A method and apparatus for assessing the risk of sudden drought in agriculture based on the dynamic water demand of crops.

Abstract

This invention discloses a method and apparatus for assessing the risk of sudden drought in agriculture based on dynamic crop water demand, addressing the problems of existing assessments that fail to incorporate dynamic crop water demand, resulting in low accuracy and poor practicality. The method involves collecting multi-source monitoring data from the target area, preprocessing it to generate standardized basic data, and then calibrating and validating it locally. An adapted crop growth model is used to simulate the dynamics of the entire growth period and soil moisture consumption. A DIT (Digital Interval Indicator) unit is constructed to couple the growth period with the time scale, and an AFDI (Agro-Acid Difference in Water Demand) index is built by combining crop water consumption and soil moisture parameters. Sudden drought events are identified and core features are extracted. The sudden drought risk is calculated using a joint probability distribution model, resulting in standardized and visualized output. This invention improves the accuracy and practicality of the assessment, providing support for agricultural drought early warning and precision irrigation.

Description

Technical Field

[0001] This invention relates to the field of agricultural disaster monitoring and risk assessment technology, and in particular to a method and apparatus for assessing the risk of sudden drought in agriculture based on the dynamic water demand of crops. Background Technology

[0002] Agricultural drought is a major natural disaster that restricts crop growth and causes a sharp decline in yield. Accurate identification and scientific assessment of agricultural drought conditions are of great significance for guiding water-saving irrigation in agriculture, reducing disaster losses, and strengthening the food security defense line.

[0003] Among existing technologies, the most widely used is the fixed-time-scale meteorological drought index method. Its process involves: pre-setting fixed time scales such as months and seasons; cumulatively calculating the total water balance of the target area and standardizing it to construct a meteorological drought index; then classifying drought levels according to fixed thresholds and statistically analyzing drought event characteristics to identify and assess agricultural drought. However, this method does not incorporate crop growth rhythms and dynamic water requirements, resulting in significant drawbacks: First, the fixed time scale does not match the crop growth period, failing to accurately reflect the degree of drought stress throughout the crop's entire growth cycle; second, it only statistically analyzes meteorological water balances without coupling them to the actual water consumption process of crops, making it difficult to reflect the real differences in water requirements at different growth stages; third, it easily overlooks sudden drought signals during critical crop growth periods, lacking sufficient ability to identify sudden droughts, which can lead to delayed drought warnings and fail to provide timely and effective support for agricultural prevention and control.

[0004] In summary, existing fixed-timescale meteorological drought index methods are not targeted enough and are seriously out of touch with the actual water requirements of crops, making it impossible to accurately identify and scientifically assess sudden agricultural droughts. There is an urgent need for a technical solution that is tailored to the realities of agricultural production and can address the above-mentioned shortcomings. Summary of the Invention

[0005] The main objective of this invention is to provide a method and apparatus for assessing the risk of sudden drought in agriculture based on the dynamic water demand of crops. This invention aims to solve the technical problems of existing methods for assessing the risk of sudden drought in agriculture, such as insufficient integration of the dynamic water demand characteristics of crops, unreasonable time scale discretization, insufficient accuracy in the quantification of sudden drought, and difficulty in directly using the assessment results to serve the precise scheduling of agricultural production and risk prevention and control decisions.

[0006] To achieve the above objectives, this invention provides a method for assessing the risk of sudden agricultural drought based on dynamic crop water demand, comprising the following steps: Acquire multi-source monitoring data of the target area, preprocess the multi-source monitoring data, and obtain basic data suitable for crop water requirement simulation; Based on the aforementioned basic data, a crop water requirement simulation model is run to simulate the crop's growth process and soil moisture changes throughout its entire growth period, thereby obtaining the crop's water consumption parameters and soil moisture parameters. A dynamic time identification unit adapted to the crop growth stage and regional precipitation pattern is constructed, and the entire crop growth period is segmented according to the dynamic time identification unit to achieve coupling and matching between the dynamic time identification unit and the crop growth stage. Using the dynamic time recognition unit as the calculation scale, and combining the water consumption parameter and the soil moisture parameter, an agricultural drought index is constructed. The agricultural drought index is then used to quantify the degree of crop water deficit in each unit. Identify sudden drought units according to preset judgment criteria, define complete sudden drought events based on the sudden drought units, and extract the core feature parameters of the sudden drought events; A risk calculation model is constructed based on the core feature parameters, and the risk of sudden drought in agriculture is assessed using the risk calculation model and the assessment results are output.

[0007] Optionally, the multi-source monitoring data includes meteorological data, basic soil data, and crop management data; the preprocessing includes standardization of the multi-source monitoring data, missing value completion, and outlier removal.

[0008] Optionally, the crop water requirement simulation model undergoes regional localization calibration and effectiveness verification; the regional localization calibration determines the parameter range by combining the genetic characteristics of the target crop with the regional agricultural production conditions, and the effectiveness verification uses the simulation errors of phenological period, biomass and soil moisture content as verification indicators.

[0009] Optionally, the duration of the dynamic time recognition unit is determined by the fluctuation range of the crop growth stage and the regional precipitation change cycle, and the duration of the dynamic time recognition unit does not exceed 1 / 3 of the duration of the corresponding crop growth stage.

[0010] Optionally, the dynamic time recognition unit assigns each reproductive stage to the dominant reproductive stage with the highest percentage of days within the unit, based on the percentage of days each reproductive stage has in the unit.

[0011] Optionally, the agricultural drought index is constructed based on the cumulative evapotranspiration, cumulative precipitation, and average soil moisture content within the dynamic time recognition unit; when the precipitation meets the crop's water requirements, the agricultural drought index is 0, and the higher the degree of water deficit, the larger the agricultural drought index value.

[0012] Optionally, the core feature parameters include the duration and intensity of the drought; the risk calculation model is constructed based on the marginal distribution and Copula joint distribution of the core feature parameters, and is used to calculate the probability and recurrence period of the drought.

[0013] Optionally, the marginal distribution is selected as the optimal distribution type using the Bayesian information criterion, and the parameters of the Copula joint distribution are determined and their rationality verified based on the correlation coefficient.

[0014] To achieve the above objectives, the present invention also provides an agricultural drought risk assessment device oriented towards crop dynamic water demand, comprising: The data preprocessing module is used to acquire and preprocess multi-source monitoring data to obtain basic data, which is then sent to the growth simulation module and the time coupling module respectively. The growth simulation module receives the basic data output by the data preprocessing module, runs the crop water requirement simulation model, outputs water consumption parameters and soil moisture parameters, and sends them to the index calculation module. The time coupling module is used to receive the basic data output by the data preprocessing module, extract crop growth period information, construct dynamic time recognition units and couple them with crop growth stages, and output time matching parameters to the index calculation module. The index calculation module is used to receive water consumption parameters and soil moisture parameters output by the growth simulation module and time matching parameters output by the time coupling module, construct and calculate the agricultural drought index to quantify the degree of crop water deficit, and send the quantified data to the drought identification module. The sudden drought identification module is used to receive the quantitative data output by the index calculation module, identify sudden drought units, define sudden drought events, and extract core feature parameters, and then send the core feature parameters to the risk assessment module. The risk assessment module receives the core feature parameters output by the drought identification module, constructs a risk calculation model, and outputs the agricultural drought risk assessment results.

[0015] To achieve the above objectives, the present invention also provides a computer-readable storage medium, characterized in that the storage medium stores a computer program, which, when executed by a processor, implements the above-described method for assessing the risk of agricultural drought oriented towards dynamic crop water demand.

[0016] The beneficial effects that this invention can achieve are as follows: This invention proposes a method and apparatus for assessing agricultural drought risk based on dynamic crop water demand. By collecting and preprocessing multi-source crop growth-related data, and combining it with a growth model adapted to the target crop to simulate the dynamic growth of the entire growth period and soil moisture consumption characteristics, it solves the problems of existing assessment methods failing to fully incorporate dynamic crop water demand characteristics and having poor data adaptability. By constructing an adaptive dynamic time identification unit (DIT), it achieves precise coupling between crop growth stages and time scales, avoiding drought identification bias caused by the discretization of traditional fixed time scales and improving the accuracy of drought event identification. By constructing an agricultural drought index (AFDI) and combining it with soil moisture consumption parameters to quantify the degree of water deficit, it solves the problems of insufficient quantification and low accuracy of core drought driving indicators, making drought judgment more scientific and targeted. By fitting a joint probability distribution model of core feature parameters, it calculates the probability of drought risk at different levels, generates standardized and reusable assessment results, and provides visual output, solving the problems of weak practicality of assessment results and difficulty in serving production decisions.

[0017] Specifically, this scheme ensures the standardization and reliability of input data through multi-source data preprocessing, providing a solid foundation for subsequent simulation and evaluation; parameter calibration and validity verification of the crop growth model ensure the accuracy of crop growth dynamics and soil moisture consumption simulation, achieving precise characterization of crop dynamic water demand characteristics; the adaptive design of DIT units adapts to the fluctuations in crop growth stages and precipitation changes, enabling each time unit to accurately correspond to the crop growth stage and ensuring the synchronization of drought identification with crop growth status; the AFDI index combined with regional correction coefficients eliminates errors caused by regional differences, improving the universality and accuracy of drought quantification; the construction of a joint probability distribution model enables quantitative assessment of drought risks at different levels, and the standardized assessment results and visualization output can directly provide a scientific basis for agricultural drought monitoring and early warning, precise irrigation scheduling, and regional agricultural drought risk prevention and control decisions, effectively improving the regional agricultural drought resistance and disaster reduction capabilities and ensuring stable agricultural production. Attached Figure Description Attached image description: Figure 1 This is a flowchart of the agricultural drought risk assessment method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the agricultural drought identification process according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the DIT partitioning method according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the device structure according to an embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] In related technologies, agricultural drought risk assessment often relies on meteorological drought indices at fixed time scales (such as monthly, ten-day, and seasonal), failing to fully integrate dynamic crop water requirements and growth rhythms, resulting in several shortcomings: First, the fixed time scale does not match the crop growth stage, easily leading to misjudgment or delayed identification of drought; second, it does not couple with actual crop water consumption, resulting in low assessment accuracy; third, quantitative indicators do not incorporate soil moisture characteristics, failing to reflect the core driving factors of drought; and fourth, the assessment results are singular, lack standardized output, have poor practicality, and are difficult to use to support production decisions. Furthermore, crop growth model parameters lack localized calibration, resulting in large simulation biases, reducing assessment reliability, and failing to provide accurate technical support.

[0021] refer to Figure 1 , 2 As shown in Figure 3, to solve the above-mentioned technical problems, the agricultural drought risk assessment method based on crop dynamic water demand adopted in this embodiment includes the following core steps: multi-source data acquisition and preprocessing, crop growth model calibration and growth dynamic simulation, construction of dynamic time identification unit (DIT) and coupling with growth period, construction and calculation of agricultural drought index (AFDI), drought event identification and core feature extraction, drought risk assessment and standardized output. The specific implementation process of each step is as follows: Step 1, Multi-source data acquisition and preprocessing: Acquire multi-source monitoring data of the target area and preprocess the multi-source monitoring data to obtain basic data for adapting the crop growth model; The core of this step is to collect multi-source data related to summer maize growth in the target area, and generate a standardized dataset that meets the input standards of crop growth models through standardized preprocessing, so as to provide reliable basic data support for subsequent crop growth simulation and drought assessment.

[0022] Multi-source data collection: Data from multiple sources was collected for 60 consecutive years from 1961 to 2020 in the Huaibei Plain of Anhui Province. This included three core data categories: First, meteorological data, derived from observations at surface and automatic weather stations within the region. Key data collected included daily precipitation, daily maximum temperature, daily minimum temperature, and sunshine duration. Daily radiation missing for some stations was calculated using the Angstrom empirical formula calibrated for the Huaibei Plain region. Quality control was implemented on the collected meteorological data, removing outliers such as daily precipitation >100mm and daily maximum temperature >40℃ to ensure data reliability. Second, crop management data, including summer maize variety parameters (genetic characteristics of Zhengdan 958), planting density (4500 plants / mu), and fertilization records (200kg / mu basal fertilizer and 150kg / mu topdressing), all derived from years of field management data measured within the region. Third, the basic soil parameters were obtained through field sampling and laboratory testing. The sampling density was set at 5 samples per square kilometer to meet the needs of soil heterogeneity characterization in the Huaibei Plain, and the testing error was controlled within 5%. Specific parameters include soil moisture content, field water holding capacity (25%), wilting coefficient (8%), soil bulk density (1.3 g / cm³), and organic matter content (1.2%) in the 0-20cm, 20-40cm, and 40-60cm layers.

[0023] Data preprocessing: Linear interpolation was used to complete missing values ​​in meteorological data, and mean interpolation was used to complete missing values ​​in field management data to ensure data continuity and integrity. Finally, a standardized dataset conforming to the input standards of the Crop Growth Model (CERES-Maize) was generated, covering three major categories: daily meteorological data, crop management parameters, and basic soil parameters, providing unified and standardized basic data for subsequent steps.

[0024] The multi-source data includes at least long-term meteorological data and crop management data; the preprocessing includes at least missing value completion processing; the crop management data includes at least crop variety parameters, planting density, fertilization records, and basic soil parameters; the basic soil parameters include stratified soil moisture content, field water holding capacity, wilting coefficient, soil bulk density, and organic matter content, which are obtained through field sampling and laboratory testing. The sampling density is adapted to the regional soil heterogeneity characterization requirements, and the detection error is controlled within a preset range.

[0025] Based on multi-source data acquisition and preprocessing, the specific requirements for data acquisition and preprocessing are further clarified as follows: Specific acquisition of multi-source data: Long-term meteorological data are selected from observation data of surface meteorological stations and regional automatic meteorological stations in the Huaibei Plain of Anhui Province. Crop management data are selected from multi-year field management and measured data of summer maize in the region. The crop management data specifically includes: crop variety parameters (genetic characteristic parameters of summer maize variety Zhengdan 958), planting density (4500 plants / mu), fertilization records (basal fertilizer 200 kg / mu, topdressing 150 kg / mu), and basic soil parameters, ensuring that the data sources are consistent with the actual production in the region and that the data types comprehensively cover the assessment needs.

[0026] Soil baseline parameter acquisition: Core parameters such as soil moisture content (0-10cm, 10-20cm, 20-30cm, 30-40cm), field water holding capacity (25%), wilting coefficient (8%), soil bulk density (1.3g / cm³), and organic matter content (1.2%) were obtained through a combination of field sampling and laboratory testing. The sampling density was set to 5 samples / square kilometer to accurately meet the needs of soil heterogeneity characterization in the Huaibei Plain, while controlling the detection error within 5% to ensure the authenticity and accuracy of soil baseline parameters.

[0027] Data preprocessing: For missing values ​​in meteorological data, linear interpolation was used to complete them; for missing values ​​in field management data, mean interpolation was used to complete them, ensuring the continuity and integrity of the data. Finally, a standardized dataset that conforms to the input standards of the Crop Growth Model (CERES-Maize) was generated, providing reliable data support for subsequent steps.

[0028] The refinement of the multi-source data acquisition and preprocessing process improves data quality and adaptability, and makes up for the shortcomings of existing methods such as vague data sources and non-standard processing procedures. The specific benefits are as follows: it clarifies the specific types and scope of multi-source data, deeply integrates meteorological data with crop management data, especially refines the specific content of crop management data, solves the problems of vague data sources and single data types in existing methods, makes the collected data more targeted, can fully support crop growth simulation and drought assessment, and ensures that the assessment results are consistent with the actual agricultural production.

[0029] The specific types, acquisition methods, sampling densities, and detection error requirements of basic soil parameters are refined. By combining field sampling with laboratory testing, the authenticity and accuracy of soil parameters are ensured. The sampling density is adapted to the regional soil heterogeneity to avoid simulation errors caused by uneven soil parameters. The detection error is controlled within the preset range, which further improves the reliability of soil parameters and provides accurate basic parameters for subsequent crop water consumption simulation and AFDI index calculation.

[0030] By clearly defining the specific preprocessing methods (missing value completion), the problem of missing data is eliminated, ensuring that the standardized dataset meets the input standards of the crop growth model, avoiding model simulation deviations caused by data anomalies, providing a guarantee for the smooth implementation of subsequent steps, and further improving the reliability and accuracy of the entire evaluation method.

[0031] Meteorological data are derived from one or more of the following sources: ground meteorological station observation data, regional automatic weather station data, reanalysis datasets, and gridded meteorological driven datasets. Outliers are removed after quality control before use. The meteorological data includes at least daily precipitation, daily maximum temperature, daily minimum temperature, and sunshine duration. Daily radiation can be calculated by combining sunshine duration with an empirical formula calibrated for the target area.

[0032] The sources, types, processing methods, and methods for supplementing missing meteorological data are as follows: Meteorological data sources: Observational data from ground meteorological stations in Anhui Province and data from regional automatic meteorological stations are selected as the main sources of meteorological data to ensure the timeliness and accuracy of the data; In response to the problem that some stations cannot directly obtain daily radiation data, the Angstrom empirical formula calibrated for the Huaibei Plain region is used for conversion to make up for the data gaps.

[0033] The specific types of meteorological data include: precise collection of four core meteorological elements: daily precipitation P (mm), daily maximum temperature Tmax (°C), daily minimum temperature Tmin (°C), and sunshine duration Rt (h). Daily radiation Ra (MJ / m²) is obtained through conversion to ensure that the meteorological data can fully meet the input requirements of the crop growth model (CERES-Maize) and avoid redundant data that would increase processing costs.

[0034] Meteorological data quality control: Strict quality control is carried out on the collected meteorological data to remove outliers that exceed reasonable ranges (such as daily precipitation > 100 mm, daily maximum temperature > 40℃, etc.) to avoid interference from outliers in subsequent preprocessing, model simulation and drought assessment, and to ensure the reliability of meteorological data. After quality control, the data is used for subsequent preprocessing and model input.

[0035] Daily radiation conversion: using the Angstrom empirical formula: ; In the formula a s =0.25、b s =0.50 (based on FAO recommendation) The maximum sunshine duration (calculated based on the latitude and seasonal characteristics of the Huaibei Plain). , An empirical coefficient related to the atmosphere is used to ensure that the converted daily radiation is consistent with the actual radiation situation in the region, and to avoid simulation errors in crop evaporation and transpiration caused by deviations in radiation data.

[0036] The refinement of the multi-source data acquisition and preprocessing process further enhances the integrity, reliability, and adaptability of meteorological data. The specific benefits are as follows: it clarifies the multiple sources of meteorological data, allowing the use of various data such as ground meteorological stations and regional automatic weather stations, thereby improving the flexibility and convenience of data acquisition, solving the problems of data loss and incomplete coverage that may exist with a single data source, and ensuring the integrity and representativeness of meteorological data.

[0037] Clearly define the specific types of meteorological data, and select the core meteorological elements (precipitation, temperature, sunshine duration) required for crop growth simulation in a targeted manner. Ensure that the meteorological data can meet the input requirements of the crop growth model, avoid redundant data that increases processing costs, and at the same time ensure the relevance and accuracy of the model simulation.

[0038] The method for converting daily radiation was clarified, and the empirical formula was required to be calibrated for the target area. This solved the problem that some stations could not directly obtain daily radiation, while ensuring that the conversion results were consistent with the actual situation in the region. This avoided the simulation error of crop evapotranspiration caused by the deviation of radiation data, and further improved the accuracy of crop growth simulation.

[0039] Adding meteorological data quality control steps to remove outliers ensures the reliability of meteorological data and prevents outlier data from interfering with subsequent preprocessing, model simulation, and drought assessment, thus guaranteeing the stability and accuracy of the entire assessment process.

[0040] Step 2, Crop growth model calibration and growth dynamics simulation: A crop growth model calibrated and validated for regional localization was adopted. Based on the aforementioned basic data, the dynamic growth of crops throughout their entire growth period and the soil moisture consumption process were simulated, and crop moisture consumption parameters and soil moisture parameters were output. The core of this step is to select a crop growth model that is suitable for the agronomic characteristics of summer maize. After parameter calibration and validity verification, the growth dynamics of summer maize throughout its entire growth period are simulated based on a standardized dataset, and key crop growth indicators and soil moisture consumption characteristic parameters are output.

[0041] Model selection: The CERES-Maize model (belonging to the DSSAT agricultural extension decision support system) was selected. This model can accurately simulate the phenological process, evapotranspiration and soil moisture consumption and other core indicators of maize throughout its growth period, and is suitable for the agronomic characteristics of summer maize and the actual agricultural production in the Huaibei Plain.

[0042] Parameter calibration: The GLUE (Generalized Likelihood Uncertainty Estimation) method was used as the parameter optimization method. Combining the genetic characteristics of summer maize Zhengdan 958 with the actual agricultural production in the Huaibei Plain, the prior distribution range of the genetic parameters (P1, P2, P5, G2, G3, PHINT) to be calibrated was determined. The Latin hypercube sampling method was used to randomly generate 10,000 parameter sets from the prior distribution to ensure that the parameters cover a reasonable range and meet the physiological constraints of summer maize growth. Each set of parameters was input into the model, and the Gaussian likelihood function was calculated in combination with the measured phenological period, yield and soil moisture content to determine the weight of each parameter set and construct the posterior distribution. The parameter set with the smallest error between the simulated value and the observed value and which conforms to the crop growth law was selected as the optimal parameter set.

[0043] Validation: Absolute relative error (ARE) and relative root mean square error (RRMSE) were used as validation indicators, with preset error thresholds of ARE ≤ 10% and RRMSE ≤ 15%. Data from 2016 to 2020 were selected as validation years. The optimal parameter set after calibration was input into the model to simulate the growth process of summer maize. The simulated values ​​were compared with the measured values. After verification, the errors of phenological period, yield and soil moisture content all met the preset thresholds, and the model was determined to have regional application reliability.

[0044] Growth dynamics simulation: The standardized dataset generated after multi-source data preprocessing is input into the calibrated and validated CERES-Maize model to simulate the growth dynamics of summer maize throughout its entire growth period from 1961 to 2020. The model outputs key indicators such as daily evapotranspiration, phenological processes (seedling stage, jointing stage, tasseling stage, milk stage, maturity stage), and soil moisture consumption characteristics, providing core data for subsequent drought identification and index calculation.

[0045] The parameter calibration employs conventional parameter optimization methods in the field. It combines the genetic characteristics of the target crop with the actual agricultural production conditions of the planting area to determine the prior distribution range, generating a parameter set that meets the physiological constraints of crop growth. The parameter weights are determined using the likelihood function, and a posterior distribution is constructed. The parameter set with the smallest error between simulated and measured values, conforming to crop growth patterns, is selected as the optimal parameter set. The effectiveness verification uses conventional model verification indicators in the field. When the simulated and measured values ​​of crop phenological period, yield, and soil moisture content meet preset error thresholds, the model is deemed to have regional application reliability. Preferably, the verification indicators include absolute relative error (ARE) and relative root mean square error (RRMSE), with preset error thresholds of ARE ≤ 10% and RRMSE ≤ 15%. The parameter optimization method is preferably the GLUE method, and the parameter set contains no fewer than 10,000 sets.

[0046] Based on crop growth process simulation, the specific procedures, methods, and standards for parameter calibration and validity verification are further clarified as follows: Parameter calibration: The GLUE (Generalized Likelihood Uncertainty Estimation) method was selected as the parameter optimization method. Combining the genetic characteristics of the summer maize variety Zhengdan 958 with the actual agricultural production in the Huaibei Plain, the prior distribution range of the genetic parameters (P1, P2, P5, G2, G3, PHINT) to be calibrated was determined. The Latin hypercube sampling method was used to randomly generate 10,000 parameter sets from the prior distribution to ensure that the parameters cover a reasonable range and meet the physiological constraints of summer maize growth, thus avoiding the parameters from deviating from the actual growth law.

[0047] Weight determination and posterior distribution construction: Each set of parameters is input into the CERES-Maize model to obtain the simulation output results. Combined with the measured phenological period, yield and soil moisture content data of the region, the Gaussian likelihood function is calculated. The weights of each parameter set are determined according to the likelihood value, and then the posterior distribution of the parameters is constructed. The mean and variance of the posterior distribution are calculated. Finally, the parameter set with the smallest error between the simulated value and the observed value and which conforms to the crop growth law is selected as the optimal parameter set to ensure the scientificity and accuracy of parameter calibration.

[0048] Validation: Absolute relative error (ARE) and relative root mean square error (RRMSE) were used as validation indicators, with preset error thresholds of ARE ≤ 10% and RRMSE ≤ 15%. Data from 2016 to 2020 were selected as validation years. The optimal parameter set after calibration was input into the model to simulate the growth process of summer maize. The simulated values ​​were compared with the measured values. When the errors of phenological period, yield and soil moisture content all met the preset thresholds, the model was deemed to have regional application reliability, ensuring that the model can accurately adapt to the actual growth of summer maize in the Huaibei Plain.

[0049] The refinement of parameter calibration and validity verification further enhances the reliability and adaptability of crop growth models, solving the problems of non-standard parameter calibration and large simulation bias in existing methods. Specific beneficial effects include: clarifying the specific methods for parameter calibration, the requirements for parameter set generation, and the criteria for selecting the optimal parameter set; prioritizing the GLUE method and requiring a parameter set of no less than 10,000 sets to ensure the scientific rigor and comprehensiveness of parameter calibration; and determining the prior distribution range by combining the genetic characteristics of the target crop and the actual regional production conditions to avoid parameter calibration deviating from reality and ensure that the optimal parameter set closely matches the actual crop growth in the region, thus solving the problems of lack of scientific procedures and large simulation bias in existing methods.

[0050] The specific indicators and preset error thresholds for validity verification were clearly defined. ARE and RRMSE were selected as verification indicators to quantify the accuracy of model simulation. The preset error thresholds (ARE≤10% and RRMSE≤15%) provide a clear standard for judging model reliability, avoid subjectivity in model verification, ensure the model has regional application reliability, and provide accurate model support for subsequent crop growth dynamic simulation and drought assessment.

[0051] By constructing the posterior distribution of parameters and determining the parameter weights, the rationality and accuracy of the optimal parameter set were further improved, ensuring that the crop growth dynamics and soil moisture consumption characteristics simulated by the model were highly consistent with the measured values. This provided accurate basic data for subsequent AFDI index calculation and drought identification, and improved the accuracy of the entire evaluation method.

[0052] like Figure 2 As shown, in step 3, the construction of the Dynamic Time Identification Unit (DIT) is coupled with the reproductive period: A dynamic time identification unit (DIT) adapted to crop growth stages and regional precipitation variation patterns is constructed. The entire crop growth period is segmented according to the DIT unit, so as to achieve a unique correspondence between the DIT unit and the crop growth stage. The core of this step is to construct an adaptive dynamic time identification unit (DIT). Using the DIT as a discretized time scale, the entire growth period of summer maize is segmented to achieve precise coupling between crop growth stages and DIT units, ensuring that each DIT unit uniquely corresponds to a crop growth stage.

[0053] Determination of DIT length: Based on 10 consecutive years of phenological observation data of summer maize from 1961 to 2020, the time fluctuation range of each growth stage was determined (seedling stage 25-30 days, jointing stage 20-25 days, tasseling stage 15-20 days, milk stage 30-35 days, maturity stage 15-20 days); combined with cross-analysis of the typical minimum variation cycle (10 days) of precipitation process in the Huaibei Plain, the DIT length was determined to be 5 days; it was verified that the DIT length (5 days) does not exceed 1 / 3 of the duration of any crop growth stage (e.g., the shortest seedling stage is 25 days, 1 / 3 is about 8.3 days, and 5 days meets the requirement), ensuring that the DIT unit can accurately correspond to the crop growth stage.

[0054] DIT Unit Segmentation and Assignment: The entire growing season of summer maize is discretized into 5-day units to form continuous DIT units. For DIT units that span two growth stages, they are assigned to the growth stage with the higher percentage of days (e.g., if a DIT unit has 3 days in the seedling stage and 2 days in the jointing stage, it is assigned to the seedling stage). This ensures that each DIT unit uniquely corresponds to one crop growth stage, achieving precise coupling between crop growth stages and DIT units.

[0055] DIT Adaptive Adjustment: DIT is an adaptive time window unit. If there are fluctuations in the growth stage of summer maize in a certain year (such as the seedling period being extended to 32 days) or changes in the regional precipitation variation cycle (such as the minimum precipitation cycle becoming 7 days), the length of DIT will be dynamically adjusted (adjusted to 6 days, still not exceeding 1 / 3 of the seedling period length) to ensure that the DIT unit always adapts to the crop growth stage and precipitation variation pattern, avoiding the identification error of sudden drought caused by a fixed time scale.

[0056] The length of the DIT (Digital Interval Time) is determined as follows: the temporal fluctuation range of each growth stage is determined based on the continuous multi-year phenological observation data of the target crop, and the appropriate DIT length is determined by cross-analysis of the typical minimum variation cycle of precipitation processes in the target area. The length of the DIT does not exceed 1 / 3 of the duration of any crop growth stage. DIT units that span two growth stages are assigned to the growth stage with the higher proportion of days. The DIT is an adaptive time window unit, and its length can be dynamically adjusted according to the fluctuation of crop growth stages and the precipitation variation cycle.

[0057] Based on the reconstruction of drought identification units, the specific methods, attribution rules, and adaptive adjustment mechanisms for DIT unit construction are further clarified as follows: Determination of DIT Length: Based on 10 consecutive years of phenological observation data of summer maize from 1961 to 2020, the time fluctuation range of each growth stage was determined (seedling stage 25-30 days, jointing stage 20-25 days, tasseling stage 15-20 days, milk stage 30-35 days, maturity stage 15-20 days). Cross-analysis was conducted using the typical minimum variation cycle (10 days) of precipitation processes in the Huaihe Plain, and the DIT length was determined to be 5 days. At the same time, it was verified that the DIT length (5 days) does not exceed 1 / 3 of the duration of any crop growth stage (e.g., the shortest seedling stage is 25 days, 1 / 3 is about 8.3 days, and 5 days meets the requirement), ensuring that the DIT unit can accurately match the crop growth stage and avoid the identification bias of sudden drought caused by unreasonable time scale.

[0058] DIT Unit Assignment: The entire growing season of summer maize is discretized into 5-day units to form continuous DIT units. For DIT units that span two growth stages, they are assigned to the growth stage with the higher percentage of days (e.g., if a DIT unit has 3 days in the seedling stage and 2 days in the jointing stage, it is assigned to the seedling stage). This ensures that each DIT unit uniquely corresponds to one crop growth stage, achieving precise coupling between crop growth stages and DIT units.

[0059] DIT Adaptive Adjustment: DIT is an adaptive time window unit. If there are fluctuations in the growth stage of summer maize in a certain year (such as the seedling period being extended to 32 days) or changes in the regional precipitation variation cycle (such as the minimum precipitation cycle becoming 7 days), the length of DIT will be dynamically adjusted (adjusted to 6 days, still not exceeding 1 / 3 of the seedling period), to ensure that the DIT unit always adapts to the crop growth stage and precipitation variation pattern, thereby improving the versatility and flexibility of the method.

[0060] By refining the construction of DIT units, the adaptability and flexibility of DIT units have been further improved, solving the problem of mismatch between traditional fixed time scales and crop growth stages. The specific benefits are as follows: the specific method for determining the length of DIT is clarified, and the cross-analysis of the phenological fluctuation range of crops over many years and the regional precipitation change cycle is combined to ensure that the length of DIT is both consistent with the crop growth stage and can capture the precipitation change pattern. This solves the problem of mismatch between traditional fixed time scales and crop growth stages, and enables the identification of sudden drought to accurately correspond to the differences in water demand of crops at different growth stages.

[0061] Clearly define the limitations on the length of the DIT (not exceeding 1 / 3 of the duration of any growth stage) to avoid the DIT unit being too long and unable to capture short-term drought signals during the critical growth period of crops, or too short and causing data redundancy and reduced evaluation efficiency. Ensure the rationality of the DIT unit and improve the accuracy and efficiency of drought identification.

[0062] Clearly define the rules for classifying DIT units that span two growth stages, ensuring that each DIT unit uniquely corresponds to one growth stage. This avoids errors in identifying sudden drought due to ambiguous classification, ensures that sudden drought assessment accurately corresponds to crop growth stages, and improves the relevance of assessment results.

[0063] By clarifying the adaptive characteristics of the DIT unit, its length can be dynamically adjusted according to the fluctuations in crop growth stages and the cycle of precipitation changes. This allows the DIT unit to adapt to the actual crop growth and precipitation in different years and regions, improving the versatility and flexibility of the entire assessment method and avoiding assessment bias caused by a fixed DIT length.

[0064] Step 4, Construction and calculation of the Agricultural Strike Drought Index (AFDI): Using the DIT unit as the calculation scale, the agricultural drought index AFDI is constructed by combining the crop water consumption parameter and the soil moisture parameter. The degree of crop water deficit in each DIT unit is quantified by the AFDI index. The core of this step is to use the DIT unit as the core calculation unit, combine soil moisture consumption characteristic parameters to calculate the difference between crop water supply and demand, construct the Agricultural Strike Drought Index (AFDI), and calculate the AFDI index value of each DIT unit to quantify the core driving indicators of strike drought occurrence.

[0065] AFDI Index Construction: The AFDI index is used to quantify the degree of crop water deficit within a DIT unit. The calculation formula is as follows: ; in For the first The first DIT unit Rainfall of the day, For the first The first DIT unit Evapotranspiration per day, where n is the number of days in the DIT (Diverterless Water Deficiency Index). This example focuses on the Huaibei Plain (primarily loam soil, plain terrain). The AFDI index is a dimensionless quantity with a value range of [0,1]. A higher value indicates a greater degree of water deficit and a higher risk of sudden drought. When there is no moisture deficit in each DIT unit An index value of 0 clearly defines a drought-free state.

[0066] AFDI index calculation: Using each DIT cell as the core calculation unit, the cumulative evapotranspiration within each DIT cell output from the crop growth dynamics simulation is extracted. and the cumulative precipitation within each DIT unit after pretreatment Substituting the values ​​into the AFDI index calculation formula, the AFDI index value for each DIT unit from 1961 to 2020 is calculated unit by unit, accurately quantifying the degree of crop water deficit in each time unit, and providing core quantitative indicators for subsequent judgment of drought conditions.

[0067] Index characteristics: The AFDI index is a dimensionless quantity with a fixed value range of [0,1]. The larger the value, the higher the degree of water deficit and the greater the risk of flash drought. It can clearly quantify the potential risk of flash drought in each DIT unit, providing a clear and unified basis for subsequent flash drought status determination, and facilitating comparative analysis of different units and regions.

[0068] The refinement of the AFDI index clarifies the calculation logic, formula, and parameter meanings of the index, further improving the accuracy and universality of drought quantification and solving the problem of poor specificity of existing drought quantification indicators. The specific beneficial effects are as follows: It clarifies that the core function of the AFDI index is to quantify the degree of crop water deficit within the DIT unit, and establishes a positive correlation between the index and the degree of water deficit. When there is no water deficit, the index value is 0, making the meaning of the index clear and intuitive, and solving the problem that existing drought quantification indicators are vague in meaning and cannot accurately represent the degree of water deficit.

[0069] The AFDI index provides a specific calculation formula that integrates three core parameters: crop evapotranspiration, precipitation, and soil moisture content. It considers both meteorological water supply and crop water consumption and soil moisture consumption characteristics, enabling the index to accurately reflect the actual water deficit of crops and solving the problem that existing indicators only consider meteorological data and are disconnected from actual crop water consumption.

[0070] The index is a dimensionless quantity with a fixed value range of [0,1], which facilitates the setting of subsequent drought thresholds and the comparison of drought severity in different DIT units and regions. It provides a standardized quantitative indicator for drought status determination and drought intensity extraction, thereby improving the standardization of drought identification and assessment.

[0071] Step 5: Identification of sudden drought events and extraction of core features Based on a preset AFDI index threshold, sudden drought units are identified, complete sudden drought events are defined, and the core feature parameters of the sudden drought events are extracted. The core of this step is to determine the drought state of each DIT unit based on the preset AFDI drought threshold, identify spatiotemporally continuous drought DIT units to define complete drought events, and extract their core feature parameters.

[0072] Sudden drought status determination: 20% of crop evapotranspiration intensity is used as the drought threshold. The preset AFDI drought threshold is AFDI≥0.2. Combined with the AFDI index value of each DIT unit, the sudden drought status of each DIT unit is determined: when AFDI≥0.2, the DIT unit is determined to be a sudden drought unit; when AFDI<0.2, it is determined to be a non-sudden drought unit.

[0073] Definition of a complete drought event: Identify spatiotemporally continuous drought DIT units and define continuous drought DIT units as a complete drought event; if the interval between two drought DIT units does not exceed 1 non-drought DIT unit, they are determined to be the same drought event; if the interval exceeds 1 non-drought DIT unit, they are determined to be two independent drought events.

[0074] Core feature parameter extraction: For each complete drought event, core feature parameters are extracted, including drought duration (the number of DIT units contained in the drought event), drought intensity (the average AFDI index of all DIT units in the drought event), drought occurrence time (the crop growth stage corresponding to the first DIT unit of the drought event), and drought spatial distribution range (the specific area of ​​the Huaihe Plain covered by the drought event). This forms a feature parameter dataset for all drought events from 1961 to 2020, providing accurate feature data for subsequent risk assessment.

[0075] Step 6, Sudden Drought Risk Assessment and Standardized Output: Based on the core feature parameters, a probability distribution model is constructed to calculate the occurrence risk of different levels of sudden drought and output the agricultural sudden drought risk assessment results.

[0076] The core of this step is to fit the marginal distribution of the core characteristic parameters of sudden drought and construct a joint probability distribution model, calculate the probability of occurrence of different levels of sudden drought, form standardized and reusable agricultural sudden drought risk assessment results, and complete the visualization output.

[0077] Marginal distribution fitting: Based on extracted drought duration (d) AFD ) and intensity (s) AFD The sequence was analyzed using the Bayesian Information Criterion (BIC) as the distribution selection method. From various distributions including Weibull, Gamma, Lognormal, and Exponential, the distribution with the smallest BIC value was selected as the optimal marginal distribution. After fitting, the duration of the sudden drought was determined to be d... AFD The optimal marginal distribution is the Weibull distribution, and the drought intensity s AFD The optimal marginal distribution is the Gamma distribution.

[0078] Joint probability distribution model construction: The Gumbel Copula function is selected to construct the joint distribution of dAFD and sAFD. This function can effectively describe the right-tail correlation between the two (i.e., the correlation of long-duration, high-intensity sudden drought events). The parameter θ of the Gumbel Copula function is calculated using Kendall's τ correlation coefficient. In this embodiment, τ=0.72, and θ=2.86 is calculated. The rationality and stability of the parameter θ are verified to ensure that the parameter can accurately describe dAFD. AFD With s AFD The correlation ensures the reliability of the joint distribution model.

[0079] Calculation and classification of drought risk probability: Based on the joint probability distribution model, the probability and recurrence period of drought occurrence under different combinations of duration thresholds (2, 3, and 4 DIT units, corresponding to 10 days, 15 days, and 20 days) and intensity thresholds (0.3, 0.5, and 0.7) are calculated. The drought risk level is classified according to the AFDI index value and drought duration, specifically: mild drought (occurrence probability P≤0.3, corresponding to AFDI≥0.2 and duration≤15 days), moderate drought (0.3<P≤0.6, corresponding to AFDI≥0.3 and duration15-20 days), and severe drought (P>0.6, corresponding to AFDI≥0.5 and duration≥20 days).

[0080] Standardized Results and Visualized Output: Standardized and reusable agricultural drought risk assessment results will be generated, including the probability of drought occurrence, recurrence period, and risk level classification results under different threshold combinations; outputting spatial scale risk distribution maps, risk level zoning maps (intuitively presenting the drought risk levels of different regions of summer maize in the Huaibei Plain), and temporal scale risk change sequences (presenting the interannual changes in drought risk from 1961 to 2020); the assessment results can be directly applied to agricultural drought monitoring and early warning, precision irrigation scheduling (developing irrigation plans for high-risk areas and growth stages), and regional agricultural drought risk prevention and control decisions, providing a scientific basis for water management during the key growth period of summer maize and improving the regional agricultural drought prevention and control capabilities.

[0081] The marginal distribution fitting adopts conventional distribution screening methods in this field, preferably Bayesian Information Criterion (BIC) to screen the optimal distribution type; the joint probability distribution model is constructed using a Copula function, preferably the GumbelCopula function, and the parameters of the Copula function are obtained by calculating the correlation coefficient and verifying its rationality and stability, preferably by calculating the parameters using Kendall's τ correlation coefficient.

[0082] Based on the risk assessment of sudden agricultural drought, the specific methods, parameter calculations, and verification procedures for marginal distribution fitting and joint probability distribution model construction are further clarified as follows: Marginal distribution fitting: Based on the duration of sudden drought d AFD With intensity s AFD The sequence was analyzed using the Bayesian Information Criterion (BIC) as the distribution selection method. From various common distributions such as Weibull, Gamma, Lognormal, and Exponential, the distribution with the smallest BIC value was selected as the optimal marginal distribution. Through actual fitting, the duration of the sudden drought was determined to be d... AFD The optimal marginal distribution is the Weibull distribution, and the drought intensity s AFD The optimal marginal distribution is the Gamma distribution, which ensures that the marginal distribution can accurately fit the sample patterns of the drought feature parameters.

[0083] Joint probability distribution model construction: The GumbelCopula function is used to construct d. AFD and s AFD The joint distribution of the function can effectively describe the right-tail correlation between the two (i.e., the correlation between long-duration, high-intensity drought events), which meets the actual needs of drought risk assessment for extreme events and makes up for the shortcomings of traditional distributions in accurately describing the correlation between two variables.

[0084] Copula function parameter determination and verification: The parameter θ of the Gumbel Copula function is calculated using Kendall's τ correlation coefficient. In this example, τ = 0.72, and θ = 2.86 is calculated. The rationality and stability of the parameter θ are verified to ensure that the parameter can accurately describe d. AFD With s AFD To ensure the reliability and stability of the joint distribution model, the correlation between parameters is determined, avoiding deviations in risk probability calculations caused by unreasonable parameters.

[0085] By scientifically fitting the marginal distribution and constructing a joint probability model, the calculation of the probability of sudden drought risk is more scientific and targeted. It can accurately quantify the probability of occurrence of different levels of sudden drought, solving the problems of inaccurate risk quantification and lack of scientific basis in existing methods.

[0086] The standardized and reusable agricultural drought risk assessment results include the probability of drought occurrence, recurrence period, and risk level classification results under different durations and intensity thresholds. The assessment results can output spatial scale risk distribution maps, risk level zoning maps, and temporal scale risk change sequences, which can be used for agricultural drought monitoring and early warning, precision irrigation scheduling, and regional agricultural drought risk prevention and control decisions. The drought risk level is classified according to the AFDI index value and drought duration, and includes at least three levels: mild, moderate, and severe.

[0087] The content, risk level classification standards, output format, and application scenarios of the assessment results are clearly defined as follows: Assessment results content: Generate standardized and reusable agricultural drought risk assessment results, specifically including: the probability of drought occurrence and recurrence period under different combinations of duration thresholds D (2, 3, and 4 DIT units, corresponding to 10 days, 15 days, and 20 days) and intensity thresholds S (0.3, 0.5, and 0.7), as well as the risk classification results of mild, moderate, and severe levels, ensuring that the results are comprehensive, highly targeted, and can meet different decision-making needs.

[0088] Risk level classification: The risk level is classified based on the AFDI index value and the duration of the drought. The specific classification criteria are as follows: mild drought (probability of occurrence P≤0.3, corresponding to AFDI≥0.2 and duration≤15 days), moderate drought (0.3<P≤0.6, corresponding to AFDI≥0.3 and duration15-20 days), and severe drought (P>0.6, corresponding to AFDI≥0.5 and duration≥20 days). This ensures that the risk level classification is scientific and reasonable, and has clear quantitative standards to avoid subjective judgment.

[0089] Assessment Results Output and Application: The assessment results include spatial-scale risk distribution maps, risk level zoning maps (intuitively presenting the drought risk levels of different regions of summer maize in the Huaibei Plain), and temporal-scale risk change sequences (presenting the interannual changes in drought risk from 1961 to 2020), making the assessment results more intuitive and easier to understand. The assessment results are directly applied to agricultural drought monitoring and early warning, precision irrigation scheduling (developing irrigation plans for high-risk areas and growth stages), and regional agricultural drought risk prevention and control decisions, providing a scientific basis for water management during the key growth period of summer maize and effectively improving the regional agricultural drought prevention and control capabilities.

[0090] The refinement of the assessment results clarifies the content, output format, and application scenarios of the results, further enhancing their practicality and reusability. It addresses the problem of poor practicality in existing assessment results, with the following specific benefits: It clarifies the specific content of the assessment results, including the probability of occurrence, recurrence period, and risk level classification results under different threshold combinations, making the assessment results more targeted and comprehensive. It solves the problem that existing assessment results are too simplistic and cannot meet different decision-making needs, providing precise support for agricultural production decisions in different scenarios.

[0091] The criteria for classifying the risk level of sudden drought (combined with the duration of the sudden drought) and the number of levels (at least three levels) should be clearly defined to make the risk level classification more scientific and standardized, avoid the problems of vague risk level classification and strong subjectivity, and make it easier for users to intuitively judge the degree of sudden drought risk.

[0092] The assessment results are clearly defined in multiple output formats (spatial scale distribution map, temporal scale change sequence), which intuitively present the spatial distribution and temporal variation characteristics of the risk of sudden drought. This solves the problem that existing assessment results are obscure and difficult to understand and apply intuitively, making it easier for users to quickly grasp the regional risk of sudden drought.

[0093] The specific application scenarios of the assessment results (drought monitoring and early warning, precision irrigation scheduling, and risk prevention and control decision-making) are clearly defined, which realizes the deep integration of assessment results with agricultural production practice, solves the problem that the existing assessment results are not very practical and difficult to serve production decision-making, improves the reusability of assessment results, and can directly provide scientific basis for the prevention and control of sudden drought in regional agriculture, thereby reducing disaster losses.

[0094] This embodiment takes the summer maize planting area in the Huaibei Plain of Anhui Province as the research area and Zhengdan 958 as the target crop. It fully implements the agricultural drought risk assessment method based on the dynamic water demand of crops for 60 consecutive years from 1961 to 2020, effectively solving the core defects of existing agricultural drought assessment methods.

[0095] The core advantages of this embodiment are: taking crop dynamic water demand as the core, integrating multi-source data and standardizing the preprocessing process, accurately simulating crop growth and water consumption characteristics through a locally calibrated crop growth model; constructing an adaptive dynamic time recognition unit to achieve precise coupling between crop growth stages and time scales; constructing an AFDI index that combines soil moisture characteristics to accurately quantify the driving factors of sudden drought; quantifying different levels of sudden drought risk through a scientific joint probability model, outputting standardized and visualized assessment results, and ultimately achieving accurate assessment of summer maize sudden drought risk.

[0096] The implementation process of this embodiment is closely aligned with the actual agricultural production in the region, and the assessment results have high accuracy and practicality. It can directly serve agricultural drought monitoring and early warning, precise irrigation scheduling, and decision-making on sudden drought risk prevention in the summer maize planting area of ​​the Huaibei Plain. At the same time, it provides a technical framework and implementation paradigm that can be referenced for agricultural sudden drought risk assessment in other crops and other regions.

[0097] This embodiment, through the complete implementation of the above six steps, specifically addresses the core problems of existing agricultural drought assessment methods, such as disconnection from crop dynamic water requirements, unreasonable time scales, low accuracy in drought quantification, and poor practicality of assessment results. It achieves accurate assessment of summer maize drought risk, and the assessment results are consistent with the actual agricultural production in the Huaibei Plain, which can directly serve regional agricultural disaster prevention and mitigation and production decision-making.

[0098] like Figure 4 As shown, this embodiment of the invention provides an agricultural drought risk assessment device for dynamic crop water demand, used to implement the above-mentioned agricultural drought risk assessment method. The two are in a corresponding relationship: each module of the device corresponds to each step of the method, working together to complete the drought risk assessment and ensuring that the method can be implemented efficiently and accurately. The method embodiment provides specific technical logic and operation procedures for the device embodiment, and the device embodiment provides hardware / software carriers for the method embodiment, enabling the method to be practically applied to agricultural production practices.

[0099] The device includes a data preprocessing module, a growth simulation module, a time coupling module, an index calculation module, a drought surge identification module, and a risk assessment module. These modules are interconnected and collaboratively complete agricultural drought surge risk assessment. The modules communicate via a bus to collaboratively complete the drought surge risk assessment for summer maize in the Huaibei Plain. The hardware / software configuration of each module is adapted to the implementation requirements of the methodology, ensuring efficient execution of each step. The specific functions of each module correspond to each step of the methodology, enabling the full-process implementation of the methodology.

[0100] The data preprocessing module collects multi-source data related to crop growth in the target area, performs standardization and missing value completion on the multi-source data, and generates a standardized dataset that conforms to the input standards of crop growth models. The growth simulation module stores various crop growth models adapted to different crop agronomic characteristics, completes model parameter calibration and validity verification, simulates the dynamic growth of crops throughout their entire growth period based on the standardized dataset, and outputs key crop growth indicators and soil moisture consumption characteristics. The time coupling module constructs adaptive dynamic time identification units (DITs), using DITs as a discrete time scale to segment the entire crop growth period, achieving precise coupling between the entire crop growth period and DIT units, ensuring that each DIT unit uniquely corresponds to one crop. The system comprises four modules: a growth stage; an index calculation module, which uses DIT units as the core calculation unit to calculate the difference between crop water supply and demand, calls the AFDI index calculation logic to calculate the AFDI index value of each DIT unit, and quantifies the core driving indicators of sudden drought occurrence; a sudden drought identification module, which stores preset AFDI drought thresholds, determines the sudden drought status of each DIT unit based on the thresholds, identifies spatiotemporally continuous sudden drought DIT units to define complete sudden drought events, and extracts the core feature parameters of sudden drought events; and a risk assessment module, which fits the marginal distribution of the core feature parameters of sudden drought, constructs a joint probability distribution model, calculates the probability of occurrence of different levels of sudden drought, generates standardized and reusable agricultural sudden drought risk assessment results, and completes the visualization output and storage of the assessment results.

[0101] Based on the above methodology, the specific implementation of each module is as follows: Structure A: Data Preprocessing Module: Used to collect multi-source data (meteorological data, crop management data, and basic soil parameters) from the summer maize planting area in the Huaibei Plain of Anhui Province, perform quality control, missing value completion, and standardization and normalization processing to generate a standardized dataset that meets the input standards of the crop growth model, corresponding to method step S1, and provide reliable input data for the crop growth simulation module.

[0102] Structure B: Growth Simulation Module: Stores multiple crop growth models such as CERES-Maize, CERES-Wheat, and CERES-Rice. Calls the CERES-Maize model adapted for summer maize, completes model genetic parameter calibration (GLUE method) and validity verification (ARE, RRMSE indicators), simulates the growth dynamics of summer maize throughout its entire growth period based on a standardized dataset, and outputs key indicators such as daily evapotranspiration and soil moisture consumption characteristics, corresponding to method step S2, providing data support for subsequent modules.

[0103] Structure C: Time Coupling Module: Used to construct adaptive DIT units, combining summer maize phenological data and regional precipitation variation cycles to determine the DIT length (5 days), segmenting the entire growth period of summer maize, achieving precise coupling between crop growth stages and DIT units, ensuring that each DIT unit uniquely corresponds to a growth stage, corresponding to method step S3, providing a timescale basis for AFDI index calculation and drought surge identification.

[0104] Structure D: Index Calculation Module: Using the DIT unit as the core calculation unit, it calculates the moisture supply-demand difference for each DIT unit and calls the AFDI index calculation logic. Calculate the AFDI index value of each DIT unit to quantify the core driving indicator of sudden drought occurrence, corresponding to method step S4, and provide quantitative basis for the sudden drought event identification module.

[0105] Structure E: Drought Identification Module: Stores a preset AFDI drought threshold (AFDI≥0.2), determines the drought status of each DIT unit based on the threshold, identifies spatiotemporally continuous drought DIT units, defines complete drought events, and extracts core feature parameters such as drought duration and intensity, corresponding to method step S5, and provides feature data for the risk assessment and visualization module.

[0106] Structure F: Risk Assessment Module: Fits the marginal distribution of drought duration and intensity (selects the optimal distribution using the BIC criterion), constructs the GumbelCopula joint probability distribution model, calculates the probability of occurrence of different levels of drought, classifies risk levels, generates standardized assessment results, and completes the visualization output and storage at spatial and temporal scales, corresponding to method step S6, providing support for agricultural production decision-making.

[0107] This device embodiment completely corresponds to the method embodiment, with each module working collaboratively. Its beneficial effects, combined with the device structure and method flow, are explained in detail below: Data preprocessing module: By accurately collecting multi-source data and performing standardized preprocessing, the module ensures the standardization and reliability of input data, solves the problem of inconsistent data quality, provides basic data support for the stable operation of the entire device, and improves the evaluation accuracy of the device.

[0108] The growth simulation module stores multiple growth models adapted to different crops, completes localized parameter calibration and validity verification, ensures the accuracy of crop growth dynamic simulation, solves the problems of poor model adaptability and large simulation deviation in existing devices, provides accurate data for subsequent drought identification and assessment, and improves the versatility and reliability of the device.

[0109] Time-coupling module: By constructing an adaptive DIT unit and achieving precise coupling with the crop growth stage, the problem of fixed time scale and disconnection from crop growth in traditional devices is solved, ensuring the accuracy of drought detection and improving the adaptability and flexibility of the device.

[0110] Index Calculation Module: By accurately calculating the difference between water supply and demand and calculating the AFDI index, the core driving indicators of sudden drought are quantified, which solves the problem of low accuracy in the quantification of sudden drought in existing devices, provides a clear basis for the identification of sudden drought, and improves the device's ability to identify sudden drought.

[0111] The sudden drought identification module accurately defines sudden drought events by setting preset thresholds, identifying continuous sudden drought units, and extracting core features. This solves the problems of incomplete identification and incomplete feature extraction in existing devices, providing accurate feature data for risk assessment and improving the basic risk assessment capabilities of the device.

[0112] Risk assessment module: By constructing a joint probability model, generating standardized assessment results and visualizing the output, it solves the problem that the assessment results of existing devices are not very practical and difficult to apply to production decisions, so that the assessment results can directly serve agricultural production and improve the practicality and reusability of the device.

[0113] The modules work together: efficient data transmission and process connection are achieved through bus communication, ensuring that each step of the method embodiment can be accurately implemented through the corresponding module, forming a complete process of "data acquisition-simulation-identification-evaluation-output". This solves the problem of independent modules and poor coordination in existing devices, improves the overall operating efficiency and evaluation accuracy of the device, and can provide reliable technical support for the prevention and control of sudden drought in regional agriculture.

[0114] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for assessing the risk of sudden drought in agriculture based on the dynamic water demand of crops, characterized in that, include: Acquire multi-source monitoring data of the target area, preprocess the multi-source monitoring data, and obtain basic data suitable for crop water requirement simulation; Based on the aforementioned basic data, a crop water requirement simulation model is run to simulate the crop's growth process and soil moisture changes throughout its entire growth period, thereby obtaining the crop's water consumption parameters and soil moisture parameters. A dynamic time identification unit adapted to the crop growth stage and regional precipitation pattern is constructed, and the entire crop growth period is segmented according to the dynamic time identification unit to achieve coupling and matching between the dynamic time identification unit and the crop growth stage. Using the dynamic time recognition unit as the calculation scale, and combining the water consumption parameter and the soil moisture parameter, an agricultural drought index is constructed. The agricultural drought index is then used to quantify the degree of crop water deficit in each unit. Identify sudden drought units according to preset judgment criteria, define complete sudden drought events based on the sudden drought units, and extract the core feature parameters of the sudden drought events; A risk calculation model is constructed based on the core feature parameters, and the risk of sudden drought in agriculture is assessed using the risk calculation model and the assessment results are output.

2. The method according to claim 1, characterized in that, The multi-source monitoring data includes meteorological data, basic soil data, and crop management data; The preprocessing includes filling in missing values ​​and removing outliers from the multi-source monitoring data.

3. The method according to claim 1, characterized in that, The crop water requirement simulation model underwent regional localization calibration and effectiveness verification. The regional localization calibration determined the parameter range by combining the genetic characteristics of the target crop with the regional agricultural production conditions. The effectiveness verification used the simulation errors of phenological period, biomass, and soil moisture content as verification indicators.

4. The method according to claim 1, characterized in that, The duration of the dynamic time recognition unit is determined by the fluctuation range of the crop growth stage and the regional precipitation change cycle, and the duration of the dynamic time recognition unit does not exceed 1 / 3 of the duration of the corresponding crop growth stage.

5. The method according to claim 4, characterized in that, The dynamic time recognition unit assigns each reproductive stage to the dominant reproductive stage with the highest percentage of days within the unit, based on the percentage of days each reproductive stage has in the unit.

6. The method according to claim 1, characterized in that, The agricultural drought index is constructed based on the cumulative evapotranspiration, cumulative precipitation, and average soil moisture content within the dynamic time recognition unit. When the precipitation meets the crop's water requirements, the agricultural drought index is 0, and the higher the degree of water deficit, the larger the agricultural drought index value.

7. The method according to claim 1, characterized in that, The core feature parameters include the duration and intensity of the drought; the risk calculation model is constructed based on the marginal distribution and Copula joint distribution of the core feature parameters, and is used to calculate the probability of drought occurrence and the recurrence period.

8. The method according to claim 7, characterized in that, The marginal distribution is selected as the optimal distribution type using the Bayesian information criterion, and the parameters of the Copula joint distribution are determined and their rationality is verified based on the correlation coefficient.

9. An agricultural drought risk assessment device oriented towards dynamic crop water demand, characterized in that, include: The data preprocessing module is used to acquire and preprocess multi-source monitoring data to obtain basic data, which is then sent to the growth simulation module and the time coupling module respectively. The growth simulation module receives the basic data output by the data preprocessing module, runs the crop water requirement simulation model, outputs water consumption parameters and soil moisture parameters, and sends them to the index calculation module. The time coupling module is used to receive the basic data output by the data preprocessing module, extract crop growth period information, construct dynamic time recognition units and couple them with crop growth stages, and output time matching parameters to the index calculation module. The index calculation module is used to receive water consumption parameters and soil moisture parameters output by the growth simulation module and time matching parameters output by the time coupling module, construct and calculate the agricultural drought index to quantify the degree of crop water deficit, and send the quantified data to the drought identification module. The sudden drought identification module is used to receive the quantitative data output by the index calculation module, identify sudden drought units, define sudden drought events, and extract core feature parameters, and then send the core feature parameters to the risk assessment module. The risk assessment module receives the core feature parameters output by the drought identification module, constructs a risk calculation model, and outputs the agricultural drought risk assessment results.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program that, when executed by a processor, implements the method described in any one of claims 1 to 8.

Images