Irrigation index prediction method based on meteorological data and field water holding capacity

Abstract

This invention discloses an irrigation index prediction method based on meteorological data and field capacity, belonging to the field of smart agriculture technology. The method includes collecting real-time meteorological data, real-time soil volumetric water content data, soil matrix potential data, and future precipitation forecast data, and performing quality control and time alignment processing to obtain a multi-source dataset. Based on the real-time meteorological data, a reference crop evapotranspiration is calculated, and combined with the crop coefficient, the actual water requirement of the crop is obtained. Based on the real-time soil volumetric water content data and the static parameter value of field capacity, a water stress coefficient is calculated, and combined with future precipitation forecast data, a basic irrigation demand index is obtained. This invention uses the Shuttleworth-Wallace model to separate evapotranspiration and, based on multi-period rolling water balance calculation and risk decision rules, generates an irrigation index that integrates crop evapotranspiration, water deficit, and effective precipitation, achieving online calibration of soil parameters and optimization of irrigation decisions.

Description

Technical Field

[0001] This invention relates to the field of smart agriculture technology, and in particular to a method for predicting irrigation index based on meteorological data and field water holding capacity. Background Technology

[0002] The core of precision agricultural irrigation management lies in the accurate assessment and prediction of the actual water requirements of crops. Based on the Penman-Monteith equation recommended by FAO, reference crop evapotranspiration is calculated, and irrigation strategies are formulated by combining crop coefficients and soil moisture sensor data. By deploying field weather stations and soil moisture sensors, real-time data on temperature, humidity, wind speed, solar radiation, and soil volumetric water content are obtained, and then the irrigation index is calculated. This fully considers meteorological factors and the current soil moisture status, and achieves a preliminary quantification of crop water requirements, providing important data support for water-saving irrigation, and has been widely used in practice.

[0003] Existing methods generally treat field capacity as a static parameter that remains fixed after laboratory measurement. Field capacity is affected by factors such as soil structure, organic matter content, and tillage practices, and changes slowly over time. Using a static parameter will cause the calculation benchmark of soil moisture deficit to gradually deviate from the actual field conditions, introducing imperceptible systematic errors. In the long run, this will affect the accuracy of irrigation decisions. Although precipitation forecasts have been introduced, there is a lack of quantitative assessment and proactive response mechanisms for the risks that future weather uncertainties bring to irrigation decisions, making it difficult to make an optimal trade-off between making full use of precipitation and avoiding drought. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an irrigation index prediction method based on meteorological data and field water holding capacity, which solves the benchmark drift problem and insufficient meteorological risk response problem caused by static parameters in the prior art.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides an irrigation index prediction method based on meteorological data and field water holding capacity, which includes collecting real-time meteorological data, real-time soil volumetric water content data, soil matrix potential data and future precipitation forecast data, and performing quality control and time alignment processing to obtain a multi-source dataset.

[0008] The reference crop evapotranspiration is calculated based on real-time meteorological data, and the actual crop water requirement is obtained by combining the crop coefficient. The water stress coefficient is calculated based on real-time soil volumetric water content data and field water holding capacity static parameter values. The basic irrigation demand index is obtained by combining future precipitation forecast data.

[0009] Based on real-time soil volumetric water content data and soil matrix potential data, soil moisture characteristic curves are fitted, and field water holding capacity dynamic values ​​are calculated by inversion.

[0010] Based on the dynamic values ​​of field water holding capacity and soil volumetric water content data, the water deficit is calculated. The vegetation index is calculated by acquiring multispectral images of farmland through remote sensing equipment. Combined with real-time meteorological data, the actual water requirement of crops is separated into soil evaporation and crop transpiration.

[0011] An irrigation index is generated by combining crop transpiration, water deficit, and effective precipitation from future precipitation forecast data.

[0012] As a preferred embodiment of the irrigation index prediction method based on meteorological data and field water holding capacity described in this invention, the method includes: collecting real-time meteorological data, real-time soil volumetric water content data, soil matric potential data, and future precipitation forecast data, and performing quality control and time alignment processing to obtain a multi-source dataset, comprising the following steps.

[0013] By using real-time meteorological data from field weather stations, real-time soil volumetric water content data, and soil matrix potential data, future precipitation forecast data can be obtained through API interfaces.

[0014] Real-time meteorological data, real-time soil volumetric water content data, soil matrix potential data, and future precipitation forecast data are input into the quality control module. The Dixon test is used to identify and remove outliers to obtain a clean real-time data sequence. The clean real-time data sequence is then input into the data repair module. For missing points caused by the removal of outliers, time series linear interpolation is used to fill in the missing points to obtain a complete real-time data sequence.

[0015] The complete real-time data sequence and the future precipitation forecast data are input into the time alignment module and unified to the UTC hourly timestamp to obtain a preliminary aligned multi-source data sequence;

[0016] The future precipitation forecast data from the initially aligned multi-source data sequence is input into the hydrological calculation module, and the effective precipitation is obtained by subtracting canopy interception and surface runoff loss from soil texture parameters.

[0017] The effective precipitation data is merged and packaged with the complete real-time data sequence to obtain a multi-source dataset.

[0018] As a preferred embodiment of the irrigation index prediction method based on meteorological data and field water holding capacity described in this invention, the method includes the following steps: calculating reference crop evapotranspiration based on real-time meteorological data and combining it with crop coefficients to obtain the actual water requirement of the crop.

[0019] Based on real-time meteorological data, the reference crop evapotranspiration was calculated using the Penman-Monteith formula.

[0020] Using historical meteorological data, soil volumetric water content data, crop type and growth stage data, and actual irrigation effect data as training sets, a gradient boosting decision tree algorithm is used for supervised learning to predict the optimal dynamic crop coefficient and build a machine learning model for the crop coefficient.

[0021] By inputting crop type, growth stage, and historical yield data into a machine learning model for crop coefficients, dynamic crop coefficients are obtained.

[0022] Based on the reference crop evapotranspiration and dynamic crop coefficient, the actual water requirement of the crop is obtained by multiplication.

[0023] As a preferred embodiment of the irrigation index prediction method based on meteorological data and field capacity described in this invention, the method includes the following steps: calculating the water stress coefficient based on real-time soil volumetric water content data and static field capacity parameter values, and obtaining the basic irrigation demand index by combining future precipitation forecast data.

[0024] The static parameter value of field water holding capacity was obtained by measuring soil samples from farmland using the ring cutter method.

[0025] Substitute real-time soil volumetric water content data and field water holding capacity static parameter values ​​into the water stress function to calculate the water stress coefficient.

[0026] Based on the actual water requirement of crops and the water stress coefficient, the basic irrigation demand index is obtained by using the coefficient product method.

[0027] As a preferred embodiment of the irrigation index prediction method based on meteorological data and field capacity described in this invention, the method includes the following steps: fitting a soil moisture characteristic curve based on real-time soil volumetric water content data and soil matric potential data, and inverting to calculate the dynamic value of field capacity.

[0028] The real-time collected soil volumetric water content and soil matrix potential data are subjected to physical rule verification and outlier removal to obtain a soil hydrodynamic dataset.

[0029] The soil hydrodynamic dataset is input into the physical information neural network. Through the fitting error and physical law constraint terms in the physical information neural network, the trained neural network model of soil hydraulic properties is output.

[0030] The standard soil matrix potential value of field water holding capacity in soil science is input into a trained neural network model of soil hydraulic properties, and the dynamic value of field water holding capacity is generated through forward propagation calculation of the network.

[0031] As a preferred embodiment of the irrigation index prediction method based on meteorological data and field capacity described in this invention, the method includes the following steps: calculating the water deficit based on the dynamic value of field capacity and soil volumetric water content data.

[0032] The relative soil moisture content is obtained by dividing the dynamic value of field water holding capacity and calculating the ratio.

[0033] The water deficit is calculated by combining the current relative soil moisture content and crop root depth data and using the soil water deficit calculation formula.

[0034] As a preferred embodiment of the irrigation index prediction method based on meteorological data and field water holding capacity described in this invention, the method includes: acquiring multispectral images of farmland using remote sensing equipment to calculate vegetation indices; combining this with real-time meteorological data to separate the actual water requirement of crops into soil evaporation and crop transpiration, comprising the following steps.

[0035] By acquiring multispectral images of farmland through satellite remote sensing platforms and performing radiometric calibration and atmospheric correction, clean remote sensing image data is obtained.

[0036] Using clean remote sensing image data, the reflectance of the near-infrared and red bands was extracted and substituted into the normalized differential vegetation index calculation formula to obtain the normalized vegetation index. The crop canopy structure was measured in the field and the data was standardized to obtain the field measured biophysical parameters.

[0037] A statistical-physical hybrid prediction model was established using normalized vegetation index and field-measured biophysical parameters, employing a nonlinear regression method.

[0038] The normalized vegetation index was input into the statistical-physical hybrid prediction model to obtain the leaf area index and canopy surface resistance.

[0039] The leaf area index, canopy surface resistance, and real-time meteorological data obtained from the inversion were input into the Shuttleworth-Wallace model. By solving the energy balance equation, the data were separated into soil evaporation and crop transpiration.

[0040] As a preferred embodiment of the irrigation index prediction method based on meteorological data and field water holding capacity described in this invention, the method involves synthesizing crop transpiration, water deficit, and effective precipitation from future precipitation forecast data to generate an irrigation index, including the following steps:

[0041] The soil moisture deficit sequence is predicted by using a recursive calculation method based on crop transpiration, effective precipitation and current water deficit.

[0042] By analyzing and predicting the soil moisture deficit sequence, the minimum, maximum and trend of change are extracted to obtain characteristic indicators that characterize the risk of future water stress.

[0043] Characteristic indicators that represent the risk of future water stress are input into the decision rule set to generate an irrigation index.

[0044] In a second aspect, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the irrigation index prediction method based on meteorological data and field water holding capacity as described in the first aspect of the present invention.

[0045] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the irrigation index prediction method based on meteorological data and field water holding capacity as described in the first aspect of the present invention.

[0046] The beneficial effects of this invention are as follows: By dynamically calibrating field water holding capacity parameters and quantifying weather forecast uncertainties, it solves the benchmark drift problem and insufficient meteorological risk response caused by static parameters in existing technologies. It collects multi-source data and performs quality control alignment, calculates crop water requirements and basic irrigation indices, dynamically fits soil moisture characteristic curves through physical information neural networks, and inverts the dynamic value of field water holding capacity, thus solving the static parameter drift problem. Combined with remote sensing inversion of vegetation indices, it uses the Shuttleworth-Wallace model to separate evapotranspiration, and generates an irrigation index that integrates crop evapotranspiration, water deficit, and effective precipitation based on multi-period rolling water balance calculation and risk decision rules. This realizes online calibration of soil parameters and optimization of irrigation decisions. Attached Figure Description

[0047] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a flowchart of an irrigation index prediction method based on meteorological data and field water holding capacity.

[0049] Figure 2 This is a flowchart showing the dynamic values ​​of field water holding capacity.

[0050] Figure 3 This is a flowchart for multi-source data acquisition.

[0051] Figure 4Flowchart of irrigation index Detailed Implementation

[0052] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0053] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0054] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0055] Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for predicting irrigation index based on meteorological data and field water holding capacity, including the following steps:

[0056] S1. Collect real-time meteorological data, real-time soil volumetric water content data, soil matrix potential data, and future precipitation forecast data, and perform quality control and time alignment processing to obtain a multi-source dataset.

[0057] S1.1. Obtain future precipitation forecast data through API interface by using real-time meteorological data, real-time soil volumetric water content data, and soil matrix potential data from field weather stations.

[0058] Furthermore, by deploying field weather stations in farmland, temperature, relative humidity, wind speed, and sunshine duration are continuously measured and output as real-time meteorological data. At the same time, soil volumetric water content data and soil matrix potential data are collected by frequency domain reflectance soil moisture sensors and tensiometers buried in the main root layer of crops, respectively. Through application programming interfaces, gridded numerical weather prediction products released by the European Centre for Medium-Range Weather Forecasts are accessed to obtain precipitation forecast data for three hours, for example, with a time resolution of 72 hours.

[0059] S1.2 Input real-time meteorological data, real-time soil volumetric water content data, soil matrix potential data, and future precipitation forecast data into the quality control module. Use the Dixon test to identify and remove outliers to obtain a clean real-time data sequence. Input the clean real-time data sequence into the data repair module. Use time series linear interpolation to fill in the missing points caused by the removal of outliers to obtain a complete real-time data sequence.

[0060] Furthermore, the real-time meteorological data collected by field weather stations, the real-time soil volumetric water content data collected by frequency domain reflectance soil moisture sensors, the soil matrix potential data collected by tensiometers, and the future precipitation forecast data obtained through application programming interfaces are used to identify and remove outliers using the Dixon test. The Dixon test uses the ratio of the range of the data sequence to the suspicious values ​​and compares it with a critical value table to determine and remove outlier data points that exceed the allowable range, forming a clean real-time data sequence. Subsequently, for the missing points in the clean real-time data sequence caused by outlier removal, time series linear interpolation is used to linearly fit the adjacent valid data points before and after the missing points to complete the missing values, generating a complete real-time data sequence.

[0061] S1.3. The complete real-time data sequence and the future precipitation forecast data are input into the time alignment module and unified to the UTC hourly timestamp to obtain a preliminary aligned multi-source data sequence.

[0062] Furthermore, the complete real-time data sequence and future precipitation forecast data are time-resampled using linear interpolation based on the Coordinated Universal Time (UTC) hourly timestamps. The complete real-time data sequence includes real-time meteorological data collected from field weather stations, real-time soil volumetric water content data collected from frequency domain reflectance soil moisture sensors, and soil matrix potential data collected from tensiometers. Future precipitation forecast data is obtained through an application programming interface (API). The linear interpolation method obtains the corresponding values ​​at the hourly time based on the values ​​of adjacent timestamps of data points, making the timestamps of all data sequences completely consistent, forming a preliminary aligned multi-source data sequence.

[0063] S1.4 Input the future precipitation forecast data from the initially aligned multi-source data sequence into the hydrological calculation module, and combine it with soil texture parameters to deduct canopy interception and surface runoff loss to obtain the effective precipitation.

[0064] The expression for effective precipitation is:

[0065]

[0066] Among them, P e ff represents effective precipitation, P f orecast is the forecast value for future precipitation, I a S represents the initial interception and depression filling losses, and S represents the maximum potential retention capacity of the watershed at that time.

[0067] S1.5. Merge and encapsulate the effective precipitation data with the complete real-time data sequence to obtain a multi-source dataset.

[0068] Furthermore, the effective precipitation data obtained through the hydrological calculation module is structurally integrated with the complete real-time data sequence after time alignment processing. The complete real-time data sequence includes real-time meteorological data collected by field weather stations, real-time soil volumetric water content data collected by frequency domain reflectance soil moisture sensors, and soil matrix potential data collected by tensiometers. The integration process uses the Coordinated Universal Time (UTC) timestamp as the index key value, combining the effective precipitation data with the real-time meteorological data, real-time soil volumetric water content data, and soil matrix potential data at the same timestamp into a unified structured data record, forming a multi-source dataset.

[0069] S2. Calculate the reference crop evapotranspiration based on real-time meteorological data, and combine it with the crop coefficient to obtain the actual water requirement of the crop.

[0070] S2.1 Based on real-time meteorological data, the reference crop evapotranspiration is calculated using the Penman-Monteith formula.

[0071] The formula for crop evapotranspiration is as follows:

[0072]

[0073] Where ET0 is the reference crop evapotranspiration, Δ is the slope of the saturated vapor pressure-temperature curve, and R... n Where γ is the net radiation from the crop canopy, G is the soil heat flux, γ is the psychrometric constant, T is the average air temperature, U2 is the wind speed at a height of 2 meters, and e is the mean air temperature. s e is the saturated vapor pressure. a This is the actual water vapor pressure.

[0074] S2.2 Using historical meteorological data, soil volumetric water content data, crop type and growth stage data, and actual irrigation effect data as training sets, a gradient boosting decision tree algorithm is used for supervised learning to predict the optimal dynamic crop coefficient and construct a machine learning model for the crop coefficient.

[0075] Furthermore, a training sample set is constructed using historical meteorological data, historical soil volumetric water content data, crop type identification data, crop growth stage coding data, and historical actual irrigation effect data. Supervised learning training is performed using the gradient boosting decision tree algorithm. The gradient boosting decision tree algorithm generates multiple weak decision tree models iteratively and combines them with weights to minimize the error between the predicted value and the true value, thereby optimizing the model parameters. The trained gradient boosting decision tree model can output the optimal dynamic crop coefficient prediction value based on the input crop type identification, crop growth stage coding, real-time meteorological data, and soil volumetric water content data, thus constructing a machine learning model for the crop coefficient.

[0076] S2.3 Input crop type, growth stage and historical yield data into the machine learning model of crop coefficient to obtain dynamic crop coefficient.

[0077] Furthermore, crop variety identifiers, digital growth stage codes, and normalized historical yield data are used as input features and fed into a pre-trained machine learning model for crop coefficients. The machine learning model for crop coefficients is constructed using a gradient boosting decision tree algorithm and generates dynamic crop coefficient prediction values ​​through forward propagation. The dynamic crop coefficient values ​​can reflect the actual water requirements of a specific crop variety under specific growth stages and historical yield backgrounds, replacing the traditional table lookup method to obtain dynamic crop coefficients.

[0078] S2.4. Based on the reference crop evapotranspiration and dynamic crop coefficient, the actual water requirement of the crop is obtained by multiplication.

[0079] The expression for the actual water requirement of crops is:

[0080] ET c =Kc d ynamic*ET0;

[0081] Among them, ET c Kc represents the actual water requirement of the crop. d ynamic is the dynamic crop coefficient.

[0082] S3. Calculate the water stress coefficient based on real-time soil volumetric water content data and field water holding capacity static parameter values, and obtain the basic irrigation demand index by combining future precipitation forecast data.

[0083] S3.1 The static parameter value of field water holding capacity was obtained by measuring soil samples from farmland using the ring cutter method.

[0084] Furthermore, at representative sampling points in the target farmland, undisturbed soil samples were collected using a ring cutter. The saturated soil samples were then placed in a gravity drainage device to drain naturally. After the gravity water was fully drained, the soil moisture content inside the ring cutter was measured. The measured value is the static parameter value of field water holding capacity, which serves as a benchmark parameter characterizing the maximum water holding capacity of the soil.

[0085] S3.2 Substitute the real-time soil volumetric water content data and the static parameter value of field water holding capacity into the water stress function to calculate the water stress coefficient.

[0086] The expression for the water stress coefficient is as follows:

[0087]

[0088] Among them, K s θ is the water stress coefficient, and θ is the real-time soil volumetric water content. lowerθ is the lower limit threshold of water stress. upper This represents the upper limit threshold for water stress.

[0089] S3.3. Based on the actual water requirement of crops and the water stress coefficient, the basic irrigation demand index is obtained by using the coefficient product method.

[0090] The basic irrigation demand index is expressed as follows:

[0091] ET cac =ET c *K s ;

[0092] Among them, ET cac It is the basic irrigation demand index.

[0093] S4. Based on real-time soil volumetric water content data and soil matrix potential data, fit the soil moisture characteristic curve and inversely calculate the dynamic value of field water holding capacity.

[0094] S4.1. Physical rule verification and outlier removal are performed on the real-time collected soil volumetric water content and soil matrix potential data to obtain a soil hydrodynamic dataset.

[0095] Furthermore, real-time soil volumetric water content data is collected using a frequency domain reflectance soil moisture sensor, and real-time soil matrix potential data is collected using a tensiometer. The collected data is then subjected to physical rule verification, which includes checking whether the soil matrix potential is negative and whether the soil volumetric water content is within a reasonable range. Data points that do not conform to physical laws are removed. Subsequently, the Dixon test is used to identify and remove statistically significant outliers, forming a soil hydrodynamic dataset.

[0096] S4.2 Input the soil hydrodynamic dataset into the physical information neural network, and output the trained neural network model of soil hydraulic properties through the fitting error and physical law constraint terms in the physical information neural network.

[0097] Furthermore, the soil hydrodynamic dataset is input into the physical information neural network for training. The loss function of the physical information neural network includes a data fitting error term and a physical constraint term. The data fitting error term calculates the mean square error between the soil volumetric water content predicted by the neural network and the measured soil volumetric water content in the soil hydrodynamic dataset. The physical constraint term calculates the derivative of the neural network output with respect to the input soil matrix potential using automatic differentiation technology and compares it with the constraints of soil hydraulic theory. The neural network parameters are optimized using the gradient descent algorithm to minimize the total loss function and output the trained neural network model of soil hydraulic properties.

[0098] S4.3 Input the standard soil matrix potential value of field water holding capacity from soil science into the trained neural network model of soil hydraulic properties, and generate the dynamic value of field water holding capacity through forward propagation calculation of the network.

[0099] The dynamic value expression for field water holding capacity is:

[0100] θ F C d ynamic=f P INN(ψ F A);

[0101] Where, θ F C d ynamic is the dynamic value of field water holding capacity, f P INN is a trained neural network model of soil hydraulic properties, ψ F Let A be the standard soil matrix potential, and let A be the set of internal weight parameters in the neural network model of soil hydraulic properties.

[0102] S5. Calculate the water deficit based on the dynamic value of field water holding capacity and soil volumetric water content data.

[0103] S5.1. Use division to calculate the ratio of the dynamic value of field water holding capacity to obtain the current relative soil moisture content.

[0104] The current expression for the relative moisture content of the soil is:

[0105]

[0106] Where, θ r This represents the current relative moisture content of the soil.

[0107] S5.2 Combine the current relative soil moisture content and crop root depth data, and calculate the water deficit using the soil water deficit calculation formula.

[0108] The expression for water deficit is:

[0109] D=(1-θ r )*θ F C d ynamic*Z r *1000;

[0110] Where D represents the water deficit.

[0111] S6. Calculate the vegetation index by acquiring multispectral images of farmland through remote sensing equipment, and combine it with real-time meteorological data to separate the actual water requirement of crops into soil evaporation and crop transpiration.

[0112] S6.1. Obtain multispectral images of farmland through a satellite remote sensing platform, perform radiometric calibration and atmospheric correction to obtain clean remote sensing image data.

[0113] Furthermore, multispectral image data of farmland areas were acquired by a multispectral imager carried on the Sentinel-2 satellite. The raw image data was radiometrically calibrated to convert digital quantization values ​​into surface reflectance. Subsequently, the 6S atmospheric correction model was used to eliminate the influence of atmospheric scattering and absorption on the surface reflectance signal, remove interference caused by atmospheric components such as aerosols and water vapor, and generate clean remote sensing image data.

[0114] S6.2. Using clean remote sensing image data, extract the reflectance of the near-infrared and red light bands and substitute it into the normalized differential vegetation index calculation formula to obtain the normalized vegetation index. Measure the crop canopy structure in the field and perform standardized data processing to obtain the measured biophysical parameters in the field.

[0115] The normalized vegetation index is expressed as follows:

[0116]

[0117] Wherein, NDVI is the normalized vegetation index, ρ NIR ρ represents the surface reflectance in the near-infrared band. Red This refers to the surface reflectance in the red light band.

[0118] S6.3. By using the normalized vegetation index and field-measured biophysical parameters, a statistical-physical hybrid prediction model is established using nonlinear regression.

[0119] Furthermore, using normalized vegetation index (NVI) data and field-measured biophysical parameter data as the modeling basis, a statistical-physical hybrid prediction model was constructed using a nonlinear regression method. The field-measured biophysical parameter data included leaf area index data measured by the LAI-2200 plant canopy analyzer and canopy spectral reflectance data measured by the ASD FieldSpec spectroradiometer. The nonlinear regression method fitted the mathematical relationship between the NVI and the field-measured biophysical parameters using the least squares method to establish the statistical-physical hybrid prediction model.

[0120] S6.4 Input the normalized vegetation index into the statistical-physical hybrid prediction model to obtain the leaf area index and canopy surface resistance.

[0121] The leaf area index is expressed as follows:

[0122]

[0123] Wherein, LAI is the leaf area index, LAI maxNDVI is the maximum leaf area index, and NDVI is the normalized difference vegetation index. max The maximum normalized vegetation index (NDVI) min is the minimum normalized vegetation index, and k is the adjustment coefficient.

[0124] The expression for canopy surface resistance is,

[0125]

[0126] Where, r s For canopy surface resistance, r s,min f1(SWC) represents the minimum stomatal resistance, f2(VPD) represents the soil moisture stress function, and f2(VPD) represents the atmospheric moisture stress function.

[0127] S6.5 Input the inverted leaf area index, canopy surface resistance and real-time meteorological data into the Shuttleworth-Wallace model, and obtain soil evaporation and crop transpiration by solving the energy balance equation.

[0128] The expression for soil evaporation is:

[0129]

[0130] Among them, E s Let λ be the soil evaporation rate, λ be the latent heat of vaporization of water, Δ be the slope of the saturated water vapor pressure-temperature curve, and R be the temperature. n For net canopy radiation, e s e is the saturated vapor pressure. a Where γ is the actual water vapor pressure, r is the wet / dry gauge constant, and γ is the actual water vapor pressure. a For aerodynamic drag.

[0131] The expression for crop transpiration is:

[0132]

[0133] Where T represents crop transpiration. Let G be the net radiation at the soil surface, G be the soil heat flux, and ρ be the net radiation at the soil surface. a For air density, c p The specific heat capacity of air at constant pressure. This refers to the aerodynamic drag above the canopy.

[0134] S7. Combine crop transpiration, water deficit, and effective precipitation from future precipitation forecast data to generate an irrigation index.

[0135] S7.1 Predict the soil moisture deficit sequence by using a recursive calculation method based on crop transpiration, effective precipitation, and current water deficit.

[0136] The expression for the soil moisture deficit sequence is as follows:

[0137] ΔS=P e ff(t)-T(t);

[0138] Where ΔS represents the soil moisture deficit sequence, P e ff(t) represents the effective precipitation at time t, T(t) represents the crop transpiration at time t, and t is the time period index.

[0139] S7.2 Analyze and predict the soil moisture deficit sequence, extract the minimum, maximum and trend of change, and obtain characteristic indicators that characterize the risk of future water stress.

[0140] Furthermore, statistical analysis was performed on the predicted soil moisture deficit sequence to extract the minimum, maximum, and linear regression slope characteristics. The minimum value reflects the most favorable water conditions during the prediction period, the maximum value represents the most severe water shortage during the prediction period, and the linear regression slope is fitted with the trend of the sequence over time using the least squares method to obtain characteristic indicators representing the risk of future water stress.

[0141] S7.3 Input the characteristic indicators representing the risk of future water stress into the decision rule set to generate the irrigation index.

[0142] Furthermore, characteristic indicators representing future water stress risks are input into the decision rule set, which includes drought resistance rules, water conservation rules, and optimization rules. The drought resistance rule stipulates that a high irrigation index is output when the maximum water deficit exceeds the crop tolerance threshold. The water conservation rule stipulates that a zero irrigation index is output when the minimum water deficit is significantly reduced due to precipitation. The optimization rule stipulates that an appropriate irrigation index is derived based on the trend of water deficit changes in the absence of extreme conditions. The rule engine performs logical judgments and numerical values ​​to generate the irrigation index.

[0143] This embodiment also provides a computer device applicable to the irrigation index prediction method based on meteorological data and field water holding capacity, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the irrigation index prediction method based on meteorological data and field water holding capacity as proposed in the above embodiment.

[0144] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0145] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the irrigation index prediction method based on meteorological data and field water holding capacity as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0146] In summary, this invention solves the benchmark drift problem and insufficient meteorological risk response caused by static parameters in existing technologies by dynamically calibrating field water holding capacity parameters and quantifying weather forecast uncertainties. It collects multi-source data and performs quality control alignment, calculates crop water requirements and basic irrigation indices, dynamically fits soil moisture characteristic curves through physical information neural networks, and inverts the dynamic value of field water holding capacity, thus solving the static parameter drift problem. Combined with remote sensing inversion of vegetation indices, it uses the Shuttleworth-Wallace model to separate evapotranspiration, and generates an irrigation index that integrates crop evapotranspiration, water deficit, and effective precipitation based on multi-period rolling water balance calculations and risk decision rules. This achieves online calibration of soil parameters and optimization of irrigation decisions.

[0147] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for predicting irrigation index based on meteorological data and field water holding capacity, characterized in that: include, Real-time meteorological data, real-time soil volumetric water content data, soil matrix potential data, and future precipitation forecast data were collected and subjected to quality control and time alignment processing to obtain a multi-source dataset. The reference crop evapotranspiration is calculated based on real-time meteorological data, and the actual crop water requirement is obtained by combining the crop coefficient. The water stress coefficient is calculated based on real-time soil volumetric water content data and field water holding capacity static parameter values. The basic irrigation demand index is obtained by combining future precipitation forecast data. Based on real-time soil volumetric water content data and soil matrix potential data, soil moisture characteristic curves are fitted, and field water holding capacity dynamic values ​​are calculated by inversion. Based on the dynamic values ​​of field water holding capacity and soil volumetric water content data, the water deficit is calculated. The vegetation index is calculated by acquiring multispectral images of farmland through remote sensing equipment. Combined with real-time meteorological data, the actual water requirement of crops is separated into soil evaporation and crop transpiration. The irrigation index is generated by combining crop transpiration, water deficit, and effective precipitation from future precipitation forecast data. The water stress coefficient is calculated based on real-time soil volumetric water content data and static field capacity parameters. Combined with future precipitation forecasts, the basic irrigation demand index is then obtained. Includes the following steps, The static parameter value of field water holding capacity was obtained by measuring soil samples from farmland using the ring cutter method. Substitute real-time soil volumetric water content data and field water holding capacity static parameter values ​​into the water stress function to calculate the water stress coefficient. Based on the actual water requirement of crops and the water stress coefficient, the basic irrigation demand index is obtained by using the coefficient product method. The vegetation index is calculated by acquiring multispectral images of farmland using remote sensing equipment. Combined with real-time meteorological data, the actual water requirement of crops is separated into soil evaporation and crop transpiration. This includes the following steps. By acquiring multispectral images of farmland through satellite remote sensing platforms and performing radiometric calibration and atmospheric correction, clean remote sensing image data is obtained. Using clean remote sensing image data, the reflectance of the near-infrared and red bands was extracted and substituted into the normalized differential vegetation index calculation formula to obtain the normalized vegetation index. The crop canopy structure was measured in the field and the data was standardized to obtain the field measured biophysical parameters. A statistical-physical hybrid prediction model was established using normalized vegetation index and field-measured biophysical parameters, employing a nonlinear regression method. The normalized vegetation index was input into the statistical-physical hybrid prediction model to obtain the leaf area index and canopy surface resistance. The leaf area index, canopy surface resistance, and real-time meteorological data obtained from the inversion are input into the Shuttleworth-Wallace model. By solving the energy balance equation, the data are separated into soil evaporation and crop transpiration.

2. The irrigation index prediction method based on meteorological data and field water holding capacity as described in claim 1, characterized in that: Real-time meteorological data, real-time soil volumetric water content data, soil matric potential data, and future precipitation forecast data are collected and processed for quality control and time alignment to obtain a multi-source dataset. This process includes the following steps: By using real-time meteorological data from field weather stations, real-time soil volumetric water content data, and soil matrix potential data, future precipitation forecast data can be obtained through API interfaces. Real-time meteorological data, real-time soil volumetric water content data, soil matrix potential data, and future precipitation forecast data are input into the quality control module. The Dixon test is used to identify and remove outliers to obtain a clean real-time data sequence. The clean real-time data sequence is then input into the data repair module. For missing points caused by the removal of outliers, time series linear interpolation is used to fill in the missing points to obtain a complete real-time data sequence. The complete real-time data sequence and the future precipitation forecast data are input into the time alignment module and unified to the UTC hourly timestamp to obtain a preliminary aligned multi-source data sequence; The future precipitation forecast data from the initially aligned multi-source data sequence is input into the hydrological calculation module, and the effective precipitation is obtained by subtracting canopy interception and surface runoff loss from soil texture parameters. The effective precipitation data is merged and packaged with the complete real-time data sequence to obtain a multi-source dataset.

3. The irrigation index prediction method based on meteorological data and field water holding capacity as described in claim 2, characterized in that: The actual water requirement of crops is obtained by calculating reference crop evapotranspiration based on real-time meteorological data and combining it with crop coefficients, including the following steps. Based on real-time meteorological data, the reference crop evapotranspiration was calculated using the Penman-Monteith formula. Using historical meteorological data, soil volumetric water content data, crop type and growth stage data, and actual irrigation effect data as training sets, a gradient boosting decision tree algorithm is used for supervised learning to predict the optimal dynamic crop coefficient and build a machine learning model for the crop coefficient. By inputting crop type, growth stage, and historical yield data into a machine learning model for crop coefficients, dynamic crop coefficients are obtained. Based on the reference crop evapotranspiration and dynamic crop coefficient, the actual water requirement of the crop is obtained by multiplication.

4. The irrigation index prediction method based on meteorological data and field water holding capacity as described in claim 3, characterized in that: Based on real-time soil volumetric water content data and soil matrix potential data, a soil moisture characteristic curve is fitted, and the dynamic value of field capacity is calculated by inversion. Includes the following steps, The real-time collected soil volumetric water content and soil matrix potential data are subjected to physical rule verification and outlier removal to obtain a soil hydrodynamic dataset. The soil hydrodynamic dataset is input into the physical information neural network. Through the fitting error and physical law constraint terms in the physical information neural network, the trained neural network model of soil hydraulic properties is output. The standard soil matrix potential value of field water holding capacity in soil science is input into a trained neural network model of soil hydraulic properties, and the dynamic value of field water holding capacity is generated through forward propagation calculation of the network.

5. The irrigation index prediction method based on meteorological data and field water holding capacity as described in claim 4, characterized in that: The water deficit is calculated based on field water holding capacity dynamics and soil volumetric water content data, including the following steps. The relative soil moisture content is obtained by dividing the dynamic value of field water holding capacity and calculating the ratio. The water deficit is calculated by combining the current relative soil moisture content and crop root depth data with the soil water deficit calculation formula.

6. The irrigation index prediction method based on meteorological data and field water holding capacity as described in claim 5, characterized in that: An irrigation index is generated by combining crop transpiration, water deficit, and effective precipitation from future precipitation forecasts. Includes the following steps, The soil moisture deficit sequence is predicted by using a recursive calculation method based on crop transpiration, effective precipitation and current water deficit. By analyzing and predicting the soil moisture deficit sequence, the minimum, maximum and trend of change are extracted to obtain characteristic indicators that characterize the risk of future water stress. Characteristic indicators that represent the risk of future water stress are input into the decision rule set to generate an irrigation index.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the irrigation index prediction method based on meteorological data and field water holding capacity as described in any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the irrigation index prediction method based on meteorological data and field water holding capacity as described in any one of claims 1 to 6.

Images