A multi-source estimation method for stable isotopes of water vapor flux in terrestrial ecosystems

Abstract

The present invention discloses a multi-source estimation method for the stable isotopes of water vapor flux in a terrestrial ecosystem, which includes collecting monitoring data and environmental data of a preset area, and preprocessing the monitoring data and the environmental data; analyzing the water vapor movement characteristics of the monitoring data according to stable isotopes to obtain partition data, and inputting the partition data into a stable isotope model to obtain water vapor flux; analyzing the time-varying influence of the environmental data according to the water vapor flux to obtain influence factors, and constructing a multi-source estimation model for the water vapor flux in a terrestrial ecosystem according to the influence factors and the monitoring data; optimizing the multi-source estimation model for the water vapor flux in a terrestrial ecosystem, inputting the data to be estimated into the multi-source estimation model for the water vapor flux in a terrestrial ecosystem, and outputting an estimation result. This method can not only improve the accuracy of multi-source estimation of water vapor flux in a terrestrial ecosystem, but also has good interpretability and can be directly applied to a multi-source estimation system for water vapor flux.

Description

Technical Field

[0001] The present invention relates to the field of estimation, and particularly to a multi-source estimation method for stable isotopes of water vapor flux in terrestrial ecosystems. Background Art

[0002] In current terrestrial ecosystem research, the accurate estimation of water vapor flux is crucial for understanding the hydrological cycle, assessing the impact of climate change, and formulating water resource management strategies. Currently, there are various technical methods for estimating water vapor flux, and among them, the technology based on stable isotopes has attracted much attention due to its unique tracer ability. However, existing technical solutions mostly rely on single data sources or simplified models, making it difficult to comprehensively consider the influence of multi-source data fusion and dynamic changes in environmental factors on water vapor flux estimation.

[0003] Specifically, traditional methods usually estimate by combining data collected from a single monitoring site with empirical formulas. Although this method is simple and easy to implement, it is limited by the insufficient spatial representativeness of the data and the limitations of model assumptions, resulting in limited accuracy of the estimation results and difficulty in reflecting the complex water vapor exchange process in terrestrial ecosystems. In addition, existing technologies often ignore the time-varying influence of environmental factors on water vapor flux and fail to construct a multi-source estimation model that can dynamically adapt to environmental changes.

[0004] The present invention proposes a multi-source estimation method for stable isotopes of water vapor flux in terrestrial ecosystems, aiming to overcome the above defects of the existing technology and improve the accuracy and reliability of water vapor flux estimation. Summary of the Invention

[0005] The object of the present invention is to provide a multi-source estimation method for stable isotopes of water vapor flux in terrestrial ecosystems.

[0006] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0007] The present invention includes the following steps:

[0008] Collect monitoring data and environmental data of a preset area, and preprocess the monitoring data and the environmental data;

[0009] Analyze the water vapor movement characteristics of the monitoring data according to stable isotopes to obtain partition data, and input the partition data into a stable isotope model to obtain water vapor flux;

[0010] Analyze the time-varying influence of the environmental data according to the water vapor flux to obtain influence factors, and construct a multi-source estimation model for water vapor flux in terrestrial ecosystems based on the influence factors and the monitoring data;

[0011] Optimize the multi-source estimation model of the water vapor flux of the terrestrial ecosystem, input the data to be estimated into the multi-source estimation model of the water vapor flux of the terrestrial ecosystem, and output the estimation result.

[0012] Furthermore, a method for analyzing the water vapor movement characteristics of the monitoring data according to stable isotopes to obtain partition data includes:

[0013] Locate the position information according to the stable isotopes of the monitoring data, and sort the position information according to time sequence to obtain time-position data;

[0014] Establish a starting point of movement. Taking the starting point of movement as the center, obtain the maximum and minimum distances of movement according to the movement ability characterized by evapotranspiration. The expression is:

[0015]

[0016] Among them, the horizontal coordinate point of the movement starting point at the th moment is The th moment vertical coordinate point of the movement starting point is The th moment horizontal direction time-position data is The maximum value of the horizontal evapotranspiration velocity is The th moment and the th moment interval time is The coordinate position of the time-position data is (g, d), and the maximum value of the vertical evapotranspiration velocity is The th moment time-position data horizontal direction coordinate position is The th moment time-position data vertical direction coordinate position is

[0017] Calculate the distance vector:

[0018]

[0019] Among them, the distance vector of the time-position data is Calculate the squared regularization distance:

[0020]

[0021] Among them, the covariance matrix in the horizontal direction is The covariance matrix in the vertical direction is The transpose of the matrix is T, and the squared regularization distance of the time-position data is

[0022] Given a distance threshold, select the nearest time-position data and associate it with the movement trajectory. When the squared regularization distance is less than the distance threshold, the associated time-position data conforms to the water vapor movement trajectory; otherwise, replace the time-position data.

[0023] Calculate the decision value:

[0024]

[0025] Where the number of associated positioning points is q, the duration of the water vapor movement trajectory is S, and the decision value is D;

[0026] When the decision value is greater than 0.45, establish a reasonable water vapor movement trajectory. If both the squared regularization distance and the decision value do not meet the requirements, move the time window backward until a reasonable water vapor movement trajectory is established.

[0027] Classify the monitoring data according to the water vapor movement trajectory using similarity, and output the classified monitoring data source and the corresponding water vapor movement trajectory as partitioned data.

[0028] Furthermore, a method for obtaining the water vapor flux by inputting the partitioned data into a stable isotope model includes:

[0029] The stable isotope model obtains the change in the stable isotope ratio of the target according to the partitioned data, and calculates the water vapor flux in the vertical direction based on the ratio change:

[0030]

[0031] Where the water vapor flux in the vertical direction is The net radiation at the vegetation surface is H, the soil heat flux is Q, the slope of the vapor pressure curve is γ, the average temperature is M, the wind speed is C, the psychrometer constant is χ, and the saturation vapor pressure is a o , the actual vapor pressure is a, and the soil moisture content function is The vegetation coefficient is The soil moisture stress coefficient is The vegetation quantity is Q, the Bowen ratio is υ, and the first improvement coefficient is The second improvement coefficient is The slope of the saturation vapor pressure change with temperature is w, the total net radiation is F, the latent heat of vaporization is μ, and the conversion coefficient is The soil coefficient is

[0032] Calculate the water vapor flux in the horizontal direction:

[0033]

[0034] Where the water vapor flux in the horizontal direction is The air density is The specific humidity is The mass of water vapor is m, and the condensation rate is The area of water vapor is S.

[0035] Furthermore, a method for analyzing the time-varying impact of environmental data based on the water vapor flux to obtain an impact factor includes:

[0036] Sort the environmental data according to the time sequence and calculate the cumulative membership degrees at different times:

[0037]

[0038] Among them, the cumulative membership degrees of the environmental data at the b-th moment and the z-th moment are The i-th type of environmental data at the b-th moment is h i (b), and the i-th type of environmental data at the z-th moment is h i (z), and the number of types of environmental data is The water vapor flux at the z-th moment is The water vapor flux at the z-th moment is

[0039] Calculate the importance:

[0040]

[0041] Among them, the importance of the c-th type of environment is k c , the number of environmental types is N, and the cumulative membership degree of the c-th type of environment is The control factor is β, and the water vapor flux of the c-th type of environment at the z-th moment is The water vapor flux of the c-th type of environment at the b-th moment is The environmental data h i (b) and the environmental data h i (z) have a correlation degree of ψ(h i (b),h i (z));

[0042] Take the environment with an importance greater than 0.732 as the cluster center and obtain the deviation threshold of the environmental data through a threshold regression model;

[0043] Calculate the similarity between the environmental data and the cluster center:

[0044]

[0045] Among them, the regulation coefficient is λ, the sign function is sgn(·), the fixed deviation constant is ε, and the deviation threshold of the i-th environmental data is ζ i , and the similarity between the c-th environmental data and the cluster center is k c ;

[0046] Classify the environmental data to the cluster center with the greatest similarity to obtain classification data, and calculate the influence factor according to the classification data:

[0047]

[0048] where the influence factor is The systematic error is σ, and the L2 norm is ‖·‖2.

[0049] Furthermore, a method for constructing a multi-source estimation model of water vapor flux in terrestrial ecosystems according to the influence factor and the monitoring data includes:

[0050]

[0051] where the objective function is The actual value of the water vapor flux is The predicted value of the water vapor flux is The actual value and the predicted value The loss function of is The influence factor is The water vapor flux in the horizontal direction is The water vapor flux in the vertical direction is

[0052] The multi-source estimation model of water vapor flux in terrestrial ecosystems includes a decision tree clustering algorithm, a long short-term memory network algorithm, and a neural network algorithm;

[0053] The decision tree clustering algorithm classifies the data by constructing a decision tree, and then divides the data points into different clusters according to the branches of the tree to obtain class data;

[0054] The long short-term memory network algorithm extracts the characteristics of class data according to long-term dependencies through a recursive structure and a gating mechanism to obtain water vapor evapotranspiration characteristics;

[0055] The neural network algorithm adjusts the water vapor flux in the horizontal direction and the water vapor flux in the vertical direction in the objective function according to the water vapor evapotranspiration characteristics of the historical input data, and performs multi-source estimation of the water vapor flux according to the detection data through the adjusted objective function.

[0056] Furthermore, a method for optimizing the multi-source estimation model of stable isotopes of water vapor flux includes:

[0057] The updated bias, the expression is:

[0058]

[0059]

[0060] where the bias of the t-th iteration is The bias for the (t - 1)-th iteration is The learning rate is α, and the gradient of the loss function for the (t - 1)-th iteration is ρ t-1 , and the first moment of the t-th iteration of the gradient is y t , and the second moment of the t-th iteration of the gradient is s t , and the momentum decay hyperparameter is The scaling decay hyperparameter is The smoothing term is σ, and the bias correction for the first moment is The bias correction for the second moment is The first moment of the (t - 1)-th iteration of the gradient is y t-1 , and the second moment of the (t - 1)-th iteration of the gradient is s t-1 ;

[0061] L2 regularization is introduced to update the loss function, and the expression is:

[0062]

[0063] where the loss function is The L2 regularization coefficient is η, the weight vector of the neural network is ω, the output dimension is U, and the number of input data is The predicted value of the x-th output dimension of the b-th input data is The actual value of the x-th output dimension of the b-th input data is Z b,x , and the mean squared error is

[0064] Batch normalization zero-centers the input of the hidden layer of the neural network and introduces scaling and offset parameters to correct the data distribution change. The expression is:

[0065]

[0066] where the estimated input standard deviation vector in the mini-batch v is l v , and the zero-centered and normalized processing vector of the x-th input data is f x , the scaling parameter is θ, the offset parameter is ξ, and the Kronecker product is The output of the x-th batch normalization is φ x , and the number of input data in the mini-batch is

[0067] Adaptive learning is performed according to the error loss, and the expression is:

[0068]

[0069] where the error loss is The hyperparameter is δ, and it is continuously iterated until the error loss is less than 0.217, otherwise the bias is updated.

[0070] In a second aspect, an embodiment of the present application further provides an electronic device, including:

[0071] a processor; and a memory arranged to store computer-executable instructions, which when executed cause the processor to execute the method steps described in the first aspect.

[0072] In a third aspect, an embodiment of the present application further provides a computer-readable storage medium storing one or more programs, which when executed by an electronic device including a plurality of application programs, cause the electronic device to execute the method steps described in the first aspect.

[0073] The beneficial effects of the present invention are as follows:

[0074] The present invention is a multi-source estimation method for stable isotopes of water vapor flux in terrestrial ecosystems. Compared with the prior art, the present invention has the following technical effects:

[0075] Through steps such as preprocessing, analysis of water vapor movement characteristics, calculation of water vapor flux, analysis of time-varying effects, model construction, and model optimization, the present invention can improve the accuracy of multi-source estimation of stable isotopes of water vapor flux in terrestrial ecosystems, thereby improving the precision of multi-source estimation of stable isotopes of water vapor flux in terrestrial ecosystems. Optimizing the multi-source estimation of stable isotopes of water vapor flux in terrestrial ecosystems can greatly save resources, improve work efficiency, can realize the automatic estimation of multi-source water vapor flux in terrestrial ecosystems, and perform data fusion and estimation optimization on the multi-source estimation of stable isotopes of water vapor flux in terrestrial ecosystems in real time, which is of great significance for the multi-source estimation of stable isotopes of water vapor flux in terrestrial ecosystems, and can adapt to multi-source estimation of stable isotopes of water vapor flux in terrestrial ecosystems with different standards and multi-source estimation requirements of stable isotopes of water vapor flux in different terrestrial ecosystems, and has a certain universality. Description of the Drawings

[0076] Figure 1 is a flowchart of the steps of a multi-source estimation method for stable isotopes of water vapor flux in a terrestrial ecosystem according to the present invention;

[0077] Figure 2 is a schematic structural diagram of an electronic device in an embodiment of this specification. Detailed Embodiments

[0078] The present invention will be further described below through specific embodiments. The illustrative embodiments and explanations of the present invention are used to explain the present invention, but do not limit the present invention.

[0079] A multi-source estimation method for stable isotopes of water vapor flux in a terrestrial ecosystem according to the present invention includes the following steps:

[0080] As Figure 1 shown, in this embodiment, the following steps are included:

[0081] Collect monitoring data and environmental data of a preset area, and preprocess the monitoring data and the environmental data;

[0082] In actual evaluation, the monitoring data includes rainfall, evaporation, air temperature and humidity, vegetation index, soil humidity and temperature, ecosystem type, vegetation transpiration, water vapor pressure, radiation, saturated water vapor pressure; the environmental data includes temperature, humidity, wind speed, wind direction, vegetation coverage, topography, soil type, precipitation, weather, evaporation, soil temperature and humidity, atmospheric circulation pattern;

[0083] Take XXX wetland as the research object, estimate the water vapor flux in spring 2020, and the actual water vapor flux in spring 2020 is 14.329 mm / d;

[0084] Analyze the water vapor movement characteristics of the monitoring data according to stable isotopes to obtain classification data, and input the classification data into a stable isotope model to obtain the water vapor flux;

[0085] In actual evaluation, the horizontal water vapor flux is 9.259 mm / d, and the vertical water vapor flux is 5.070 mm / d;

[0086] Analyze the time-varying influence of the environmental data according to the water vapor flux to obtain influence factors, and construct a multi-source estimation model of the water vapor flux of the terrestrial ecosystem according to the influence factors and the monitoring data;

[0087] In actual evaluation, the influence factor is 0.381;

[0088] Optimize the multi-source estimation model of the water vapor flux of the terrestrial ecosystem, input the data to be estimated into the multi-source estimation model of the water vapor flux of the terrestrial ecosystem, and output the estimation result;

[0089] In actual evaluation, the predicted water vapor flux is 14.319 mm / d.

[0090] In this embodiment, the method for analyzing the water vapor movement characteristics of the monitoring data according to stable isotopes to obtain classification data includes:

[0091] Locate the coordinates of the monitoring data according to stable isotopes to obtain position information, and sort the position information according to time sequence to obtain time-position data;

[0092] Establish a starting point of movement, take the starting point of movement as the center, and obtain the maximum and minimum distances of movement according to the movement ability characterized by evapotranspiration. The expression is:

[0093]

[0094] Among them, the horizontal coordinate point of the motion starting point at the moment is The vertical coordinate point of the motion starting point at the moment is The time position data in the horizontal direction at the moment is The maximum value of the evapotranspiration rate in the horizontal direction is The time interval between the moment and the moment is The coordinate position of the time position data is (g, d), and the maximum value of the evapotranspiration rate in the vertical direction is The horizontal coordinate position of the time position data at the moment is The vertical coordinate position of the time position data at the moment is

[0095] Calculate the distance vector:

[0096]

[0097] Among them, the distance vector of the time position data is Calculate the squared regularized distance:

[0098]

[0099] Among them, the covariance matrix in the horizontal direction is The covariance matrix in the vertical direction is The transpose of the matrix is T, and the squared regularized distance of the time position data is

[0100] Given a distance threshold, select the nearest time position data to associate with the water vapor movement trajectory. When the squared regularized distance is less than the distance threshold, the associated time position data conforms to the water vapor movement trajectory; otherwise, replace the time position data;

[0101] Calculate the decision value:

[0102]

[0103] Among them, the number of associated positioning points is q, the duration of the water vapor movement trajectory is S, and the decision value is D;

[0104] When the decision value is greater than 0.45, establish a reasonable water vapor movement trajectory. If both the squared regularized distance and the decision value do not meet the requirements, move the time window backward until a reasonable water vapor movement trajectory is established;

[0105] Classify the monitoring data according to the similarity based on the water vapor movement trajectory, and output the classified monitoring data source and the corresponding water vapor movement trajectory as partitioned data;

[0106] In the actual evaluation, the distance threshold is 0.921.

[0107] In this embodiment, the method for obtaining the water vapor flux by inputting the partitioned data into the stable isotope model includes:

[0108] The stable isotope model obtains the change in the stable isotope ratio of the target according to the partitioned data, and calculates the water vapor flux in the vertical direction according to the ratio change:

[0109]

[0110] where the water vapor flux in the vertical direction is The net radiation on the vegetation surface is H, the soil heat flux is Q, the slope of the vapor pressure curve is γ, the average temperature is M, the wind speed is C, the psychrometer constant is χ, and the saturation vapor pressure is a o , the actual vapor pressure is a, and the soil water content function is The vegetation coefficient is The soil water stress coefficient is The vegetation quantity is Q, the Bowen ratio is υ, and the first improvement coefficient is The second improvement coefficient is The slope of the saturation vapor pressure changing with temperature is w, the total net radiation is F, the latent heat of vaporization is μ, and the conversion coefficient is The soil coefficient is

[0111] Calculate the water vapor flux in the horizontal direction:

[0112]

[0113] where the water vapor flux in the horizontal direction is The air density is The specific humidity is The mass of water vapor is m, and the condensation rate is The water vapor area is S.

[0114] In this embodiment, the method for obtaining the impact factor by performing time-varying impact analysis on the environmental data according to the water vapor flux includes:

[0115] Sort the environmental data according to the time sequence, and calculate the cumulative membership degrees at different times:

[0116]

[0117] where the cumulative membership degrees of the environmental data at the b-th moment and the z-th moment are The i-th environmental data at time b is h i (b), and the i-th environmental data at time z is h i (z), and the number of types of environmental data is The water vapor flux at time z is The water vapor flux at time z is

[0118] Calculate the importance:

[0119]

[0120] Among them, the importance of the c-th environment is k c , the number of environmental types is N, and the cumulative membership degree of the c-th environment is The control factor is β, and the water vapor flux of the c-th environment at time z is The water vapor flux of the c-th environment at time b is Environmental data h i (b) and environmental data h i (z) has a correlation degree of ψ(h i (b), h i (z));

[0121] Take the environment with an importance greater than 0.732 as the cluster center, and obtain the deviation threshold of the environmental data through the threshold regression model;

[0122] Calculate the similarity between the environmental data and the cluster center:

[0123]

[0124] Among them, the regulation coefficient is λ, the sign function is sgn(·), the fixed deviation constant is ε, and the deviation threshold of the i-th environmental data is ζ i , and the similarity between the c-th environmental data and the cluster center is k c ;

[0125] Assign the environmental data to the cluster center with the maximum similarity to obtain the classification data, and calculate the influence factor according to the classification data:

[0126]

[0127] Among them, the influence factor is The systematic error is σ, and the L2 norm is ‖·‖2.

[0128] In this embodiment, the method for constructing a multi-source estimation model of the water vapor flux of the terrestrial ecosystem according to the influence factor and the monitoring data includes:

[0129]

[0130] where the objective function is The actual value of the water vapor flux is The predicted value of the water vapor flux is Actual value and predicted value The loss function of is The influencing factor is The water vapor flux in the horizontal direction is The water vapor flux in the vertical direction is

[0131] The multi-source estimation model of water vapor flux in terrestrial ecosystems includes decision tree clustering algorithm, long short-term memory network algorithm, and neural network algorithm;

[0132] The decision tree clustering algorithm classifies data by constructing a decision tree, and then divides data points into different clusters according to the branches of the tree to obtain categorical data;

[0133] The long short-term memory network algorithm extracts categorical data features according to long-term dependencies through a recursive structure and gating mechanism to obtain water vapor evapotranspiration features;

[0134] The neural network algorithm adjusts the water vapor flux in the horizontal direction and the water vapor flux in the vertical direction in the objective function according to the water vapor evapotranspiration features learned from historical input data, and performs multi-source estimation of water vapor flux according to the detected data through the adjusted objective function.

[0135] In this embodiment, the method for optimizing the multi-source estimation model of water vapor flux stable isotopes includes:

[0136] The updated bias, the expression is:

[0137]

[0138] where the bias of the t-th iteration is The bias of the (t-1)-th iteration is The learning rate is α, and the gradient of the loss function of the (t-1)-th iteration is ρ t-1 , the first moment of the t-th iteration of the gradient is y t , the second moment of the t-th iteration of the gradient is s t , the momentum decay hyperparameter is The scaling decay hyperparameter is The smoothing term is σ, and the bias correction of the first moment is The bias correction of the second moment is The first moment of the (t-1)-th iteration of the gradient is y t-1 , the second moment of the (t-1)-th iteration of the gradient is s t-1 ;

[0139] Introduce L2 regularization and update the loss function. The expression is:

[0140]

[0141] where the loss function is The L2 regularization coefficient is η, the weight vector of the neural network is ω, the output dimension is U, and the number of input data is The predicted value of the x-th output dimension of the b-th input data is The actual value of the x-th output dimension of the b-th input data is Z b,x , and the mean squared error is

[0142] Batch normalization zero-centers the input of the hidden layer of the neural network and introduces scaling parameters and offset parameters to correct the data distribution change. The expression is:

[0143]

[0144]

[0145] where the estimated input standard deviation vector in the mini-batch v is l v , the zero-centered and normalized processing vector of the x-th input data is f x , the scaling parameter is θ, the offset parameter is ξ, and the Kronecker product is The output of the x-th batch normalization is φ x , and the number of input data in the mini-batch is

[0146] Adaptive learning is performed according to the error loss. The expression is:

[0147]

[0148] where the error loss is The hyperparameter is δ, and it is continuously iterated until the error loss is less than 0.217, otherwise the bias is updated.

[0149] Figure 2 is a schematic structural diagram of an electronic device according to an embodiment of the present application. Please refer to Figure 2 , at the hardware level, the electronic device includes a processor, and optionally also includes an internal bus, a network interface, and a memory. Among them, the memory may include a memory, such as a high-speed random access memory (Random-Access Memory, RAM), and may also include a non-volatile memory, such as at least one disk memory, etc. Of course, the electronic device may also include other hardware required for other services.

[0150] The processor, network interface, and memory can be interconnected through an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of representation, Figure 2 only a single bidirectional arrow is used in Figure 2 , but it does not mean that there is only one bus or one type of bus.

[0151] Memory, which is used to store programs. Specifically, the program can include program code, and the program code includes computer operation instructions. The memory can include a memory and a non-volatile memory, and provide instructions and data to the processor.

[0152] The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it, forming a multi-source estimation device for the stable isotopes of water vapor flux in a terrestrial ecosystem at the logical level. The processor executes the program stored in the memory and is specifically used to execute any of the foregoing multi-source estimation methods for the stable isotopes of water vapor flux in a terrestrial ecosystem.

[0153] The above as in this application Figure 1A multi-source estimation method for stable isotopes of water vapor flux in a terrestrial ecosystem disclosed in the illustrated embodiment can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in the processor or the instructions in the form of software. The above-mentioned processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as being executed and completed by a hardware decoding processor, or executed and completed by a combination of the hardware and software modules in the decoding processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method.

[0154] The electronic device can also execute Figure 1 a multi-source estimation method for stable isotopes of water vapor flux in a terrestrial ecosystem, and implement Figure 1 the functions of the illustrated embodiment, which will not be elaborated in the embodiments of the present application.

[0155] The embodiments of the present application also propose a computer-readable storage medium that stores one or more programs. The one or more programs include instructions that, when executed by an electronic device including a plurality of application programs, execute any of the foregoing multi-source estimation methods for stable isotopes of water vapor flux in a terrestrial ecosystem.

[0156] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0157] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0158] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means realizes the functions specified in Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0159] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0160] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and a memory.

[0161] The memory may include non-permanent memory in the computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash memory (flash RAM). The memory is an example of a computer-readable medium.

[0162] A computer-readable medium includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transitory medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer-readable medium does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0163] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover a non-exclusive inclusion, such that a process, method, article or apparatus comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or apparatus. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or apparatus comprising the element.

[0164] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code.

[0165] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A multi-source estimation method for stable isotopes of water vapor flux in terrestrial ecosystems, characterized in that: The following steps are involved: Collecting monitoring data and environmental data of a preset area, and preprocessing the monitoring data and the environmental data; Performing water vapor movement characteristic analysis on the monitoring data according to stable isotopes to obtain partition data, and inputting the partition data into a stable isotope model to obtain water vapor flux; Conducting a time-varying impact analysis on environmental data based on the water vapor flux to obtain an impact factor, and constructing a multi-source estimation model for water vapor flux of a terrestrial ecosystem based on the impact factor and the monitoring data; Optimizing the multi-source estimation model of water vapor flux of terrestrial ecosystems, inputting the data to be estimated into the multi-source estimation model of water vapor flux of terrestrial ecosystems, and outputting the estimation result; The method for constructing a multi-source estimation model of water vapor flux of a terrestrial ecosystem based on the influencing factors and the monitoring data comprises: The objective function is The actual value of the water vapor flux is The predicted value of water vapor flux is Actual value and predicted values The loss function is The impact factor is The horizontal water vapor flux is The vertical water vapor flux is The multi-source estimation model of water vapor flux in terrestrial ecosystems includes decision tree clustering algorithm, long short-term memory network algorithm, and neural network algorithm; The decision tree clustering algorithm classifies data by building a decision tree, and then divides the data points into different clusters according to the branches of the tree to obtain category data; The long short-term memory network algorithm extracts the characteristics of category data according to long-term dependencies through a recursive structure and gating mechanism to obtain water vapor evapotranspiration characteristics; The neural network algorithm adjusts the horizontal and vertical water vapor fluxes in the objective function according to the water vapor evaporation characteristics of the historical input data, and estimates the multi-source water vapor flux according to the detection data through the adjusted objective function.

2. The multi-source estimation method of stable isotopes of water vapor flux in terrestrial ecosystems according to claim 1, characterized in that: The method of analyzing the water vapor movement characteristics of the monitoring data according to stable isotopes to obtain the divided data comprises: The monitoring data is coordinate-located according to the stable isotope to obtain the position information, and the position information is sorted according to the time sequence to obtain the time position data; Establish the starting point of the movement, take the starting point of the movement as the center, and obtain the maximum and minimum distance of the movement according to the movement capacity represented by evapotranspiration. The expression is: Among them The horizontal coordinate point of the starting point of the movement at the moment is No. The vertical coordinate point of the starting point of the movement at the moment is No. The time position data in the horizontal direction at the moment is The maximum horizontal evapotranspiration rate is No. Moment and The time interval is The coordinate position of the time position data is (g, d), and the maximum vertical evapotranspiration rate is No. The horizontal coordinate position of the time position data is No. The vertical coordinate position of the time position data is Calculate the distance vector: The distance vector of the time position data is Compute the squared regularized distance: The covariance matrix in the horizontal direction is The covariance matrix in the vertical direction is The transpose of the matrix is ​​T, and the squared normalized distance of the time position data is Given a distance threshold, select the nearest time position data to associate with the motion trajectory. When the square regularized distance is less than the distance threshold, the associated time position data conforms to the water vapor motion trajectory, otherwise the time position data is replaced. Calculate the decision value: The number of associated positioning points is q, the duration of the water vapor movement trajectory is S, and the decision value is D; When the decision value is greater than 0.45, a reasonable water vapor movement trajectory is established. If both the square regularized distance and the decision value are not satisfied, the time window is moved back until a reasonable water vapor movement trajectory is established. The monitoring data are classified according to the water vapor movement trajectory using similarity, and the classified monitoring data sources and the corresponding water vapor movement trajectories are output as partitioned data.

3. The multi-source estimation method of stable isotopes of water vapor flux in terrestrial ecosystems according to claim 1, characterized in that: The method of inputting the partitioned data into a stable isotope model to obtain water vapor flux comprises: The stable isotope model obtains the change in the stable isotope ratio of the target according to the partitioning data, and calculates the vertical water vapor flux according to the change in ratio: The vertical water vapor flux is The net radiation of the vegetation surface is H, the soil heat flux is Q, the slope of the vapor pressure curve is γ, the average temperature is M, the wind speed is C, the psychrometer constant is χ, and the saturated water vapor pressure is a o , the actual water vapor pressure is a, and the soil moisture function is The vegetation coefficient is The soil water stress coefficient is The amount of vegetation is Q, the Bowen ratio is υ, and the first improvement coefficient is The second improvement factor is The slope of the saturated water vapor pressure changing with temperature is w, the total net radiation is F, the latent heat of vaporization is μ, and the conversion coefficient is The soil coefficient is Calculate the horizontal water vapor flux: The horizontal water vapor flux is The air density is Specific humidity The mass of water vapor is m, and the condensation rate is The water vapor area is S.

4. The multi-source estimation method of stable isotopes of water vapor flux in terrestrial ecosystems according to claim 1, characterized in that: The method for obtaining the impact factor by performing time-varying impact analysis on environmental data according to the water vapor flux includes: Sort the environmental data in chronological order and calculate the cumulative membership at different times: The cumulative membership of the environmental data at time b and time z is The i-th environmental data at the b-th moment is h i (b), the i-th environmental data at the z-th moment is h i (z), the number of types of environmental data is The water vapor flux at time b is The water vapor flux at time z is Calculate importance: The importance of the cth environment is k c , the number of environmental types is N, and the cumulative membership of the cth environment is The control factor is β, and the water vapor flux of the cth environment at the zth time is The water vapor flux of the cth environment at time b is Environmental data i (b) and environmental data h i The correlation degree of (z) is ψ(h i (b), h i (z)); The environments with importance greater than 0.732 are taken as cluster centers, and the deviation threshold of the environmental data is obtained through the threshold regression model; Calculate the similarity between the environment data and the cluster center: The control coefficient is λ, the sign function is sgn(·), the fixed deviation constant is ε, and the deviation threshold of the i-th environmental data is ζ i , the similarity between the cth environmental data and the cluster center is k c ; The environmental data is assigned to the center of the cluster with the largest similarity to obtain the classified data, and the impact factor is calculated based on the classified data: The impact factor is The system error is σ and the L2 norm is ||·||2.

5. The multi-source estimation method of stable isotopes of water vapor flux in terrestrial ecosystems according to claim 1, characterized in that: Optimizing the multi-source estimation model of water vapor flux in terrestrial ecosystems, including: The updated bias is expressed as: The bias of the tth iteration is The bias for the t-1th iteration is The learning rate is α, and the gradient of the loss function at the t-1th iteration is ρ t-1 , the first-order moment of the gradient at the tth iteration is y t , the second-order moment of the gradient at the tth iteration is s t , the momentum decay hyperparameter is The scaling decay hyperparameter is The smoothing term is σ, and the bias correction of the first-order moment is The bias correction of the second-order moment is The first-order moment of the gradient at the t-1th iteration is y t-1 , the second-order moment of the gradient at the t-1th iteration is st-1; Introduce L2 regularization and update the loss function, the expression is: The loss function is The L2 regularization coefficient is η, the weight vector of the neural network is ω, the output dimension is U, and the number of input data is The predicted value of the xth output dimension of the bth input data is The actual value of the xth output dimension of the bth input data is Z b，x , the mean square error is Batch normalization zero-centers the input of the hidden layer of the neural network and introduces scaling parameters and offset parameters to correct data distribution changes. The expression is: The estimated input standard deviation vector in the mini-batch v is The zero-centered and normalized vector of the xth input data is f x , the scaling parameter is θ, the offset parameter is ξ, and the Kronecker product is The output of the xth batch normalization is φ x , the number of input data in a mini-batch is Adaptive learning is performed based on the error loss, and the expression is: The error loss is The hyperparameter is δ, and the iteration is continued until the error loss is less than 0.217, otherwise the bias is updated.

6. An electronic device comprising: processor; as well as A memory arranged to store computer executable instructions, which, when executed, cause the processor to perform the method according to any one of claims 1 to 5.

7. A computer-readable storage medium storing one or more programs, which, when executed by an electronic device including a plurality of application programs, enables the electronic device to execute the method according to any one of claims 1 to 5.

Images