model, establish bidirectional communication between the simulation box device layer and the data interaction layer; An HMI visualization interface is constructed to generate a digital twin system for estimating the RH of the wetland ecosystem to be evaluated; wherein, the visualization interface includes a target data display module, a model iteration and adjustment module, and an RH result output function module.

[0007] Optionally, the acquisition of target data of the wetland ecosystem to be evaluated through the simulation box device layer includes: Through the built-in infrared of the simulation box The analyzer continuously monitors the changes in gas concentration inside the chamber at a first preset frequency, and calculates the concentration using the chamber method formula. Emission flux data; The simulation chamber uses a multi-parameter sensing module to synchronously collect environmental factor data at a second preset frequency; the environmental factor data includes water level, soil temperature, soil moisture content, salinity, and redox potential data. All collected Emission flux data and environmental factor data are transmitted to the data interaction layer in real time to obtain target data and are then transmitted synchronously to the virtual model layer.

[0008] Optionally, the above will collect all the data. Emission flux data and environmental factor data are transmitted in real time to the data interaction layer to obtain target data and are simultaneously transmitted to the virtual model layer, including: Perform a decomposition operation on the environmental factor data to obtain intrinsic mode functions; Based on the intrinsic mode function, noise reduction and data reconstruction operations are performed on the environmental factor data to obtain the noise-reduced environmental factor data. Based on the above The target data is generated from emission flux data and noise-reduced environmental factor data.

[0009] Optionally, performing the decomposition operation on the environmental factor data to obtain the intrinsic mode functions includes: Set the preset number of decompositions for environmental factor data; Obtain the noisy environmental factor data corresponding to the environmental factor data, and perform the decomposition operation on the noisy environmental factor data based on the empirical mode decomposition algorithm to obtain the intrinsic mode function components; If the current decomposition count has not reached the preset decomposition count, return to the step of obtaining the noisy environmental factor data corresponding to the environmental factor data, and perform the decomposition operation on the noisy environmental factor data based on the empirical mode decomposition algorithm until the preset number of intrinsic mode function components are obtained. The intrinsic mode functions are determined based on the preset number of intrinsic mode function components.

[0010] Optionally, the step of performing the decomposition operation on the noisy environmental factor data based on the empirical mode decomposition algorithm to obtain intrinsic mode function components includes: The decomposition operation of the noisy environmental factor data is performed using the following formula: in, , , , This represents Gaussian white noise with a mean of 0. This represents the local mean of the noisy environmental factor data. The mathematical expectation operator, This represents the preset amplitude of the noise-containing environmental factor data. This represents the signal-to-noise ratio (SNR) of the i-th decomposition stage. This represents the eigenmode function component of the i-th decomposition stage. This represents the residual term in the i-th decomposition stage.

[0011] Another embodiment of this application provides a system for assessing the net ecosystem productivity of wetlands, the system comprising: The first acquisition module is used to acquire initial data of the wetland ecosystem to be evaluated; wherein, the initial data includes hyperspectral image data, basic environmental data, and historical biomass dynamic data; The construction module is used to construct a wetland net primary productivity (NPP) inversion model based on the initial data through three-dimensional spectral feature extraction and multi-model weighted fusion, and generate an NPP spatial distribution map. The module is used to obtain a simulation box for simulating the wetland ecosystem to be evaluated, and to construct a corresponding digital twin system based on the simulation box; wherein, the digital twin system includes a simulation box device layer, a virtual model layer and a data interaction layer; The second acquisition module is used to acquire target data of the wetland ecosystem to be evaluated through the simulation box device layer; wherein, the target data includes Emission flux data and environmental factor data; The transmission module is used to transmit the target data to the virtual model layer using the data interaction layer, and drive the RH inversion model based on the hybrid algorithm to iteratively update in order to generate RH spatiotemporal distribution data. The generation module is used to generate, based on the NPP spatial distribution map and the RH spatiotemporal distribution data, a wetland net ecosystem productivity spatial distribution map, a sensitivity analysis map, a community impact map, and an annual change map of the wetland ecosystem to be evaluated.

[0012] Another embodiment of this application provides a storage medium storing a computer program, wherein the computer program is configured to implement the method described in any of the above-described embodiments when running.

[0013] Another embodiment of this application provides an electronic device including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to implement the method described in any of the above embodiments.

[0014] Compared with existing technologies, this invention first collects initial data on the wetland ecosystem to be evaluated; based on the initial data, a wetland net primary productivity (NPP) inversion model is constructed through three-dimensional spectral feature extraction and multi-model weighted fusion, generating an NPP spatial distribution map; a simulation chamber is obtained to simulate the wetland ecosystem to be evaluated, and a corresponding digital twin system is constructed based on the simulation chamber; target data of the wetland ecosystem to be evaluated is collected through the simulation chamber device layer; the target data is transmitted to the virtual model layer using the data interaction layer, driving the iterative update of the RH inversion model based on a hybrid algorithm to generate RH spatiotemporal distribution data; based on the NPP spatial distribution map and the RH spatiotemporal distribution data, a wetland net ecosystem productivity spatial distribution map, sensitivity analysis map, community impact map, and annual change map of the wetland ecosystem to be evaluated are generated. It achieves high-precision estimation of wetland ecosystem net ecosystem productivity (NEP), providing technical support for wetland ecosystem carbon cycle research and ecological protection. Attached Figure Description

[0015] Figure 1 Hardware structure block diagram of a computer terminal for a method of assessing the net ecosystem productivity of wetlands provided in an embodiment of the present invention; Figure 2 A flowchart illustrating a method for assessing the net ecosystem productivity of wetlands provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the framework of a wetland net ecosystem productivity assessment system provided in an embodiment of the present invention. Detailed Implementation

[0016] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0017] The present invention first provides a method for assessing the productivity of a wetland net ecosystem. This method can be applied to electronic devices, such as computer terminals, specifically ordinary computers, tablets, etc.

[0018] The following detailed explanation uses a computer terminal as an example. Figure 1 This is a hardware block diagram of a computer terminal for a method of assessing the net ecosystem productivity of wetlands provided in an embodiment of the present invention. Figure 1 As shown, the computer device includes a processor, memory, and network interface connected via a system bus, wherein the memory may include non-volatile storage media and internal memory.

[0019] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions that, when executed, cause the processor to perform any method for assessing the net ecosystem productivity of a wetland.

[0020] The processor provides computing and control capabilities, supporting the operation of the entire computer device.

[0021] Internal memory provides an environment for the execution of computer programs in non-volatile storage media, which, when executed by a processor, enable the processor to perform any method for assessing the net ecosystem productivity of wetlands.

[0022] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 1 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0023] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0024] See Figure 2 , Figure 2 A flowchart illustrating a method for assessing the net ecosystem productivity of wetlands, provided in an embodiment of the present invention, may include the following steps: S201: Collect initial data on the wetland ecosystem to be evaluated; wherein the initial data includes hyperspectral image data, basic environmental data, and historical biomass dynamic data.

[0025] Specifically, the geographical boundaries of the wetland ecosystem to be assessed should first be clearly defined. This can be done using GIS spatial analysis technology, combined with wetland hydrological characteristics, salinity distribution, and vegetation type differences, to complete three-dimensional zoning. For example, it can be divided into hydrological zones, salinity gradient zones, and vegetation type zones. Differentiated data collection plans should be developed for each zone to ensure that collection points cover all zone types, and that the number of collection points within each zone is no less than three, to accommodate the highly heterogeneous characteristics of wetlands.

[0026] A drone equipped with a hyperspectral sensor can be used for full-coverage data collection. During the collection process, a whiteboard can be carried for synchrotron radiation calibration, such as collecting whiteboard data every 30 minutes of flight for subsequent radiometric correction. Simultaneously, flight trajectory, altitude, and air pressure parameters are recorded to provide a basis for geometric correction. After collection, the raw hyperspectral data undergoes format conversion and preliminary screening to remove invalid data caused by sensor malfunction, cloud cover, or motion blur. Environmental baseline data, including topographic, meteorological, and hydrological data, is used to support the optimization of the NPP inversion model and the construction of a digital twin system for RH estimation. For example, topographic data can be obtained from a digital elevation model (DEM). Historical biomass dynamic data, including aboveground biomass (AGB), belowground biomass (BGB), and related auxiliary data, is used for subsequent model training and validation. This can be achieved by prioritizing the collection of measured biomass data from the wetland to be assessed over the past 3-5 years, collecting biomass research data from similar wetlands, or remote sensing inversion biomass data, and performing consistency correction. After all initial data collection is completed, preprocessing steps are performed to ensure data quality.

[0027] S202: Based on the initial data, a wetland net primary productivity (NPP) inversion model is constructed by extracting three-dimensional spectral features and weighted fusion of multiple models, and an NPP spatial distribution map is generated.

[0028] Specifically, based on the initial data, constructing a wetland net primary productivity (NPP) inversion model through three-dimensional spectral feature extraction and multi-model weighted fusion to generate an NPP spatial distribution map may include: Step 1: Perform radiometric correction, illumination correction, geometric correction, and image stitching sequentially on the hyperspectral image data to determine the pixel spectral data; Step 2: Extract three-dimensional spectral feature indicators from the pixel spectral data; wherein, the three-dimensional spectral feature indicators include vegetation index, sensitive narrowband features, and continuous spectral features; Step 3: Based on the validation set composed of the three-dimensional spectral feature index, the environmental basic data, and the historical biomass dynamic data, construct the random forest model, XGBoost model, and PLSR model respectively. Perform weighted fusion based on the root mean square error and coefficient of determination of the validation set to obtain the optimal inversion model of aboveground biomass AGB and the optimal inversion model of belowground biomass BGB. Step 4: Combining the data on plant growth stages, hydrological conditions, and typical plant structural parameters of the wetland ecosystem to be evaluated, optimize the AGB optimal inversion model and the BGB optimal inversion model to construct a wetland net primary productivity (NPP) inversion model and generate an NPP spatial distribution map.

[0029] For example, radiometric correction, illumination correction, geometric correction, and image stitching are sequentially performed on hyperspectral image data to obtain pixel spectral data. Then, a three-dimensional feature index system sensitive to vegetation in wetland ecosystems is constructed, including vegetation indices, sensitive narrowband features, and continuous spectral features. Among them, vegetation indices include Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Normalized Red Edge Difference Index (NDRE), Red Edge Chlorophyll Index (CIred-edge), and Modified Chlorophyll Absorption and Reflectance Index (MCARI); sensitive narrowband features include the 680–750 nm red edge band, 970 nm, 1200 nm, 1450 nm, and 1940 nm water absorption bands, and the 2100–2300 nm structure and lignin band; continuous spectral features include principal component, derivative spectrum, and envelope features.

[0030] The Normalized Difference Vegetation Index (NDVI) is calculated by comparing the red light band (630-690nm, where vegetation has strong absorption) and the near-infrared band (760-900nm, where vegetation has strong reflectance), highlighting vegetation cover and growth vitality. It can quickly identify the distribution range of wetland vegetation and provide a preliminary assessment of vegetation cover, but it is easily affected by wetland water background and soil moisture. The Enhanced Vegetation Index (EVI) introduces the blue light band (450-520nm) to suppress atmospheric scattering interference and incorporates soil modifiers to reduce the influence of soil background and atmospheric aerosols, making it more suitable for areas with high vegetation cover and complex habitats (such as wetlands). It can accurately reflect the growth status of high-density wetland vegetation, such as estimating the biomass of Spartina alterniflora invasive communities, and reduces atmospheric disturbance caused by water evaporation. The Normalized Red Edge Difference Index (NRD) replaces the red band in NDVI with the red edge band (700-750 nm, a sensitive band for vegetation chlorophyll content). The red edge band is more sensitive to early growth stages and chlorophyll changes in vegetation, capturing subtle differences in vegetation growth. It can monitor subtle changes in wetland vegetation phenology (such as the greening and jointing stages), assess the salt stress level of salt-tolerant plants such as Salicornia and Suaeda, and provide data support for phenological correction coefficients in NPP estimation. The Red Edge Chlorophyll Index is a red edge index specifically for vegetation chlorophyll content. It directly correlates chlorophyll concentration through the ratio or difference between the red edge band and the near-infrared band, avoiding interference from other pigments (such as carotenoids), and has higher accuracy than the traditional chlorophyll index. It can quantify the photosynthetic capacity of wetland vegetation, such as the dynamic monitoring of chlorophyll in Reed and Curcuma longa, providing core input for vegetation physiological characteristic parameters in NPP estimation. The modified chlorophyll absorption reflectance index enhances the sensitivity to chlorophyll absorption characteristics by combining the red light band (strong chlorophyll absorption), the near-infrared band (strong vegetation reflectance), and the green light band (530-570nm, weak chlorophyll absorption), while correcting for reflection interference from soil background (such as bare soil in wetlands and substrate in shallow water areas). It can accurately invert vegetation chlorophyll content in areas where wetland vegetation, bare soil, and shallow water areas are interspersed (such as intertidal wetlands), reducing interference from soil background and water reflection, and providing highly reliable spectral feature input for biomass inversion models.

[0031] For example, in constructing a three-dimensional spectral feature system, the improved red-edge chlorophyll index accurately reflects chlorophyll content; the lignin sensitivity index characterizes the degree of vegetation lignification; the coupling feature between the normalized water index and the absorption depth (Depth1450) at 1450 nm is used, where Depth1450 is calculated using the envelope removal method to fit the spectral envelope near the 1450 nm band, and the maximum absolute value of the difference between the band reflectance and the envelope is calculated; soil background correction features are used: based on the red-edge slope (RES) and blue-edge position (BEP) after envelope removal, soil background interference is reduced. Principal component analysis (PCA) is used to reduce the dimensionality of the above feature set, retaining principal components with a cumulative contribution rate ≥90% as modeling input.

[0032] Then, based on the validation set composed of three-dimensional spectral feature indicators, the aforementioned environmental baseline data, and the aforementioned historical biomass dynamic data, random forest, XGBoost, and PLSR models were constructed respectively. Weighted fusion was performed using the root mean square error and coefficient of determination of the validation set to obtain the optimal inversion models for aboveground biomass (AGB) and belowground biomass (BGB). Specifically, the weights for weighted fusion can be calculated as follows: Based on the root mean square error (RMSE) of the validation set data, the weights are negatively correlated with the RMSE, and the weight calculation satisfies: in, For the first The weights of the model Indicates the first The coefficient of determination of the model For the first The root mean square error of the model.

[0033] It should be noted that the optimal AGB / BGB inversion model obtained by fusion can be configured to require AGB inversion. ≥0.85, BGB inversion ≥0.78.

[0034] By combining data on plant growth stages, hydrological conditions, and typical plant structural parameters of the wetland ecosystem to be evaluated, the optimal AGB and BGB inversion models are optimized to construct a wetland net primary productivity (NPP) inversion model and generate an NPP spatial distribution map.

[0035] Specifically, based on biomass dynamic data, a wetland net primary productivity (NPP) inversion model is constructed by introducing phenological correction coefficient P, hydrological limiting factor H, and soil nutrient coefficient S: in, These represent the current aboveground biomass and the previous aboveground biomass, respectively. The values ​​represent the current underground biomass and the previous underground biomass. P is determined based on the phenological stages (0.3 for the greening stage, 0.8 for the jointing stage, 1.0 for the flowering stage, and 0.2 for the withering stage). S is determined by the ratio of soil organic matter content to the optimal value of soil organic matter, which is determined based on the organic matter content corresponding to the peak vegetation growth in the sample plot.

[0036] Finally, the trained wetland net primary productivity (NPP) inversion model was used to perform pixel-by-pixel extrapolation of the image data of the wetland ecosystem to be estimated, in order to generate spatial distribution maps of aboveground biomass, belowground biomass, and net primary productivity.

[0037] S203: Obtain a simulation box for simulating the wetland ecosystem to be evaluated, and construct a corresponding digital twin system based on the simulation box; wherein, the digital twin system includes a simulation box device layer, a virtual model layer, and a data interaction layer.

[0038] Specifically, obtaining a simulation chamber for simulating the wetland ecosystem to be evaluated, and constructing a corresponding digital twin system based on the simulation chamber, may include: 1. Simulation boxes are evenly distributed within the wetland ecosystem to be evaluated, according to the distribution characteristics of wetland hydrological zones, salinity gradients, and plant types; the simulation boxes include infrared [equipment / systems]. Analyzer and / or Analyzer, multi-parameter sensing module, and wireless communication module; 2. Construct a digital twin parameter mapping model by utilizing the respiratory mechanism of wetland soil microorganisms; 3. Based on the digital twin parameter mapping model, establish bidirectional communication between the simulation box device layer and the data interaction layer; 4. Construct an HMI visualization interface to generate a digital twin system for estimating the RH of the wetland ecosystem to be evaluated; wherein the visualization interface includes a target data display module, a model iteration and adjustment module, and an RH result output function module.

[0039] For example, depending on the type of wetland to be evaluated, such as coastal wetland tidal flats or inland freshwater wetlands, one can choose either a gate-type tidal level synchronization simulation box or a U-shaped tube weak open-circuit water level synchronization simulation box. The simulation box has a built-in water level regulation module, which responds in real time to changes in the external wetland water level through the opening and closing of the gate or the hydraulic balance principle of the U-shaped tube, maintaining the soil hydrological environment inside the box consistent with the original location.

[0040] For example, simulation boxes are set up at preset intervals in the wetland area to be evaluated, and an infrared sensor is installed on top of each simulation box. Analyzer or optional The analyzer can be equipped with multi-parameter sensing modules, such as water level sensors, soil temperature sensors, water content sensors, salinity sensors, and redox potential sensors, and can also be equipped with a Wi-Fi wireless communication module to form a simulation chamber device layer with multi-parameter acquisition and data transmission capabilities.

[0041] Based on the physicochemical mechanisms of microbial respiration in wetland soil, this study integrates substrate decomposition kinetics and gas diffusion mass transfer principles to establish an intrinsic correlation model between RH and core factors such as water level, temperature, and salinity, laying the physical foundation for parameter mapping. Three algorithms—random forest, XGBoost, and LSTM—are integrated to construct a digital twin parameter mapping model. Random forest and XGBoost are used to capture nonlinear interactions between factors, while LSTM is used to mine dynamic changes in time-series monitoring data. Weighted ensemble optimization is used to improve model prediction accuracy. Finally, a bidirectional data link between the simulation chamber device layer and the data interaction layer is constructed based on the TCP / IP protocol. The simulation chamber uploads collected gas flux and environmental parameters to the data interaction layer in real time via a wireless communication module. The data interaction layer feeds back model optimization commands to the simulation chamber's sensing module. The target data synchronization frequency is set to 5-30 minutes / time, with the gas flux parameter synchronization frequency set to 10-30 minutes / time and the environmental factor parameter synchronization frequency set to 5-20 minutes / time, ensuring that data timeliness matches model response speed.

[0042] In one alternative implementation, a web-based HMI visualization interface can be developed, integrating three core functional modules: a target data display module for real-time display of gas flux, environmental parameters, and model predictions; a model iteration and adjustment module for supporting manual correction of algorithm weights or importing offline data to optimize the model; and an RH result output module for supporting the export of single-point RH time-series curves and regional RH statistical reports.

[0043] For example, by activating the monitoring function of the simulation box device layer, the stability and integrity of sensor data transmission are verified; the initial parameters of the digital twin parameter mapping model are calibrated using historical measured data to ensure that the model prediction deviation is small; and the bidirectional communication delay of the data interaction layer is adjusted to ensure that the synchronization error is no more than 30 seconds, so as to complete the construction of the digital twin system for estimating the RH of the wetland to be evaluated.

[0044] S204: Collect target data of the wetland ecosystem to be evaluated through the simulation box device layer; wherein, the target data includes Emission flux data and environmental factor data.

[0045] Specifically, the collection of target data of the wetland ecosystem to be evaluated through the simulation box device layer may include: A. Through the built-in infrared of the simulation box The analyzer continuously monitors the changes in gas concentration inside the chamber at a first preset frequency, and calculates the concentration using the chamber method formula. Emission flux data; B. Environmental factor data are synchronously collected at a second preset frequency through the multi-parameter sensing module built into the simulation box; wherein, the environmental factor data includes water level, soil temperature, soil moisture content, salinity, and redox potential data; C. Collect all the data Emission flux data and environmental factor data are transmitted to the data interaction layer in real time to obtain target data and are then transmitted synchronously to the virtual model layer.

[0046] Specifically, based on the analog box device layer of the digital twin system, the infrared... Analyzer or optional The installation positions of the analyzer and multi-parameter sensing module should ensure unobstructed detection paths between the sensors and the soil or environment inside the simulation chamber, and the connection and debugging between the sensors and the data interaction layer should be completed through the wireless communication module.

[0047] Start the infrared sensor installed on the top of the simulation box. The analyzer can continuously monitor changes in gas concentration inside the chamber at a preset frequency of 10-30 minutes, calculate the flux value in real time based on the chamber method formula, and upload the collected data to the data interaction layer of the digital twin system in real time. Emission flux data were obtained through bin method calculation, using the following formula: in, This represents the change in gas concentration per unit time. For the monitoring time interval, To simulate the box volume, The area of ​​soil covering the simulation box.

[0048] The simulation chamber uses a built-in water level sensor to monitor water level changes in real time, simultaneously acquiring in-situ water level data of the wetland under evaluation. Water level data is collected every 5-20 minutes at a second preset frequency, recording peak and trough values ​​and their duration. The data is then calibrated and uploaded to the data exchange layer. Soil temperature, soil moisture, salinity, and redox potential sensors are inserted into the soil at a preset depth within the simulation chamber, simultaneously collecting soil temperature, soil moisture, salinity, and redox potential data every 5-30 minutes. Finally, an air temperature sensor is installed on the upper exterior of the simulation chamber, avoiding direct sunlight and water accumulation, to collect ambient air temperature data of the wetland under evaluation every 5-30 minutes, ensuring the data reflects the actual atmospheric temperature conditions of the wetland.

[0049] In one alternative implementation, the step of collecting all the data... Emission flux data and environmental factor data are transmitted in real time to the data interaction layer to obtain target data and are then synchronously transmitted to the virtual model layer, which may include: Perform a decomposition operation on the environmental factor data to obtain intrinsic mode functions (IMFs); based on the IMFs, perform denoising and data reconstruction operations on the environmental factor data to obtain denoised environmental factor data; based on the... The target data is generated from emission flux data and noise-reduced environmental factor data.

[0050] The step of performing the decomposition operation on the environmental factor data to obtain intrinsic mode functions (IMFs) includes: setting a preset number of decompositions for the environmental factor data; obtaining noisy environmental factor data corresponding to the environmental factor data; performing the decomposition operation on the noisy environmental factor data based on the empirical mode decomposition algorithm to obtain IMF components; if the current decomposition count has not reached the preset number of decompositions, returning to the steps of obtaining the noisy environmental factor data corresponding to the environmental factor data and performing the decomposition operation on the noisy environmental factor data based on the empirical mode decomposition algorithm, until the preset number of IMF components are obtained; and determining the IMFs based on the preset number of IMF components.

[0051] The step of performing the decomposition operation on the noisy environmental factor data based on the empirical mode decomposition algorithm to obtain the intrinsic mode function components may include: The decomposition operation of the noisy environmental factor data is performed using the following formula: in, , , , This represents Gaussian white noise with a mean of 0. This represents the local mean of the noisy environmental factor data. The mathematical expectation operator, This represents the preset amplitude of the noise-containing environmental factor data. This represents the signal-to-noise ratio (SNR) of the i-th decomposition stage. This represents the eigenmode function component of the i-th decomposition stage. This represents the residual term in the i-th decomposition stage.

[0052] For example, environmental factor data can be decomposed based on empirical mode decomposition algorithms to obtain a series of intrinsic mode functions, which represent components at different frequency scales in the original environmental factor data.

[0053] When using Empirical Mode Decomposition (EMD) algorithms to process environmental factor data, a fixed preset number of decomposition iterations is typically not set directly. EMD is a data-driven adaptive decomposition method that decomposes intrinsic mode functions (EMFs) at different scales based on the characteristics of the signal itself. In this application, to control the complexity of the decomposition and avoid over-decomposition, a stopping condition or a preset number of decomposition iterations can be set.

[0054] Then, noisy environmental factor data is acquired, and noise is added to simulate the noisy signal. This can be achieved by adding random noise, such as uniform noise, to the original environmental factor data signal. Before performing empirical mode decomposition, the noisy signal can be preprocessed, such as by filtering, detrending, and removing the mean, to remove components that affect the decomposition results.

[0055] If the current decomposition count has not reached the preset decomposition count, return to the step of obtaining the noisy environmental factor data corresponding to the environmental factor data, and perform the decomposition operation on the noisy environmental factor data based on the empirical mode decomposition algorithm until the preset number of intrinsic mode function components are obtained.

[0056] Finally, the intrinsic mode functions (IMFs) are determined based on the preset number of IMF components. For example, the IMFs are determined based on the preset number of IMF components using the following formula: in, Represents the intrinsic mode functions. Indicates the preset number of decompositions.

[0057] S205: The target data is transmitted to the virtual model layer using the data interaction layer to drive the RH inversion model based on the hybrid algorithm to iteratively update and generate RH spatiotemporal distribution data.

[0058] Specifically, the RH inversion model based on the hybrid algorithm can include: The system comprises a cascaded data preprocessing module, a mechanism modeling module, a machine learning module, a parameter fusion module, and a dynamic update module. The data preprocessing module is used to receive target data and perform preprocessing on the target data; The mechanism model module is used to construct the correlation between RH and various environmental factor data using a wetland soil microbial respiration mechanism model and pre-processed target data. The machine learning module is used to mine the nonlinear coupling relationships between various environmental factor data based on a hybrid algorithm. The parameter fusion module is used to fuse the correlation and nonlinear coupling relationships of various environmental factor data to generate monitoring data; The dynamic update module is used to receive and use the monitoring data to update the training parameters of the wetland soil microbial respiration mechanism model, so as to realize the dynamic adaptation of the RH inversion model based on the hybrid algorithm to changes in the wetland environment.

[0059] For example, the construction of the RH inversion model based on the hybrid algorithm first builds the model framework in a cascade structure, and integrates the data preprocessing module, the mechanism model module, the machine learning module, the parameter fusion module and the dynamic update module in sequence. Each module realizes information flow through the data interaction interface.

[0060] Firstly, through the data interaction layer of the digital twin system, the collected data... Target data such as emission flux, water level, soil temperature, soil moisture content, salinity, redox potential, and air temperature are transmitted to the virtual model layer in real time. Then, the data preprocessing module performs standardization processing on the target data to eliminate dimensional differences. At the same time, the Kalman filter algorithm is used to remove outliers and fill in missing values ​​to ensure data quality.

[0061] Based on the respiratory mechanism of wetland soil microorganisms, a linear correlation model between RH and pretreated target data is established, which can be expressed by the following formula: This represents the RH estimate output by the mechanistic model. These represent the initial values ​​of the mechanistic model coefficients determined based on the thermodynamic characteristics of wetland soil respiration and the kinetics of substrate decomposition. This represents the standardized parameters of soil temperature. This represents the standardized parameter of soil moisture content. Indicates the water level standardization parameter. This represents the salinity standardization parameter. This represents the normalized parameter of redox potential.

[0062] A hybrid algorithm combining Random Forest (RF), XGBoost, and LSTM is employed, using preprocessed target data as input to uncover nonlinear coupling features between factors and output RH prediction values. .

[0063] For example, firstly, ensemble learning is performed using multiple decision trees, as shown in the formula: in, For the number of decision trees, Let be the prediction function of the m-th decision tree. For decision tree parameters, To standardize the target data set.

[0064] Then, the prediction accuracy is optimized based on a gradient boosting strategy, as shown in the formula. in, For the number of classifiers, Let t be the output of the t-th classifier.

[0065] Finally, the dynamic correlation of time-series data is captured, the state is updated through the gating unit, and the time-series prediction value is output. .

[0066] The predicted RH value was obtained by weighted fusion. Represented as: in, , , The sum is 1, and the weights are determined after cross-validation.

[0067] It should be noted that the linear correlation result of the fusion mechanism model and the nonlinear coupling result of the machine learning module are expressed as follows: in, For the merged monitoring data, To fuse weights, the validation set error is minimized. Sure.

[0068] In one alternative implementation, the dynamic update module receives the fused monitoring data. Combined with the actual RH value measured offline Calculation error .

[0069] The mechanistic model coefficients and machine learning model parameters are updated based on error feedback, where the formula for updating the mechanistic model coefficients is: in, The machine learning model updates the decision tree split threshold, gradient boosting step size, and LSTM network weights using the backpropagation algorithm, with the learning rate as the threshold. Repeating these steps, the model iterates using new target data transmitted in real-time from the data interaction layer at a preset synchronization frequency, until the model's prediction error converges, generating a RH inversion target model based on a hybrid algorithm. Finally, the RH inversion target model based on the hybrid algorithm calculates the RH value of the wetland to be evaluated, generating RH spatiotemporal distribution data.

[0070] S206: Based on the NPP spatial distribution map and the RH spatiotemporal distribution data, generate a wetland net ecosystem productivity spatial distribution map, sensitivity analysis map, community impact map, and annual change map for the wetland ecosystem to be evaluated.

[0071] For example, based on the NPP spatial distribution map and RH spatiotemporal distribution data obtained above, and through the core accounting relationship NEP=NPP-RH, combined with wetland zoning characteristics, vegetation community types and annual dynamic monitoring data, the system generates a wetland net ecosystem productivity spatial distribution map, sensitivity analysis map, community impact map and annual change map.

[0072] First, spatiotemporal consistency calibration was performed on the NPP spatial distribution map and the RH spatial distribution data. The NPP data was resampled according to the RH grid resolution to ensure a one-to-one correspondence between the grid cells of the two; at the same time, missing grids in the NPP or RH data were removed to avoid invalid calculations.

[0073] Based on the calculation formula NEP=NPP-RH, quantitative calculations are performed on each grid cell, and the units of NPP and RH are unified. In the calculation results, NEP>0 indicates that the region is a carbon sink, NEP<0 indicates that it is a carbon source, and NEP=0 indicates that the carbon budget is balanced.

[0074] In one optional implementation, the natural breakpoint method can be used to classify the NEP calculation results into seven levels: strong carbon sink area, medium carbon sink area, weak carbon sink area, carbon balance area, weak carbon source area, medium carbon source area, and strong carbon source area. GIS software is used for color rendering, such as using green for carbon sink areas, red for carbon source areas, and yellow for carbon balance areas. Basic layers such as wetland boundaries, hydrological zones, and vegetation type zones are overlaid to generate an NEP spatial distribution map, clearly showing the spatial distribution pattern of carbon sinks / carbon sources.

[0075] NEP sensitivity analysis maps are used to identify key driving factors influencing the dynamic changes of NEP, providing a targeted basis for wetland carbon sequestration regulation. First, core driving factors are selected based on wetland ecological characteristics, including NPP-related factors such as vegetation cover, chlorophyll content, and biomass; RH-related factors such as soil temperature, soil moisture content, water level, salinity, redox potential; and external environmental factors. Sensitivity coefficients can be calculated using the controlled variable method. With other factors remaining constant, a single driving factor is varied by ±10%, and the rate of change in NEP response is calculated, i.e., the sensitivity coefficient. ,in The magnitude of the change in the driving factor. This represents the corresponding change in NEP. The larger the absolute value, the more significant the effect of the factor on NEP; Positive values ​​indicate a positive correlation between the factor and NEP, while negative values ​​indicate a negative correlation. Simultaneously, using the NEP spatial distribution map as a base map, the sensitivity coefficients of key driving factors in each grid cell can be spatially expressed to generate a single-factor sensitivity distribution map. Furthermore, average sensitivity coefficients for each partition can be summarized using radar charts or heatmaps to generate a comprehensive sensitivity analysis map, clearly identifying the dominant driving factors in different partitions.

[0076] Community impact maps focus on the differences in the contribution of different vegetation communities to NEP (Neural Energy Emission), providing a reference for wetland vegetation restoration and carbon sequestration optimization. Firstly, based on vegetation type zoning data, the spatial distribution maps of NEP, NPP, and RH are spatially aggregated by community type, such as reed-Scirpus triqueter pastoral communities, Spartina alterniflora invasive communities, Salicyces cerevisiae-Suaeda salsa extreme habitat communities, and areas without vegetation cover. The average NEP, average NPP, and average RH values ​​for each community type are calculated. The NEP contribution of each vegetation community is calculated to analyze the differences in carbon sequestration capacity among different communities. Simultaneously, the matching relationship between NPP and RH within the same community is compared to identify community types with advantageous or disadvantageous carbon sequestration capacity.

[0077] The annual variation map reflects the long-term dynamic trend of NEP, supporting the sustainable monitoring and assessment of wetland carbon sinks. For example, based on 3-5 years of continuous hyperspectral image data, simulation box monitoring data and environmental baseline data, the spatial distribution data of NEP is calculated year by year according to the aforementioned method, and an annual time series database of NEP is constructed to identify NEP growth areas, stable areas and decline areas. At the same time, the annual average NEP value and annual fluctuation range of the wetland as a whole are calculated, the long-term variation trend is analyzed, and the time series fluctuation characteristics of the overall carbon sink capacity of the wetland are displayed intuitively.

[0078] As can be seen, this invention first collects initial data on the wetland ecosystem to be evaluated; based on the initial data, a wetland net primary productivity (NPP) inversion model is constructed through three-dimensional spectral feature extraction and multi-model weighted fusion, generating an NPP spatial distribution map; a simulation chamber is obtained to simulate the wetland ecosystem to be evaluated, and a corresponding digital twin system is constructed based on the simulation chamber; target data of the wetland ecosystem to be evaluated is collected through the simulation chamber device layer; the target data is transmitted to the virtual model layer using the data interaction layer, driving the iterative update of the RH inversion model based on a hybrid algorithm to generate RH spatiotemporal distribution data; based on the NPP spatial distribution map and the RH spatiotemporal distribution data, a wetland net ecosystem productivity spatial distribution map, sensitivity analysis map, community impact map, and annual change map of the wetland ecosystem to be evaluated are generated. It achieves high-precision estimation of wetland ecosystem net ecosystem productivity (NEP), providing technical support for wetland ecosystem carbon cycle research and ecological protection.

[0079] Another embodiment of this application provides a system for assessing the net ecosystem productivity of wetlands, such as... Figure 3The diagram shows a framework for an assessment system of net ecosystem productivity in wetlands, the system comprising: The first acquisition module 301 is used to acquire initial data of the wetland ecosystem to be evaluated; wherein, the initial data includes hyperspectral image data, basic environmental data, and historical biomass dynamic data; The construction module 302 is used to construct a wetland net primary productivity (NPP) inversion model based on the initial data by extracting three-dimensional spectral features and weighting and fusing multiple models, and to generate an NPP spatial distribution map. The module 303 is used to obtain a simulation box for simulating the wetland ecosystem to be evaluated, and to construct a corresponding digital twin system based on the simulation box; wherein, the digital twin system includes a simulation box device layer, a virtual model layer and a data interaction layer; The second acquisition module 304 is used to acquire target data of the wetland ecosystem to be evaluated through the simulation box device layer; wherein, the target data includes Emission flux data and environmental factor data; The transmission module 305 is used to transmit the target data to the virtual model layer using the data interaction layer, and drive the RH inversion model based on the hybrid algorithm to iteratively update in order to generate RH spatiotemporal distribution data. The generation module 306 is used to generate a spatial distribution map of net ecosystem productivity (NPP), a sensitivity analysis map, a community impact map, and an annual change map of the wetland ecosystem to be evaluated, based on the NPP spatial distribution map and the RH spatiotemporal distribution data.

[0080] Compared with existing technologies, this invention first collects initial data on the wetland ecosystem to be evaluated; based on the initial data, a wetland net primary productivity (NPP) inversion model is constructed through three-dimensional spectral feature extraction and multi-model weighted fusion, generating an NPP spatial distribution map; a simulation chamber is obtained to simulate the wetland ecosystem to be evaluated, and a corresponding digital twin system is constructed based on the simulation chamber; target data of the wetland ecosystem to be evaluated is collected through the simulation chamber device layer; the target data is transmitted to the virtual model layer using the data interaction layer, driving the iterative update of the RH inversion model based on a hybrid algorithm to generate RH spatiotemporal distribution data; based on the NPP spatial distribution map and the RH spatiotemporal distribution data, a wetland net ecosystem productivity spatial distribution map, sensitivity analysis map, community impact map, and annual change map of the wetland ecosystem to be evaluated are generated. It achieves high-precision estimation of wetland ecosystem net ecosystem productivity (NEP), providing technical support for wetland ecosystem carbon cycle research and ecological protection.

[0081] This invention also provides a storage medium storing a computer program, wherein the computer program is configured to implement the steps in the above method embodiments when running.

[0082] Specifically, in this embodiment, the storage medium can be configured to store a computer program for performing the following steps: S201: Collect initial data on the wetland ecosystem to be evaluated; wherein, the initial data includes hyperspectral image data, basic environmental data, and historical biomass dynamic data; S202: Based on the initial data, a wetland net primary productivity (NPP) inversion model is constructed by three-dimensional spectral feature extraction and multi-model weighted fusion, and an NPP spatial distribution map is generated. S203: Obtain a simulation box for simulating the wetland ecosystem to be evaluated, and construct a corresponding digital twin system based on the simulation box; wherein, the digital twin system includes a simulation box device layer, a virtual model layer, and a data interaction layer; S204: Collect target data of the wetland ecosystem to be evaluated through the simulation box device layer; wherein, the target data includes Emission flux data and environmental factor data; S205: The target data is transmitted to the virtual model layer using the data interaction layer to drive the RH inversion model based on the hybrid algorithm to iteratively update and generate RH spatiotemporal distribution data; S206: Based on the NPP spatial distribution map and the RH spatiotemporal distribution data, generate a wetland net ecosystem productivity spatial distribution map, sensitivity analysis map, community impact map, and annual change map for the wetland ecosystem to be evaluated.

[0083] Specifically, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0084] Compared with existing technologies, this invention first collects initial data on the wetland ecosystem to be evaluated; based on the initial data, a wetland net primary productivity (NPP) inversion model is constructed through three-dimensional spectral feature extraction and multi-model weighted fusion, generating an NPP spatial distribution map; a simulation chamber is obtained to simulate the wetland ecosystem to be evaluated, and a corresponding digital twin system is constructed based on the simulation chamber; target data of the wetland ecosystem to be evaluated is collected through the simulation chamber device layer; the target data is transmitted to the virtual model layer using the data interaction layer, driving the iterative update of the RH inversion model based on a hybrid algorithm to generate RH spatiotemporal distribution data; based on the NPP spatial distribution map and the RH spatiotemporal distribution data, a wetland net ecosystem productivity spatial distribution map, sensitivity analysis map, community impact map, and annual change map of the wetland ecosystem to be evaluated are generated. It achieves high-precision estimation of wetland ecosystem net ecosystem productivity (NEP), providing technical support for wetland ecosystem carbon cycle research and ecological protection.

[0085] This invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to perform the steps described in the method embodiments above.

[0086] Specifically, the aforementioned electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the aforementioned processor, and the input / output device is connected to the aforementioned processor.

[0087] Specifically, in this embodiment, the processor can be configured to perform the following steps via a computer program: S201: Collect initial data on the wetland ecosystem to be evaluated; wherein, the initial data includes hyperspectral image data, basic environmental data, and historical biomass dynamic data; S202: Based on the initial data, a wetland net primary productivity (NPP) inversion model is constructed by three-dimensional spectral feature extraction and multi-model weighted fusion, and an NPP spatial distribution map is generated. S203: Obtain a simulation box for simulating the wetland ecosystem to be evaluated, and construct a corresponding digital twin system based on the simulation box; wherein, the digital twin system includes a simulation box device layer, a virtual model layer, and a data interaction layer; S204: Collect target data of the wetland ecosystem to be evaluated through the simulation box device layer; wherein, the target data includes Emission flux data and environmental factor data; S205: The target data is transmitted to the virtual model layer using the data interaction layer to drive the RH inversion model based on the hybrid algorithm to iteratively update and generate RH spatiotemporal distribution data; S206: Based on the NPP spatial distribution map and the RH spatiotemporal distribution data, generate a wetland net ecosystem productivity spatial distribution map, sensitivity analysis map, community impact map, and annual change map for the wetland ecosystem to be evaluated.

[0088] Compared with existing technologies, this invention first collects initial data on the wetland ecosystem to be evaluated; based on the initial data, a wetland net primary productivity (NPP) inversion model is constructed through three-dimensional spectral feature extraction and multi-model weighted fusion, generating an NPP spatial distribution map; a simulation chamber is obtained to simulate the wetland ecosystem to be evaluated, and a corresponding digital twin system is constructed based on the simulation chamber; target data of the wetland ecosystem to be evaluated is collected through the simulation chamber device layer; the target data is transmitted to the virtual model layer using the data interaction layer, driving the iterative update of the RH inversion model based on a hybrid algorithm to generate RH spatiotemporal distribution data; based on the NPP spatial distribution map and the RH spatiotemporal distribution data, a wetland net ecosystem productivity spatial distribution map, sensitivity analysis map, community impact map, and annual change map of the wetland ecosystem to be evaluated are generated. It achieves high-precision estimation of wetland ecosystem net ecosystem productivity (NEP), providing technical support for wetland ecosystem carbon cycle research and ecological protection.

[0089] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0090] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0091] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical or other forms.

[0092] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0093] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0094] If the aforementioned integrated units are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0095] The embodiments of the present invention have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for assessing the net ecosystem productivity of wetlands, characterized in that, include: Initial data on the wetland ecosystem to be evaluated are collected; wherein, the initial data includes hyperspectral image data, basic environmental data, and historical biomass dynamic data; Based on the initial data, a wetland net primary productivity (NPP) inversion model is constructed by three-dimensional spectral feature extraction and multi-model weighted fusion, and an NPP spatial distribution map is generated. A simulation chamber for simulating the wetland ecosystem to be evaluated is obtained, and a corresponding digital twin system is constructed based on the simulation chamber; wherein, the digital twin system includes a simulation chamber device layer, a virtual model layer, and a data interaction layer; The target data of the wetland ecosystem to be evaluated are collected through the simulation chamber device layer; wherein, the target data includes Emission flux data and environmental factor data; The target data is transmitted to the virtual model layer using the data interaction layer, driving the RH inversion model based on the hybrid algorithm to iteratively update in order to generate RH spatiotemporal distribution data; Based on the NPP spatial distribution map and the RH spatiotemporal distribution data, a wetland net ecosystem productivity spatial distribution map, sensitivity analysis map, community impact map, and annual change map are generated for the wetland ecosystem to be evaluated.

2. The method according to claim 1, characterized in that, Based on the initial data, a wetland net primary productivity (NPP) inversion model is constructed through three-dimensional spectral feature extraction and multi-model weighted fusion, generating an NPP spatial distribution map, including: Radiometric correction, illumination correction, geometric correction, and image stitching are performed sequentially on the hyperspectral image data to determine the pixel spectral data. Extract three-dimensional spectral feature indicators from the pixel spectral data; wherein, the three-dimensional spectral feature indicators include vegetation index, sensitive narrowband features, and continuous spectral features; Based on the validation set composed of the three-dimensional spectral feature index, the environmental basic data, and the historical biomass dynamic data, a random forest model, an XGBoost model, and a PLSR model were constructed respectively. The root mean square error and the coefficient of determination of the validation set were used as the basis for weighted fusion to obtain the optimal inversion model of aboveground biomass AGB and the optimal inversion model of belowground biomass BGB. By combining data on plant growth stages, hydrological conditions, and typical plant structural parameters of the wetland ecosystem to be evaluated, the optimal AGB and BGB inversion models are optimized to construct a wetland net primary productivity (NPP) inversion model and generate an NPP spatial distribution map.

3. The method according to claim 2, characterized in that, The process of obtaining a simulation chamber for simulating the wetland ecosystem to be evaluated, and constructing a corresponding digital twin system based on the simulation chamber, includes: Simulation boxes were evenly distributed within the wetland ecosystem to be evaluated, based on the distribution characteristics of wetland hydrological zones, salinity gradients, and plant types. Each simulation box included an infrared sensor. Analyzer and / or Analyzer, multi-parameter sensing module, and wireless communication module; A digital twin parameter mapping model was constructed using the respiratory mechanism of wetland soil microorganisms. Based on the digital twin parameter mapping model, establish bidirectional communication between the simulation box device layer and the data interaction layer; An HMI visualization interface is constructed to generate a digital twin system for estimating the RH of the wetland ecosystem to be evaluated; wherein, the visualization interface includes a target data display module, a model iteration and adjustment module, and an RH result output function module.

4. The method according to claim 3, characterized in that, The collection of target data of the wetland ecosystem to be evaluated through the simulation box device includes: Through the built-in infrared of the simulation box The analyzer continuously monitors the changes in gas concentration inside the chamber at a first preset frequency, and calculates the concentration using the chamber method formula. Emission flux data; The simulation chamber uses a multi-parameter sensing module to synchronously collect environmental factor data at a second preset frequency; the environmental factor data includes water level, soil temperature, soil moisture content, salinity, and redox potential data. All collected Emission flux data and environmental factor data are transmitted to the data interaction layer in real time to obtain target data and are then transmitted synchronously to the virtual model layer.

5. The method according to claim 4, characterized in that, The collection of all Emission flux data and environmental factor data are transmitted in real time to the data interaction layer to obtain target data and are simultaneously transmitted to the virtual model layer, including: Perform a decomposition operation on the environmental factor data to obtain intrinsic mode functions; Based on the intrinsic mode function, noise reduction and data reconstruction operations are performed on the environmental factor data to obtain the noise-reduced environmental factor data. Based on the above The target data is generated from emission flux data and noise-reduced environmental factor data.

6. The method according to claim 5, characterized in that, The step of performing the decomposition operation on the environmental factor data to obtain the intrinsic mode functions includes: Set the preset number of decompositions for environmental factor data; Obtain the noisy environmental factor data corresponding to the environmental factor data, and perform the decomposition operation on the noisy environmental factor data based on the empirical mode decomposition algorithm to obtain the intrinsic mode function components; If the current decomposition count has not reached the preset decomposition count, return to the step of obtaining the noisy environmental factor data corresponding to the environmental factor data, and perform the decomposition operation on the noisy environmental factor data based on the empirical mode decomposition algorithm until the preset number of intrinsic mode function components are obtained. The intrinsic mode functions are determined based on the preset number of intrinsic mode function components.

7. The method according to claim 6, characterized in that, The step of performing the decomposition operation on the noisy environmental factor data based on the empirical mode decomposition algorithm to obtain intrinsic mode function components includes: The decomposition operation of the noisy environmental factor data is performed using the following formula: ， in, , , , This represents Gaussian white noise with a mean of 0. This represents the local mean of the noisy environmental factor data. The mathematical expectation operator, This represents the preset amplitude of the noise-containing environmental factor data. This represents the signal-to-noise ratio (SNR) of the i-th decomposition stage. This represents the eigenmode function component of the i-th decomposition stage. This represents the residual term in the i-th decomposition stage.

8. A system for assessing the net ecosystem productivity of wetlands, characterized in that, The system includes: The first acquisition module is used to acquire initial data of the wetland ecosystem to be evaluated; wherein, the initial data includes hyperspectral image data, basic environmental data, and historical biomass dynamic data; The construction module is used to construct a wetland net primary productivity (NPP) inversion model based on the initial data through three-dimensional spectral feature extraction and multi-model weighted fusion, and generate an NPP spatial distribution map. The module is used to obtain a simulation box for simulating the wetland ecosystem to be evaluated, and to construct a corresponding digital twin system based on the simulation box; wherein, the digital twin system includes a simulation box device layer, a virtual model layer and a data interaction layer; The second acquisition module is used to acquire target data of the wetland ecosystem to be evaluated through the simulation box device layer; wherein, the target data includes Emission flux data and environmental factor data; The transmission module is used to transmit the target data to the virtual model layer using the data interaction layer, and drive the RH inversion model based on the hybrid algorithm to iteratively update in order to generate RH spatiotemporal distribution data. The generation module is used to generate, based on the NPP spatial distribution map and the RH spatiotemporal distribution data, a wetland net ecosystem productivity spatial distribution map, a sensitivity analysis map, a community impact map, and an annual change map of the wetland ecosystem to be evaluated.

9. A storage medium, characterized in that, The storage medium stores a computer program, wherein the computer program is configured to implement the method of any one of claims 1 to 7 when it is run.

10. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to run the computer program to implement the method of any one of claims 1 to 7.

Images