Soil moisture prediction method and system for dryland farmland under limited irrigation

Abstract

The present invention provides a soil moisture prediction method and system for dryland farmland under limited irrigation, which relates to the technical field of soil moisture prediction. The method includes: dividing the soil into multiple layers according to depth, each layer corresponding to a different water retention and water absorption capacity coefficient, collecting the water content of each layer at multiple moments and its influencing parameters, constructing a time series data set and training a long-short-term memory network model; calculating the water deficit based on the suitable water content of the oats in the current growth stage, and after trend correction, weighting the water retention and water absorption capacity coefficients to obtain the irrigation demand index, and determining the highest priority layer; using the model to predict the water content of the layer at the next moment, combining the field water holding capacity calculation theory and the maximum allowable irrigation amount, and taking the minimum value to correct the moisture content prediction value. The present invention scientifically determines the irrigation priority, realizes on-demand and limited irrigation, and solves the problem that the existing technology has difficulty in assessing the actual moisture content of the soil layer and optimizing resource allocation, as well as the problem of lacking irrigation constraint logic.

Description

Technical Field

[0001] The present invention relates to the technical field of soil moisture prediction, and in particular to a method and system for predicting soil moisture in dryland farmland under limited supplementary irrigation. Background Art

[0002] In agricultural production, soil moisture prediction is a core component of precision irrigation management. Scientifically predicting soil moisture can effectively improve supplementary irrigation efficiency, conserve water resources, and ensure crop yields, especially for dryland farmland. Dryland farmland is affected by factors such as uneven natural rainfall, differences in soil water retention capacity, and the dynamic changes in crop water requirements during growth stages. The imbalance between water supply and demand is even more pronounced. Insufficient water can hinder crop growth and reduce yields. Excessive supplementary irrigation not only wastes limited water resources but can also cause secondary soil salinization and root hypoxia.

[0003] Publication number CN120216911A provides a method, system, and device for predicting deep soil moisture based on shallow soil moisture data. The method includes the following steps: obtaining geographic and hydrological data from several monitoring points within the watershed to be measured; stratifying the soil to obtain time-series data on soil moisture at different levels at each monitoring point; integrating the geographic, hydrological, and soil moisture time-series data from each monitoring point to obtain a time-series data matrix for each monitoring point; constructing a long-short-term memory (LSTM) network model and training it using the time-series data matrix; and using the trained LSTM network model to predict deep soil moisture based on shallow soil moisture data. This method not only effectively predicts deep soil moisture but also enables rapid learning and long-term prediction in other watersheds through transfer learning.

[0004] However, there are still the following deficiencies. As can be seen from the above statements, soil stratification only relies on depth division, but does not combine the water retention capacity coefficient and water absorption capacity coefficient of each layer, resulting in the inability to accurately assess the actual moisture of each layer and making it difficult to support the determination of irrigation priority. At the same time, the existing technology only focuses on moisture prediction itself and does not construct the logic associated with irrigation. This includes neither calculating the deficit between the current moisture and the appropriate moisture content in the crop growth stage to determine the necessity of irrigation, nor considering the maximum allowable irrigation amount as a constraint to achieve the limited goal of "irrigation on demand without waste."

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0006] The purpose of the present invention is to provide a method and system for predicting soil moisture in dryland farmland under limited supplementary irrigation, so as to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] S1. Divide dryland farmland soil into multiple layers according to different depth ranges. Each layer corresponds to a different water retention capacity coefficient and water absorption capacity coefficient. Collect the historical water content and influencing parameters of all layers. Construct a time series dataset for each layer. Use the water content and influencing parameters at m+1 consecutive moments as input and the water content at the next moment as output to train a long-short-term memory network model for each layer.

[0009] S2. Determine the current growth stage of the oats and calculate the current water deficit based on the appropriate moisture content for the current growth stage. Calculate the moisture change rate and trend correction coefficient based on the moisture content for the most recent m+1 moments. Correct the water deficit using the trend correction coefficient. A weighted calculation is performed using the water retention capacity coefficient and the water absorption capacity coefficient to determine the irrigation requirement index. The stratum with the highest irrigation requirement index is designated as the highest irrigation priority stratum.

[0010] S3. Input the water content and influencing parameters of the highest priority layer at the most recent m+1 moments into the long short-term memory network model to predict the water content of the layer at the next moment;

[0011] S4. Determine the field water holding capacity based on the soil parameters of the highest priority layer, combine the suitable water content and the predicted water content at the next moment, calculate the theoretical irrigation amount and the maximum allowable irrigation amount of the highest priority layer, compare the theoretical irrigation amount and the maximum allowable irrigation amount, take the minimum value of the two as the actual irrigation amount, correct the water content prediction value, and obtain the corrected water content prediction value of the highest priority layer.

[0012] Furthermore, the dryland farmland soil is divided into different depth ranges to obtain multiple layered soils. The specific steps are as follows:

[0013] Based on the characteristic that the root distribution range of oats is 0-60cm, the dryland farmland soil is divided into several continuous depth intervals: 0-20cm, 20-40cm, and 40-60cm, corresponding to the upper layer, middle layer, and lower layer respectively.

[0014] Furthermore, the water deficit at the current moment is calculated based on the appropriate water content corresponding to the current growth stage, according to the following formula:

[0015] ;

[0016] in, In the current growth stage, The water deficit of each layer, In the current growth stage, The appropriate water content corresponding to each layer is For the current moment The moisture content of each layer;

[0017] Where, For hierarchical indexes, , is the number of layers.

[0018] Furthermore, the moisture change rate is calculated based on the moisture content at the most recent m+1 moments, and the trend correction coefficient is calculated to correct the moisture deficit. The formula is as follows:

[0019] The water content time series data consisting of m+1 moments is ,in, The m+1th moment The water content of the first layer, that is, the water content of the first layer at the current moment The moisture content of each layer, For the The moment The moisture content of each layer, is the moment index among the m+1 moments, ;

[0020] ;

[0021] in, From the first moment to the current moment, The average rate of change of moisture in each layer is is the time interval between two adjacent moments;

[0022] ;

[0023] in, In the current growth stage, The trend correction coefficient of each layer, is the modified strength coefficient when the moisture content is reduced, , is the modified strength coefficient when moisture increases, , is the absolute value of the average rate of change of moisture;

[0024] when hour,

[0025] ;

[0026] in, In the current growth stage, Water deficit after stratification correction;

[0027] when When The moisture content of each layer is equal to or higher than the suitable moisture content, and there is no need for supplementary irrigation.

[0028] Furthermore, the corrected water deficit, water retention capacity coefficient, and water absorption capacity coefficient are combined for weighted calculation to obtain the irrigation demand index, based on the following formula:

[0029] ;

[0030] in, In the current growth stage, A stratified irrigation demand index, For the The water retention capacity coefficient of each layer, For the The moisture absorption capacity coefficient of each layer, 、 are the weight coefficients of water retention capacity coefficient and water absorption capacity coefficient, respectively. On this basis, .

[0031] Furthermore, the layer with the highest index is determined as the layer with the highest priority for replenishment. The specific steps are as follows:

[0032] Only the recharge demand index The hierarchical participation sorting;

[0033] Under the current growth stage, the irrigation demand index of each layer is sorted from large to small. The layer with the highest irrigation demand index is the layer with the highest irrigation priority. When there are two layers of irrigation demand index When the water retention coefficients are the same, first compare the water retention capacity coefficients of the two, and give priority to the layer with the lowest water retention capacity coefficient. When the water retention capacity coefficients are the same, then compare the water absorption capacity coefficients of the two, and give priority to the layer with the highest water absorption capacity coefficient.

[0034] Furthermore, the specific steps and formulas of S4 are as follows:

[0035] ;

[0036] in, It is the theoretical irrigation amount for the highest irrigation priority layer at the current growth stage. It is the appropriate water content corresponding to the highest priority level of supplementary irrigation at the current growth stage. is the water content of the highest priority layer at the next moment in the current growth stage, is the mass of solid particles per unit volume of soil, The thickness of the highest priority layer for recharging;

[0037] ;

[0038] in, The maximum allowable irrigation amount for the highest irrigation priority layer at the current growth stage. It is the field water holding capacity of the highest priority level for supplementary irrigation at the current growth stage. The water content of the highest priority layer at the current moment;

[0039] Compare the theoretical recharge volume with the maximum allowable recharge volume, and take the minimum value of the two as the actual recharge volume, which is recorded as ;

[0040] ;

[0041] ;

[0042] in, It is the predicted value of water content after correction for the highest priority layer of supplementary irrigation in the current growth stage.

[0043] To achieve the above object, the present invention further provides the following technical solutions:

[0044] A soil moisture prediction system for dryland farmland under limited supplementary irrigation, the system being used to execute any of the above-mentioned soil moisture prediction methods for dryland farmland under limited supplementary irrigation, comprising:

[0045] The model construction module is used to divide the dryland farmland soil into different depth ranges to obtain multiple layers. Each layer corresponds to a different water retention capacity coefficient and water absorption capacity coefficient. The previous moisture content and influencing parameters of all layers are collected. A time series dataset is constructed for each layer. The moisture content and influencing parameters at m+1 consecutive moments are used as input, and the moisture content at the next moment is used as output. The long-short-term memory network model of each layer is trained separately.

[0046] The calculation module is used to determine the current growth stage of the oats, calculate the water deficit at the current moment based on the appropriate water content corresponding to the current growth stage, calculate the water change rate and trend correction coefficient based on the water content at the most recent m+1 moments, correct the water deficit using the trend correction coefficient, and perform weighted calculation based on the water retention capacity coefficient and the water absorption capacity coefficient to obtain the irrigation demand index. The layer with the highest irrigation demand index is designated as the highest irrigation priority layer.

[0047] The simulation module is used to input the water content and influencing parameters of the highest priority layer at the latest m+1 moments into the long short-term memory network model to predict the water content of the layer at the next moment;

[0048] The correction module is used to determine the field water holding capacity based on the soil parameters of the highest priority layer, combine the suitable water content and the predicted water content at the next moment, calculate the theoretical irrigation amount and the maximum allowable irrigation amount of the highest priority layer, compare the theoretical irrigation amount and the maximum allowable irrigation amount, take the minimum value of the two as the actual irrigation amount, correct the water content prediction value, and obtain the corrected water content prediction value of the highest priority layer.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] The present invention divides the soil into multiple layers according to different depth ranges, introduces the water retention capacity coefficient and the root water absorption capacity coefficient, calculates the water deficit degree in combination with the appropriate water content of the crop growth stage, and dynamically adjusts the trend correction coefficient to obtain a weighted irrigation demand index to scientifically determine the irrigation priority. This solves the problem of existing technologies that make it difficult to assess the actual moisture content of the soil layer and optimize resource allocation.

[0051] By determining the field water holding capacity based on the soil parameters of the highest priority layer, combining the suitable water content and the predicted water content at the next moment, calculating the theoretical irrigation amount and the maximum allowable irrigation amount of the highest priority layer, comparing the theoretical irrigation amount and the maximum allowable irrigation amount, taking the minimum of the two as the actual irrigation amount, and correcting the water content prediction value, the corrected water content prediction value of the highest priority layer is obtained, ensuring that irrigation can meet the needs of crop growth and avoid water leakage and waste, realizing "on-demand replenishment and limited irrigation", and making up for the deficiency of existing technology in the lack of irrigation constraint logic. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 Schematic diagram of the overall method flow of the present invention;

[0053] Figure 2 It is a block diagram of the module composition of the present invention;

[0054] Figure 3 is the fitting curve of the water deficit degree and the supplementary irrigation demand index after correction in the present invention;

[0055] Figure 4 is the fitting curve of the water retention capacity coefficient and the supplementary irrigation demand index of the present invention;

[0056] Figure 5 It is the fitting curve of the water absorption capacity coefficient and the supplementary irrigation demand index of the present invention. DETAILED DESCRIPTION

[0057] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to specific embodiments.

[0058] It should be noted that, unless otherwise defined, the technical or scientific terms used in the present invention should have the usual meanings understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative position relationships. When the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0059] Example 1:

[0060] See also Figures 1 to 5 , the present invention provides a technical solution:

[0061] A method for predicting soil moisture in dryland farmland under limited irrigation, comprising the following steps:

[0062] S1. Divide dryland farmland soil into multiple layers according to different depth ranges. Each layer corresponds to a different water retention capacity coefficient and water absorption capacity coefficient. Collect the historical water content and influencing parameters of all layers. Construct a time series dataset for each layer. Use the water content and influencing parameters at m+1 consecutive moments as input and the water content at the next moment as output to train a long-short-term memory network model for each layer.

[0063] Based on the above embodiment, the dryland farmland soil is divided into different depth ranges to obtain multiple layered soils. The specific steps are as follows:

[0064] Based on the characteristic that the root distribution range of oats is 0-60cm, the dryland farmland soil is divided into several continuous depth intervals: 0-20cm, 20-40cm, and 40-60cm, corresponding to the upper layer, middle layer, and lower layer respectively.

[0065] Based on the above embodiment, the method for determining the water retention capacity coefficient and the water absorption capacity coefficient comprises the following specific steps:

[0066] For each divided layer (such as 0-20cm, 20-40cm, etc.), multiple-point sampling (such as 5-10 points per layer) is carried out separately in the corresponding depth area, and the samples are mixed evenly and impurities are removed to obtain independent representative samples of each layer.

[0067] The water retention capacity coefficient is determined as follows:

[0068] For each representative sample of each layer, perform the following operations: weigh 100g of air-dried soil sample, place it in an aluminum box of known weight, and record the total weight of the aluminum box and soil sample. , soak the aluminum box containing the soil sample in water for 24 hours, so that the soil can fully absorb water and reach saturation. Take out the aluminum box, gently absorb the water on the surface of the aluminum box with filter paper, and then weigh it, and record the total weight of the aluminum box and soil sample at this time , Place the aluminum box in an oven and dry it at 105°C until constant weight is reached. Take it out and place it in a desiccator to cool to room temperature. Weigh it again and record the total weight of the aluminum box and the dried soil sample. ;

[0069] Calculate soil saturation moisture content :

[0070] ;

[0071] in, is the weight of the aluminum box;

[0072] Water retention capacity coefficient Correction is made based on the soil texture and structural factors, and the correction coefficient is determined for each layer separately. The correction coefficient is determined by consulting relevant literature or based on local soil characteristics experience. The water retention capacity coefficient calculation formula is:

[0073] ;

[0074] in, For the The saturated water content of the soil in each layer.

[0075] The water absorption capacity coefficient is determined as follows:

[0076] 50 g of soil was taken from each mixed sample collected in each layer, placed in a petri dish, and adjusted to about 60% of the field capacity;

[0077] Place each layer of the culture dish in an independent sealed container, add saline solution of the same concentration (such as 0.1 mol / L) to the bottom, start timing at the same time, and weigh it every 1 hour (record the weight of the sample ), lasting 24 hours;

[0078] For each layer of data, the water absorption ( ) The curve of change with time is fitted to obtain the absorption rate equation of each layer;

[0079] According to the slope of the absorption rate equation of each layer at a specific time point (such as 12 hours), , the slopes of the curves of different layers are different, so different.

[0080] Based on the above embodiment, the parameters affecting moisture content include rainfall, temperature, and wind speed. The collection method of rainfall, temperature, and wind speed is as follows:

[0081] Set the same time interval as soil moisture monitoring (e.g. every hour) and automatically record the rainfall at m+1 moments through the rain gauge;

[0082] Set time nodes synchronously with rainfall and soil moisture monitoring, and automatically record the temperature values ​​at m+1 moments through sensors;

[0083] Set time nodes synchronously with rainfall and soil moisture monitoring, and record wind speed values ​​at m+1 moments;

[0084] Among them, multiple groups of water content, rainfall, temperature and wind speed at the same moment are collected, and the water content, rainfall, temperature and wind speed at the same moment are all the average values ​​at that moment.

[0085] Among them, the previously collected water content and its influencing parameters must be normalized, and subsequent data analysis and processing are all based on normalized data.

[0086] Based on the above embodiment, the specific process of constructing and training the long short-term memory network model is as follows:

[0087] 1) For each layer’s “moisture content and influencing parameters at m+1 consecutive moments” data, sort the data in chronological order to form a one-dimensional time series of “timestamp-moisture content and influencing parameters”.

[0088] The "sliding window method" is used to divide the training samples: the window length is set to m+1 (e.g. 24 hours), the water content and influencing parameters for 24 consecutive hours are used as input features, and the water content at the next moment is used as the prediction target;

[0089] 2) Construction of long short-term memory network model:

[0090] Input layer: receives time series, corresponding to the water content and influencing parameters for 24 consecutive hours;

[0091] LSTM layer: 1-2 layers of LSTM units, with 64 neurons in the first layer and 32 neurons in the second layer. Dropout layers (dropout rate 0.2) are added to prevent overfitting.

[0092] Fully connected layer: 1 layer of 64 neurons, with ReLU (rectified linear unit) activation function;

[0093] Output layer: 1 neuron, outputs the predicted water content value at the next moment;

[0094] 3) Parameter configuration of the long short-term memory network model:

[0095] Optimizer: Adam (learning rate 0.001, decay rate 1e-5);

[0096] Loss function: mean square error, which measures the deviation between the predicted value and the true value;

[0097] Training rounds: Initially set to 100 rounds, combined with an early stopping mechanism (monitoring the validation set loss and terminating if there is no decrease for 10 consecutive rounds);

[0098] 4) Training of long short-term memory network model:

[0099] The sample set of each stratification is divided into a training set (70%), a validation set (20%), and a test set (10%) in a ratio of 7:2:1 to ensure a continuous time sequence;

[0100] The model is trained with the training set, the validation set loss is calculated in each round, and the parameters are adjusted dynamically. When the validation set loss is minimized and stabilized, the current model weights are saved, which is the optimal LSTM model for that layer.

[0101] Use the test set to evaluate model performance. The core indicators include:

[0102] Root mean square error: reflects the overall prediction deviation;

[0103] Mean absolute error: measures the size of typical errors;

[0104] Coefficient of determination: evaluates the model's ability to explain data variation. A coefficient of determination greater than or equal to 0.8 is considered acceptable.

[0105] S2. Determine the current growth stage of the oats and calculate the current water deficit based on the appropriate moisture content for the current growth stage. Calculate the moisture change rate and trend correction coefficient based on the moisture content for the most recent m+1 moments. Correct the water deficit using the trend correction coefficient. A weighted calculation is performed based on the water retention capacity coefficient and the water absorption capacity coefficient to determine the irrigation requirement index. The tier with the highest irrigation requirement index is designated as the tier with the highest irrigation priority.

[0106] Based on the above embodiment, the oat growth stages include sowing period, seedling period, overwintering period, jointing period, booting period, heading and flowering period, grain filling period, and maturity period.

[0107] Based on the above embodiment, the appropriate water content corresponding to each growth stage is obtained. The specific steps are as follows:

[0108] Agricultural research institutions and universities have conducted long-term experiments on oat varieties in different regions (such as the North China Plain, the Huanghuai region, and the arid northwest region) to summarize the appropriate moisture content ranges for each growth stage. These can be directly referenced:

[0109] Among them, the suitable soil moisture content during the sowing period (as a percentage of field water holding capacity) is: 60%-70%;

[0110] Suitable soil moisture content during the seedling and overwintering stages (as a percentage of field water holding capacity): 60%-70%;

[0111] Suitable soil moisture content during jointing and booting stages (as a percentage of field water holding capacity): 70%-80%;

[0112] Suitable soil moisture content during heading and flowering (as a percentage of field water holding capacity): 70%-80%;

[0113] Suitable soil moisture content during the filling period (as a percentage of field capacity): 65%-75%;

[0114] Suitable soil moisture content at maturity (as a percentage of field capacity): 50%-60%.

[0115] Based on the above embodiment, the water deficit at the current moment is calculated in combination with the appropriate water content corresponding to the current growth stage, according to the following formula:

[0116] ;

[0117] in, In the current growth stage, The water deficit of each layer is used to combine the two index parameters of the suitable water content in the current growth stage and the actual water content at the current moment to quantify the water deficit or surplus state of the layer;

[0118] when , indicating that The actual water content of a layer is lower than the appropriate water content in the current growth stage, that is, the layer is in a water deficit state;

[0119] when , indicating that The actual water content of each layer is equal to the appropriate water content at the current growth stage, that is, the water supply and demand of the layer is balanced;

[0120] when , indicating that The actual water content of a layer is higher than the appropriate water content in the current growth stage, that is, the layer is in a water surplus state;

[0121] Where, In the current growth stage, The appropriate water content corresponding to each layer is For the current moment The moisture content of each layer;

[0122] Where, For hierarchical indexes, , is the number of layers.

[0123] Based on the above embodiment, the moisture change rate is calculated based on the moisture content at the most recent m+1 moments, and the trend correction coefficient is calculated to correct the moisture deficit. The formula is as follows:

[0124] The water content time series data consisting of m+1 moments is ,in, The m+1th moment The water content of the first layer, that is, the water content of the first layer at the current moment The moisture content of each layer, For the The moment The moisture content of each layer, is the moment index among the m+1 moments, ;

[0125] ;

[0126] in, From the first moment to the current moment, The average rate of change of moisture in each layer is used to combine the average rate of change of moisture in the first layer at the current moment. The water content of each layer, the first moment The moisture content of each layer is quantified from the initial moment to the current moment. The overall change trend and average speed of the water content of each layer, and The larger it is, the faster the average rate of change is;

[0127] when When , the moisture content shows an overall increasing trend;

[0128] when When the water content is kept stable,

[0129] when When the water content is 1%, the overall water content tends to decrease;

[0130] Where, is the time interval between two adjacent moments;

[0131] The current soil water deficit ( ), the actual risk needs to be determined in combination with the moisture trend. For example, "currently in deficit and decreasing moisture" and "currently in deficit but increasing moisture" have completely different impacts on oats and need to be distinguished by a correction factor;

[0132] 1) When (When there is a lack of current water);

[0133] Case 1: (Water continues to decrease and the deficit worsens);

[0134] At this time, oats are facing the dual risks of "deficit + deterioration", and need to be Amplify the deficit;

[0135] in, The bigger it is, the faster the water will be lost. The larger the corrected deficit The higher it is, the priority is to refill it;

[0136] Case 2: (Water continues to increase and the deficit is alleviated);

[0137] At this time, the deficit is improving and needs to be improved through Reduce deficit;

[0138] in, The bigger it is, the faster the moisture increases. The smaller it is, the lower the deficit after correction, and the lower the priority of refilling;

[0139] Case 3: (Water is stable and the deficit remains unchanged);

[0140] No trend change. , then the deficit is not corrected and the current assessment is maintained;

[0141] The specific function form is as follows:

[0142] ;

[0143] in, In the current growth stage, The trend correction coefficient is used to dynamically correct the current water deficit based on the changing trend of soil moisture, so as to more accurately reflect the actual water risk status faced by crops. The larger the trend correction coefficient, the more significant the current water risk (worsening deficit) is;

[0144] Where, is the modified strength coefficient when the moisture content is reduced, , is the modified strength coefficient when moisture increases, , is the absolute value of the average rate of change of moisture;

[0145] when hour,

[0146] ;

[0147] in, In the current growth stage, Water deficit after stratification correction;

[0148] when When The moisture content of each layer is greater than or equal to the suitable moisture content, and there is no need for supplementary irrigation.

[0149] Based on the above embodiments, Both are modified intensity factors that quantify the effect of the rate of moisture change on the "current deficit":

[0150] Corrected strength factor when moisture is reduced : When crops are in water deficit and water continues to decrease (double risk), the deficit is amplified to trigger supplementary irrigation first;

[0151] Corrected strength factor when moisture increases : When crops are in water deficit but water continues to increase (risk mitigation), reduce the deficit and lower the priority of supplementary irrigation;

[0152] Different crops (e.g. oats) have different sensitivities to water deficit exacerbation and deficit mitigation, and the coefficient range needs to match the crop tolerance threshold:

[0153] Because "deficit + continuous reduction of water" is the most direct harm to crops, a stronger correction intensity is needed to highlight the risk and give priority to supplementary irrigation. The range is higher (0.1-0.2);

[0154] In the case of "deficit but increased moisture", crop risks are already easing and overcorrection may lead to delayed re-irrigation, so Weaker than ;

[0155] The soil moisture change rate is affected by factors such as precipitation, evaporation, and crop transpiration. Its absolute value is usually in a relatively stable range, and the coefficient range needs to be consistent with The actual fluctuation of ),even though The absolute value of is small, which will also lead to A sharp rise may lead to over-filling; if the coefficient is too small (e.g. ), it is impossible to effectively distinguish the difference between "rapid water loss" and "slow water loss";

[0156] The upper limit is set to 0.15 (lower than 0.2), which can reduce the misjudgment of "excessive reduction in irrigation priority due to a short-term increase in water";

[0157] The lower limit of the coefficient is set to 0.1. If the lower limit of the coefficient is lower than 0.1 (for example, 0.05), when the moisture change rate is small, the correction effect can be almost ignored, which will weaken the impact of the "moisture change trend" on decision-making and lose the meaning of setting the correction coefficient.

[0158] At the same time, when the correction coefficient corresponding to the moisture change rate is lower than 0.1, its actual impact on crop growth has not yet reached the level that requires adjustment; and 0.1 just corresponds to the physiological critical point of the crop where "it begins to need to pay attention to trend changes", which means it will neither intervene too early nor too late.

[0159] Table 1. Changes in the supplementary irrigation demand index with the corrected water deficit, water retention capacity coefficient, and water absorption capacity coefficient

[0160]

[0161] According to Table 1, the supplementary irrigation demand index is positively correlated with the modified water deficit and water absorption capacity coefficient, and negatively correlated with the water retention capacity coefficient, as shown in the following:

[0162] The corrected water deficit decreased from 0.95 to 0.28 (decreasing), and the corresponding irrigation demand index showed a decreasing trend;

[0163] The water retention capacity coefficient increased from 0.2 to 0.7 (increasing), and the corresponding supplementary irrigation demand index showed a decreasing trend;

[0164] The water absorption capacity coefficient decreased from 0.9 to 0.5 (decreasing), and the corresponding irrigation demand index showed a decreasing trend.

[0165] Under the combined effect of the three factors, the irrigation demand index decreases as the corrected water deficit decreases, the water retention capacity coefficient increases, and the water absorption capacity coefficient decreases, reflecting the synergistic influence of each parameter on the urgency of irrigation.

[0166] according to Figure 3-Figure 5 It can be seen that the corrected water deficit is positively correlated with the supplementary irrigation demand index. When the deficit increases, the index rises faster, indicating that the more severe the water deficit, the more urgent the supplementary irrigation.

[0167] The water retention capacity coefficient is negatively correlated with the supplementary irrigation demand index. The larger the coefficient, the faster the index decreases, indicating that the better the soil water retention is, the weaker the supplementary irrigation demand is.

[0168] The water absorption capacity coefficient is positively correlated with the supplementary irrigation demand index. When the coefficient increases, the index rises faster, reflecting that the stronger the soil water absorption, the more urgent the supplementary irrigation.

[0169] Based on the above embodiment, a weighted calculation is performed by combining the corrected water deficit, water retention capacity coefficient, and water absorption capacity coefficient to obtain the supplementary irrigation demand index, according to the following formula:

[0170] ;

[0171] in, In the current growth stage, The supplementary irrigation demand index is used to combine the three index parameters of water deficit, water retention capacity coefficient and water absorption capacity coefficient after correction to determine the supplementary irrigation demand index of the first layer. The greater the supplementary irrigation demand index, the greater the urgency and actual need for supplementary irrigation in the soil layer.

[0172] On this basis, it should be noted that:

[0173] As a basic item, it directly reflects the current stratified water deficit state and is the "original driving force" of the need for supplementary irrigation - the more severe the deficit, The larger the value is, the higher the irrigation demand index is. and It is a positive correlation;

[0174] It is Water retention capacity coefficient of each layer (the larger the value, the stronger the water retention capacity), ) reflects the characteristics of "weak water retention → easy water loss". Soil with weak water retention capacity has a short water retention time after irrigation, and more urgent irrigation is needed to avoid aggravation of the deficit. Therefore, through ( ) Amplify demand; weight Control the extent of its impact, therefore, and It is a negative correlation;

[0175] It is the water absorption capacity coefficient (the larger the value, the stronger the water absorption capacity), which directly reflects the soil's "absorption efficiency" of supplementary irrigation water. Soil with strong water absorption capacity can quickly alleviate the deficit after supplementary irrigation, and the "actual effect" of supplementary irrigation is more significant. The more efficient the soil is in using water, the more worthy of priority supplementary irrigation when it is in deficit. This is not because water is not needed due to strong water absorption, but because the value of supplementary irrigation is higher due to strong water absorption, which is reflected in the greater urgency and actual demand. Therefore, and It is a positive correlation.

[0176] In the formula, and Multiplication avoids the one-sidedness of “only looking at the deficit without considering the effect” or “only looking at the ability without considering the deficit”;

[0177] The function form is not chosen arbitrarily, but strictly corresponds to the physical meaning of the three indicators and the logical relationship between the recharge decision:

[0178] use As a multiplier, ensure that the replenishment demand is based on the premise of "existence of deficit"; use ( ) reflects the negative feedback of "urgent irrigation is required when water retention is weak"; It reflects the positive feedback of "strong water absorption means high supplementary irrigation value";

[0179] The combination of weighted sum and product realizes the organic unity of "basic demand" and "efficiency regulation", and finally reasonably quantifies the urgency of supplementary irrigation.

[0180] In summary, the above function form is used to express the functional relationship between the supplementary irrigation demand index and the three indicator parameters of the corrected water deficit, water retention capacity coefficient, and water absorption capacity coefficient.

[0181] Where, is the water retention capacity coefficient, is the water absorption capacity coefficient, 、 are the weight coefficients of water retention capacity coefficient and water absorption capacity coefficient respectively;

[0182] Based on the actual needs of agricultural production, for soils with water deficit, soils that can use supplementary irrigation water faster and more efficiently have a higher cost-effectiveness ratio and therefore need to be given a higher weight. , to amplify its impact on the “urgency of recharge”.

[0183] In contrast, its impact on "Current Recharge Urgency" is weaker than "Instant Absorption Effect", so it is given a lower weight. .

[0184] When crops are experiencing water deficit, "rapidly alleviating the current stress" is usually prioritized over "preventing future water loss." Soil with strong water absorption capacity can directly solve the current problem, while soil with weak water retention capacity will more likely affect the continued effect of irrigation. Therefore, setting the weight , so that the impact of "immediate effect" is greater than "long-term maintenance", which is in line with the judgment logic of the urgency of supplementary irrigation in actual production.

[0185] Therefore, in On this basis, .

[0186] As an implementation method, The value range is 0-0.5, The value range is 0.5-1.0. The specific value is set by technical personnel according to actual conditions and is not limited here.

[0187] Based on the above embodiment, the layer with the highest index is determined as the layer with the highest priority for replenishment irrigation. The specific steps are as follows:

[0188] Only the recharge demand index The hierarchical participation sorting;

[0189] Under the current growth stage, the irrigation demand index of each layer is sorted from large to small. The layer with the highest irrigation demand index is the layer with the highest irrigation priority. When there are two layers of irrigation demand index When the water retention coefficients are the same, first compare the water retention capacity coefficients of the two, and give priority to the layer with the lowest water retention capacity coefficient. When the water retention capacity coefficients are the same, then compare the water absorption capacity coefficients of the two, and give priority to the layer with the highest water absorption capacity coefficient.

[0190] S3. Input the water content and influencing parameters of the current and previous m moments of the highest priority layer into the long short-term memory network model to predict the water content of the layer at the next moment;

[0191] Collect the water content and influencing parameters of the current and previous m moments (a total of m+1 moments) of the highest priority layer to form a time series;

[0192] Use the water content and influencing parameters at m+1 moments as model input (the input length during model training is m+1), ensuring that the time step length of the input sequence is consistent with that during model training (for example, if the previous 24 hours of data is used to predict the next moment during training, the input must be data from the most recent 24 moments);

[0193] Rebuild the LSTM model structure to be exactly the same as that in the training phase (including the number of neurons in the LSTM layer, the number of layers, and the fully connected layer settings) to ensure that the model can correctly read the pre-trained parameters;

[0194] Import the optimal weight parameters of the LSTM model corresponding to the highest priority layer saved during the training phase to enable the model to have predictive capabilities;

[0195] The sorted water content and influencing parameters of the highest priority m+1 moments are input into the loaded LSTM model. The model outputs the predicted value of "water content at the next moment" based on the learned time series rules.

[0196] S4. Determine the field water holding capacity based on the soil parameters of the highest priority layer, and calculate the theoretical irrigation amount and the maximum allowable irrigation amount of the highest priority layer in combination with the suitable water content and the predicted water content at the next moment. Compare the theoretical irrigation amount and the maximum allowable irrigation amount, take the minimum value of the two as the actual irrigation amount, and correct the water content prediction value to obtain the corrected water content prediction value of the highest priority layer.

[0197] Based on the above embodiment, the field capacity is determined according to the soil parameters at the highest priority level. The specific steps are as follows:

[0198] Soil parameters include soil texture and organic matter content. Soil texture is divided into sand, clay, and loam;

[0199] Sandy soil: large but few pores, poor water retention → low field water holding capacity (usually 10%-20%);

[0200] Clay: small but numerous pores, strong water retention → high field water holding capacity (usually 25%-35%);

[0201] Loam: Uniform pore distribution, moderate water retention → medium field capacity (usually 20%-30%).

[0202] Organic matter can increase soil porosity and water retention. For every 1% increase in organic matter content, field water holding capacity can increase by about 1%-2%.

[0203] Based on the above embodiment, the specific steps and the formulas used in S4 are as follows:

[0204] ;

[0205] in, It is the theoretical irrigation amount for the highest irrigation priority layer at the current growth stage. It is the appropriate water content corresponding to the highest priority level of supplementary irrigation at the current growth stage. is the water content of the highest priority layer at the next moment in the current growth stage, is the mass of solid particles per unit volume of soil, The thickness of the highest priority layer for recharging;

[0206] On this basis, it should be noted that:

[0207] The difference is calculated only when "refilling is required";

[0208] when When the water content is predicted to be lower than the appropriate value at the next moment, , represents the “ratio of water required to achieve the appropriate water content”;

[0209] when When the water content is predicted to meet or exceed the appropriate value at the next moment, The result is 0, which means no refilling is needed;

[0210] It is to convert the “volume difference” into the “mass of water required to be added per unit volume of soil;

[0211] The total water mass required for the entire layer to reach the appropriate moisture content, that is, the theoretical supplementary irrigation amount.

[0212] ;

[0213] in, The maximum allowable irrigation amount for the highest irrigation priority layer at the current growth stage. It is the field water holding capacity of the highest priority level for supplementary irrigation at the current growth stage. The water content of the highest priority layer at the current moment;

[0214] On this basis, it should be noted that:

[0215] Field capacity ( ) is the maximum amount of capillary suspended water that the soil can hold. This refers to the maximum water content that the soil can stably maintain after irrigation or precipitation, when excess water seeps out. It is the upper limit of water available to plant roots. If this value is exceeded, water will seep down due to gravity or accumulate, resulting in decreased soil aeration and affecting root respiration.

[0216] The formula is As the upper limit of supplementary irrigation, rather than the saturated water content, its essence is to ensure the "water-air balance" of the soil;

[0217] Where, It means “the difference between the current water content and the field capacity”, that is, the difference in volumetric water content;

[0218] If the current water content If the water content is lower than the field capacity, the difference is positive, indicating the proportion of water that the layer can safely absorb; if the current water content has reached or exceeded the field capacity, the difference is negative, and the maximum allowable supplementary irrigation amount is 0.

[0219] The difference directly reflects the "safe space" for supplementary irrigation, ensuring that supplementary irrigation will not exceed the upper limit of the soil's water holding capacity.

[0220] Similar to the theoretical recharge formula, multiply by and The purpose is to convert the "volume ratio difference" into mass units. Compare the theoretical filling volume with the maximum allowable filling volume, and take the minimum value of the two as the actual filling volume, which is recorded as ;

[0221] ;

[0222] ;

[0223] in, It is the predicted value of water content after correction for the highest priority layer of supplementary irrigation in the current growth stage.

[0224] On this basis, it should be noted that:

[0225] Maximum allowable refill volume Compared with the theoretical recharge volume mentioned above Forming a constraint relationship:

[0226] The actual irrigation volume must also meet the requirements of "reaching the appropriate water content" (no less than ) and “not exceeding field capacity” (not higher than ); If the theoretical recharge volume exceeds the maximum allowable recharge volume (i.e. ), then The limit is (to avoid water accumulation); otherwise, the theoretical refilling amount shall be followed.

[0227] Corrected water content prediction value of the highest priority layer of supplementary irrigation , is the predicted water content before supplementary irrigation + the water content increment caused by supplementary irrigation, that is, the actual supplementary irrigation amount Reversely infer the increase in soil moisture content and the predicted value before irrigation By adding them together, we can finally get the "corrected water content" after irrigation.

[0228] See also Figure 2 , the present invention also provides a technical solution:

[0229] A soil moisture prediction system for dryland farmland under limited supplementary irrigation, the system being used to execute any of the above-mentioned soil moisture prediction methods for dryland farmland under limited supplementary irrigation, comprising:

[0230] The model construction module is used to divide the dryland farmland soil into different depth ranges to obtain multiple layers. Each layer corresponds to a different water retention capacity coefficient and water absorption capacity coefficient. The previous moisture content and influencing parameters of all layers are collected. A time series dataset is constructed for each layer. The moisture content and influencing parameters at m+1 consecutive moments are used as input, and the moisture content at the next moment is used as output. The long-short-term memory network model of each layer is trained separately.

[0231] The calculation module is used to determine the current growth stage of the oats, calculate the water deficit at the current moment based on the appropriate water content corresponding to the current growth stage, calculate the water change rate and trend correction coefficient based on the water content at the most recent m+1 moments, correct the water deficit using the trend correction coefficient, and perform weighted calculation based on the water retention capacity coefficient and the water absorption capacity coefficient to obtain the irrigation demand index. The layer with the highest irrigation demand index is designated as the highest irrigation priority layer.

[0232] The simulation module is used to input the water content and influencing parameters of the highest priority layer at the latest m+1 moments into the long short-term memory network model to predict the water content of the layer at the next moment;

[0233] The correction module is used to determine the field water holding capacity based on the soil parameters of the highest priority layer, combine the suitable water content and the predicted water content at the next moment, calculate the theoretical irrigation amount and the maximum allowable irrigation amount of the highest priority layer, compare the theoretical irrigation amount and the maximum allowable irrigation amount, take the minimum value of the two as the actual irrigation amount, correct the water content prediction value, and obtain the corrected water content prediction value of the highest priority layer.

[0234] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0235] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will appreciate that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by computer software, electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software depends on the specific application and design constraints of the technical solution.

[0236] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0237] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A method for predicting soil moisture in dryland farmland under limited irrigation, characterized in that: The specific steps include: S1. Divide dryland farmland soil into multiple layers according to different depth ranges. Each layer corresponds to a different water retention capacity coefficient and water absorption capacity coefficient. Collect the historical water content and influencing parameters of all layers. Construct a time series dataset for each layer. Use the water content and influencing parameters at m+1 consecutive moments as input and the water content at the next moment as output to train a long-short-term memory network model for each layer. S2. Determine the current growth stage of the oats and calculate the current water deficit based on the appropriate moisture content for the current growth stage. Calculate the moisture change rate and trend correction coefficient based on the moisture content for the most recent m+1 moments. Correct the water deficit using the trend correction coefficient. A weighted calculation is performed using the water retention capacity coefficient and the water absorption capacity coefficient to determine the irrigation requirement index. The stratum with the highest irrigation requirement index is designated as the highest irrigation priority stratum. S3. Input the water content and influencing parameters of the highest priority layer at the most recent m+1 moments into the long short-term memory network model to predict the water content of the layer at the next moment; S4. Determine the field water holding capacity based on the soil parameters of the highest priority layer, combine the suitable water content and the predicted water content at the next moment, calculate the theoretical irrigation amount and the maximum allowable irrigation amount of the highest priority layer, compare the theoretical irrigation amount and the maximum allowable irrigation amount, take the minimum value of the two as the actual irrigation amount, correct the water content prediction value, and obtain the corrected water content prediction value of the highest priority layer.

2. The soil moisture prediction method for dryland farmland under limited irrigation according to claim 1, characterized in that: The dryland farmland soil is divided into different depth ranges to obtain multiple layered soils. The specific steps are as follows: Based on the characteristic that the root distribution range of oats is 0-60cm, the dryland farmland soil is divided into several continuous depth intervals: 0-20cm, 20-40cm, and 40-60cm, corresponding to the upper layer, middle layer, and lower layer respectively.

3. The soil moisture prediction method for dryland farmland under limited irrigation according to claim 1, characterized in that: Combined with the appropriate water content corresponding to the current growth stage, the water deficit at the current moment is calculated based on the following formula: ； in, In the current growth stage, The water deficit of each layer, In the current growth stage, The appropriate water content corresponding to each layer is For the current moment The moisture content of each layer; Where, For hierarchical indexes, , is the number of layers.

4. The soil moisture prediction method for dryland farmland under limited irrigation according to claim 3, characterized in that: The moisture change rate is calculated based on the moisture content at the most recent m+1 moments, and the trend correction coefficient is calculated to correct the moisture deficit. The formula is as follows: The water content time series data consisting of m+1 moments is ,in, The m+1th moment The water content of the first layer, that is, the water content of the first layer at the current moment The moisture content of each layer, For the The moment The moisture content of each layer, is the moment index among the m+1 moments, ; ； in, From the first moment to the current moment, The average rate of change of moisture in each layer is is the time interval between two adjacent moments; ； in, In the current growth stage, The trend correction coefficient of each layer, is the modified strength coefficient when the moisture content is reduced, , is the modified strength coefficient when moisture increases, , is the absolute value of the average rate of change of moisture; when hour, ； in, In the current growth stage, Water deficit after stratification correction; when When The moisture content of each layer is equal to or higher than the suitable moisture content, and there is no need for supplementary irrigation.

5. The soil moisture prediction method for dryland farmland under limited irrigation according to claim 4, characterized in that: The supplementary irrigation demand index is obtained by weighted calculation based on the corrected water deficit, water retention capacity coefficient, and water absorption capacity coefficient. The formula is as follows: ； in, In the current growth stage, A stratified irrigation demand index, For the The water retention capacity coefficient of each layer, For the The moisture absorption capacity coefficient of each layer, 、 are the weight coefficients of water retention capacity coefficient and water absorption capacity coefficient, respectively. On this basis, .

6. The soil moisture prediction method for dryland farmland under limited irrigation according to claim 5, characterized in that: Determine the layer with the highest index as the highest priority layer for replenishment. The specific steps are as follows: Only the recharge demand index The hierarchical participation sorting; Under the current growth stage, the irrigation demand index of each layer is sorted from large to small. The layer with the highest irrigation demand index is the layer with the highest irrigation priority. When there are two layers of irrigation demand index When the water retention coefficients are the same, first compare the water retention capacity coefficients of the two, and give priority to the layer with the lowest water retention capacity coefficient. When the water retention capacity coefficients are the same, then compare the water absorption capacity coefficients of the two, and give priority to the layer with the highest water absorption capacity coefficient.

7. The soil moisture prediction method for dryland farmland under limited irrigation according to claim 6, characterized in that: The specific steps and formulas of S4 are as follows: ； in, It is the theoretical irrigation amount for the highest irrigation priority layer at the current growth stage. It is the appropriate water content corresponding to the highest priority level of supplementary irrigation at the current growth stage. is the water content of the highest priority layer at the next moment in the current growth stage, is the mass of solid particles per unit volume of soil, The thickness of the highest priority layer for recharging; ； in, The maximum allowable irrigation amount for the highest irrigation priority layer at the current growth stage. It is the field water holding capacity of the highest priority level for supplementary irrigation at the current growth stage. The water content of the highest priority layer at the current moment; Compare the theoretical recharge volume with the maximum allowable recharge volume, and take the minimum value of the two as the actual recharge volume, which is recorded as ; ； ； in, It is the predicted value of water content after correction for the highest priority layer of supplementary irrigation in the current growth stage.

8. A soil moisture prediction system for dryland farmland under limited supplementary irrigation, the system being used to execute the soil moisture prediction method for dryland farmland under limited supplementary irrigation according to any one of claims 1 to 7, characterized in that: include: The model construction module is used to divide the dryland farmland soil into different depth ranges to obtain multiple layers. Each layer corresponds to a different water retention capacity coefficient and water absorption capacity coefficient. The previous moisture content and influencing parameters of all layers are collected. A time series dataset is constructed for each layer. The moisture content and influencing parameters at m+1 consecutive moments are used as input, and the moisture content at the next moment is used as output. The long-short-term memory network model of each layer is trained separately. The calculation module is used to determine the current growth stage of the oats, calculate the water deficit at the current moment based on the appropriate water content corresponding to the current growth stage, calculate the water change rate and trend correction coefficient based on the water content at the most recent m+1 moments, correct the water deficit using the trend correction coefficient, and perform weighted calculation based on the water retention capacity coefficient and the water absorption capacity coefficient to obtain the irrigation demand index. The layer with the highest irrigation demand index is designated as the highest irrigation priority layer. The simulation module is used to input the water content and influencing parameters of the highest priority layer at the latest m+1 moments into the long short-term memory network model to predict the water content of the layer at the next moment; The correction module is used to determine the field water holding capacity based on the soil parameters of the highest priority layer, combine the suitable water content and the predicted water content at the next moment, calculate the theoretical irrigation amount and the maximum allowable irrigation amount of the highest priority layer, compare the theoretical irrigation amount and the maximum allowable irrigation amount, take the minimum value of the two as the actual irrigation amount, correct the water content prediction value, and obtain the corrected water content prediction value of the highest priority layer.

Images