A three-dimensional visualization decision method and system for silage corn water-nitrogen management scenarios

Abstract

This invention discloses a three-dimensional visualization decision-making method and system for water and nitrogen management scenarios in silage maize. The method collects input data from field experiments and the RZWQM2 model, completes model calibration, validation, and batch simulation of irrigation-nitrogen application scenarios, extracts dry matter content, soil moisture content at maturity (0–30 cm), and phenotypic indices, constructs a three-dimensional driving table, and converts it into a GroIMP-importable .sci parameter sequence, generating three-dimensional plant architecture / canopy images at key stages. Based on yield formation capacity, soil moisture status, and input levels, the method divides the scenario into zones, outputting recommended water and nitrogen management intervals and a three-dimensional evidence chain.

Description

Technical Field

[0001] This invention belongs to the field of smart agriculture and agricultural information technology, and relates to a three-dimensional visualization decision-making method and system for water and nitrogen management scenarios of silage corn. Specifically, it relates to a three-dimensional visualization decision-making method and system for water and nitrogen management scenarios of silage corn based on the coupling of crop model and function-structure model. Background Technology

[0002] Silage maize, as a typical high-water- and high-nitrogen-demand forage crop, exhibits highly sensitive yield and biomass formation to irrigation and nitrogen management. In water-scarce regions such as the arid Northwest, inappropriate irrigation and nitrogen application not only lead to resource waste but also pose environmental risks. Existing technologies, such as the RZWQM2 mechanistic model, can simulate soil moisture, nitrogen transport, and crop growth processes, providing quantitative tools for water and nitrogen management. However, the following shortcomings remain in practical applications:

[0003] 1. Low standardization of scenario simulation: Existing methods lack standardized model calibration, batch scenario generation and output parsing processes, resulting in low efficiency in scenario library construction, inconsistent data standards, and difficulty in supporting reliable scenario comparison and selection.

[0004] 2. Poor 3D visualization interactivity: Traditional scenario analysis results are mostly presented in two-dimensional tables or graphs, which cannot intuitively show the differences in three-dimensional morphology such as crop plant type and canopy structure under different management measures, thus reducing the interpretability and acceptability of decision-making results.

[0005] 3. Data gap between models: There are significant differences in data format, field definition and unit between the output of mechanistic models (such as RZWQM2) and the input of functional-structural models (such as GroIMP projects). The lack of automated and standardized conversion bridges limits the batch generation of 3D evidence chains.

[0006] Therefore, there is an urgent need for a decision-making method that can couple mechanism simulation, phenotypic fusion and three-dimensional visualization in a closed loop to improve the scientific rigor and intuitiveness of water and nitrogen management scenario screening. Summary of the Invention

[0007] This invention aims to overcome the shortcomings of existing technologies and provide a three-dimensional visualization decision-making method and system for water and nitrogen management scenarios in silage maize. This system enables standardized scenario batch simulation, automated generation of three-dimensional visualization evidence chains, and output of feasible recommended management intervals based on yield formation capacity and soil moisture status at maturity.

[0008] Specifically, this is achieved through the following technical solutions:

[0009] First, this invention provides a three-dimensional visualization decision-making method for water and nitrogen management scenarios in silage maize, comprising the following steps:

[0010] Step 1: Collect and organize field test data of silage maize and the input data required for model operation. The field test data shall include at least the number of leaves, plant height, stem diameter, leaf length and width, and leaf area index (LAI). The input data shall include at least meteorological data, soil profile parameters, and irrigation and fertilization management records.

[0011] Step 2: Construct the RZWQM2 model based on the input data described in Step 1, and use a step-by-step calibration strategy to obtain the validation parameter set: First, calibrate the soil moisture process parameters so that the simulated and measured values ​​of soil moisture content in the 0-30 cm layer at maturity meet the preset threshold; after the moisture process parameters stabilize, adjust the crop growth parameters so that the simulated and measured values ​​of aboveground dry matter meet the preset threshold; and validate the parameters on independent annual experimental data to obtain the parameter set P that can be used for scenario expansion.

[0012] Step 3: Under the condition that the parameter set P remains unchanged, construct and simulate 25 complete irrigation-nitrogen application scenarios. The application dates and frequency of irrigation and nitrogen application are consistent with the baseline scenario. Only the amount of irrigation and nitrogen application for each application are scaled proportionally according to the irrigation scaling factor and the nitrogen application scaling factor, thereby forming the scenario simulation result database D.

[0013] Step 4: Extract key indicators with a unified caliber from the output file of the scenario simulation results database D. The key indicators include at least: Days after sowing (DAS), Leaf area index (LAI), Top DM (top dry matter), Plant height or developmental progress indicators, and Soil moisture content (SWC_0-30) at maturity (0-30 cm); and form a structured scenario long table dataset.

[0014] Step 5: Align and merge the scenario long table dataset described in Step 4 with the measured phenotypic data described in Step 1 according to the number of days after broadcast to construct a three-dimensional driving table T;

[0015] Step 6: Based on the template parameter file params_template.sci exported from the GroIMP 3D modeling project, according to the field mapping relationship and stable replacement strategy, automatically convert the 3D driving table T into an importable .sci parameter sequence file set F, and batch import the .sci parameter sequences into the GroIMP project to generate key period 3D plant type / canopy image sequence V;

[0016] Step 7: Construct scenario evaluation and partitioning rules based on the scenario long table dataset, and output the priority recommendation interval R and its corresponding 3D visualized evidence chain image sequence V. The evaluation indicators include the aboveground dry matter mass (TopDM) at maturity, the soil moisture content (SWC_0-30) at 0-30 cm at maturity, the cumulative irrigation amount, and the cumulative nitrogen application amount. Among them, TopDM is used to characterize the aboveground biomass formation capacity of silage maize, SWC_0-30 is used to characterize the soil moisture retention status at maturity, and the cumulative irrigation amount and cumulative nitrogen application amount are used to judge the rationality of resource input. Scenarios with TopDM reaching the baseline scenario level or the preset ratio requirement and SWC_0-30 not lower than the preset lower limit are classified into the priority recommendation area. Scenarios that partially meet the above conditions are classified into the candidate area. Scenarios with low TopDM, insufficient SWC_0-30, or whose TopDM improvement is not significant despite increased input are classified into the non-recommended area. The priority recommendation interval R is determined by the irrigation scaling factor range and nitrogen scaling factor range corresponding to the scenarios within the priority recommendation area.

[0017] Preferably, the preset threshold in step 2 is evaluated using root mean square error (RMSE) and normalized root mean square error (nRMSE), where RMSE is calculated based on the root mean square error of the observed and simulated value sequences, and nRMSE is the ratio of RMSE to the observed mean. Further, the nRMSE of TopDM is no greater than 15%, and the nRMSE of SWC_0-30 is no greater than 10%.

[0018] Preferably, the 25 scenarios in step 3 are formed by a complete combination of 5 irrigation scaling levels and 5 nitrogen application scaling levels, wherein the set of irrigation scaling coefficients is {0.8, 0.9, 1.0, 1.1, 1.2}, and the set of nitrogen application scaling coefficients is {0.8, 0.9, 1.0, 1.1, 1.2}. In this embodiment, the field trial treatment W1N2 is used as the baseline scenario, where W1 corresponds to an irrigation amount of 553.5 mm and N2 corresponds to a nitrogen application amount of 280 kg N·hm². -2 When the irrigation scaling factor is 1.0 and the nitrogen application scaling factor is 1.0, the absolute irrigation amount and nitrogen application amount in extended scenario S33 are the same as those in W1N2; for the other scenarios, the irrigation amount and nitrogen application amount for each irrigation and nitrogen application are scaled proportionally based on W1N2, while the date, frequency, and allocation ratio of irrigation and nitrogen application remain unchanged.

[0019] Preferably, the source files for the key indicators in step 4 include at least PlantGro.OUT, SoilWat.OUT, and ACCWAT.OUT. PlantGro.OUT is used to extract indicators such as DAS, LAID, CWAD, and CHTD; SoilWat.OUT is used to extract stratified soil moisture content indicators such as DAS, SW1D, SW2D, and SW3D; ACCWAT.OUT is used to extract water budget indicators such as IRR, RAIN, and AET; SWC_0-30 is obtained by weighting the volumetric water content of the soil layer corresponding to 0-30 cm in SoilWat.OUT according to the soil layer thickness.

[0020] Preferably, the three-dimensional driving table T in step 5 uses the daily output DAS of the RZWQM2 model as a unified time index, with a DAS step size of 1 day, covering all simulated DAS from sowing to maturity; the three-dimensional driving table T includes at least the fields step, DAS, leaf_no_obs, height_m, stem_diam, leaf_len, leaf_wid, LAI, and density_plants_m2; the key period image sequences used for three-dimensional display are extracted and generated from the complete three-dimensional driving table T according to the preset DAS.

[0021] Preferably, the stable replacement strategy in step 6 includes replacing key external driving fields and freezing internal structural parameters of the template. The key external driving fields include at least the time control field Chr_Age(N), the density-related field Projected_area(Sp), and the structural upper limit field Macro_Num(Nu_Ma). The internal structural parameters of the template include organ development order, leaf-to-segment connection relationships, basic structural templates, and related matrix parameters, which remain unchanged during batch generation of .sci parameter sequences to ensure the stability of batch import and comparability between scenarios.

[0022] Preferably, the density-related field Projected_area(Sp) is calculated from the planting density using the formula Projected_area(Sp) = 10000 / density, where density is the planting density in units of plants per meter. -2 Projected_area(Sp) is in cm²·plants -1 In this embodiment, the planting density is 113,100 plants per hectare. -2 Therefore, Projected_area(Sp) = 10000 / 11.31 ≈ 884.2 cm²·plant -1 .

[0023] Second, the present invention provides an intelligent decision-making system for implementing the method described herein, comprising the following modules:

[0024] The system includes modules for data acquisition and management, the RZWQM2 simulation engine, automated batch processing control, scenario library generation, indicator extraction and standardization, phenotypic fusion and 3D-driven table construction, .sci parameter sequence generation, 3D visualization, and scenario evaluation and recommendation.

[0025] The data acquisition and management module is used to collect, store, and manage various input data required for system operation, including meteorological data, soil profile parameters, irrigation and fertilization management records, and phenotypic data such as leaf number, plant height, stem diameter, leaf length and width, and leaf area index (LAI) observed in silage maize field trials. This module provides a unified data access interface for other modules to use.

[0026] The RZWQM2 simulation engine module encapsulates and calls the RZWQM2 model program, and is responsible for model engineering construction, parameter setting, runtime scheduling, and result output management. This module receives simulation instructions and parameter configurations from the automated batch processing control module, executes single or batch scenario simulations, and outputs raw result files.

[0027] The automated batch processing control module is used to automate the step-by-step calibration and batch scenario simulation of the RZWQM2 model. Specifically, it includes: automatically generating and modifying model input files (such as soil, meteorological, and management files), batch submitting simulation tasks, monitoring the running status, and automatically parsing the output files by calling the index extraction module. This module serves as an automated bridge connecting the simulation engine and the scenario library generation module.

[0028] The scenario library generation module is used to automatically generate complete input configurations for 25 irrigation-nitrogen application fully combined scenarios, based on the irrigation scaling factor set and the nitrogen application scaling factor set, while keeping the parameter set P unchanged. The simulation output results of each scenario are organized and stored according to a unified naming convention to form a structured scenario simulation result database D.

[0029] The indicator extraction and standardization module is used to parse and extract key indicators with standardized standards from the original result files (such as PlantGro.OUT, SoilWat.OUT, ACCWAT.OUT, etc.) output by the RZWQM2 model. These key indicators include at least the number of days after sowing (DAS), leaf area index (LAI), aboveground dry matter (TopDM), plant height or developmental progress indicators, and soil moisture content (SWC_0-30) at maturity (0-30 cm). This module automatically traverses and parses the outputs of each scenario to form a structured scenario long table dataset.

[0030] The phenotypic fusion and 3D-driven table construction module is used to align and fuse the scenario long table dataset output by the index extraction and caliber unification module with the measured phenotypic data (number of leaves, plant height, stem diameter, leaf length and width, LAI, etc.) in the data acquisition and management module according to the number of days after sowing, and generate a 3D-driven table T containing time steps, morphological parameters and density parameters, providing standardized input for the subsequent generation of 3D parameter sequences.

[0031] The .sci parameter sequence generation module is used to automatically convert the 3D driving table T into a set of .sci parameter sequence files F that can be batch imported into GroIMP projects, based on the template parameter file params_template.sci exported from the GroIMP 3D modeling project and according to predefined field mapping relationships and a stable replacement strategy. The stable replacement strategy includes replacing the time control field Chr_Age (N), the density-related field Projected_area (Sp), and the structural upper limit field Macro_Num (Nu_Ma), while freezing the parameters of the internal structural occurrence matrix block to ensure the stability of batch generation and comparability between scenarios.

[0032] The 3D visualization module is used to automatically import the parameter sequence file set F output by the .sci parameter sequence generation module into the GroIMP project in batches. It performs visualization rendering under a unified 3D scene and perspective, and outputs 3D plant type / canopy image sequence V of key growth stages in various scenarios (such as jointing stage, large trumpet stage, tasseling stage, grain filling stage, etc.). It also supports image export and management.

[0033] The scenario evaluation and recommendation module is used to construct scenario evaluation and partitioning rules based on a scenario long table dataset. These rules use the TopDM (Top Dry Matter) to characterize yield formation capacity, the SWC_0-30 (0-30 cm soil moisture content at maturity) as a water state constraint, and combine cumulative irrigation and nitrogen application to determine the rationality of inputs. The higher level of the TopDM is determined based on a preset proportion of the baseline scenario's TopDM or the maximum value of the TopDM among the 25 scenarios. The preset lower limit of SWC_0-30 is determined based on the baseline scenario's maturity soil moisture content, the suitable lower limit of soil moisture content in the study area, or field trial observations. This module compares, filters, and partitions the 25 scenarios, automatically outputting a priority recommendation interval R, and in conjunction with a 3D visualization module to output a 3D evidence chain image sequence V representing the scenarios within the recommendation interval, forming the final decision support report.

[0034] Third, the present invention provides a three-dimensional visualization decision-making system for water and nitrogen management scenarios of silage corn, including one or more processors; a memory for storing one or more programs; the processors are configured to execute program instructions stored in the memory, and the program instructions execute the three-dimensional visualization decision-making method for water and nitrogen management scenarios of silage corn described in the present invention when they are executed.

[0035] Fourth, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by one or more processors, implements the three-dimensional visualization decision-making method for water and nitrogen management scenarios of silage corn as described in the present invention.

[0036] Compared with the prior art, the present invention has at least the following advantages:

[0037] High degree of standardization and automation: The calibration, batch scenario simulation and output analysis of the RZWQM2 model were realized through automated scripts, and a structured scenario database was built, which greatly improved the efficiency and consistency of scenario analysis.

[0038] A three-dimensional evidence chain enhances interpretability: By fusing the output of the RZWQM2 mechanistic model with measured phenotypic data, a GroIMP parameter sequence is automatically generated through a stable replacement strategy, enabling a three-dimensional visualization of plant architecture / canopy morphology under key water and nitrogen scenarios. These three-dimensional images serve as intuitive evidence supporting the recommended intervals, aiding in the explanation of morphological differences between scenarios, but do not replace the RZWQM2 simulation of crop growth and soil moisture processes.

[0039] Stable and reliable parameter conversion: The strategy of "freezing the internal structural parameters of the template and replacing only the key external driving fields" is proposed to avoid the problem of disorder in the 3D model structure caused by batch parameter import, and to ensure the comparability of 3D images between scenes and the stability of generation.

[0040] Closed-loop decision support: It forms a complete closed loop from "mechanism simulation → phenotypic fusion → 3D visualization → recommended interval output". The output management suggestions have both numerical basis (ranking and partitioning) and intuitive evidence (3D images), which are more conducive to guiding agricultural production practices. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the overall process of the three-dimensional visualization decision-making method for water and nitrogen management scenarios of silage corn according to the present invention;

[0042] Figure 2 This is a schematic diagram illustrating the field mapping relationship between the three-dimensional driving table and the stable replacement strategy of the .sci parameter sequence in this invention;

[0043] Figure 3This is a schematic diagram of scenario evaluation, recommendation interval identification, and three-dimensional evidence chain output in this invention. Detailed Implementation

[0044] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment takes the water and nitrogen management of silage maize in the oasis irrigation area of ​​the Hexi Corridor as an example, but the application scope of the present invention is not limited to this.

[0045] Example 1

[0046] 1. Experimental conditions and data acquisition

[0047] The experiment was conducted at Huarui Ranch in Minle County, Zhangye City, Gansu Province, from 2022 to 2023. The tested variety was "Jinling Silage 67". A randomized block design with complete combinations of water and nitrogen was used, and the planting density was 113,100 plants per hectare. -2 The collected data is divided into two categories:

[0048] Input data for model operation includes daily meteorological data, soil profile stratification parameters (bulk density, field water holding capacity, etc.), and records of the date and frequency of irrigation and fertilization.

[0049] Field trial data: Leaf number, plant height, stem diameter, leaf length and width, and leaf area index (LAI) were observed regularly according to the growth stage. A summary of relevant experimental design and management parameters is shown in Table 1. The nitrogen application rates listed in Table 1 are all pure nitrogen amounts expressed as elemental nitrogen. The actual nitrogen fertilizer was urea, applied via fertigation with irrigation water.

[0050]

[0051] 2. RZWQM2 Calibration, Validation, and Scenario Extension

[0052] Step-by-step calibration and validation: Taking 2022 as the calibration year, soil moisture parameters were first calibrated to ensure that the error between the simulated and measured values ​​of soil moisture content in the 0-30 cm depth met a preset threshold. Based on this, crop parameters were calibrated to ensure that the simulation accuracy of aboveground dry matter (TopDM) met a preset threshold. 2023 was the validation year, and the parameter set was kept unchanged for evaluation. After meeting the thresholds, a parameter set P that can be used for scenario expansion was obtained. The calibration and validation accuracy are shown in Table 2.

[0053] Batch scenario simulation: Under parameter set P, using the field trial treatment W1N2 as the baseline scenario, 25 complete combination scenarios were constructed, consisting of 5 irrigation scaling levels (0.8, 0.9, 1.0, 1.1, 1.2) × 5 nitrogen application scaling levels (0.8, 0.9, 1.0, 1.1, 1.2). Among these, an irrigation scaling factor of 1.0 corresponds to an irrigation amount of 553.5 mm for W1, and a nitrogen application scaling factor of 1.0 corresponds to a nitrogen application amount of 280 kg N·hm² for N2.-2 The scenario with an irrigation scaling factor of 1.0 and a nitrogen scaling factor of 1.0 (denoted as S33) has the same absolute input as W1N2. The irrigation and fertilization dates, frequency, and allocation ratios for each scenario are consistent with W1N2, with only the application amount scaled proportionally for each application. After batch running RZWQM2, the scenario simulation result database D is obtained.

[0054]

[0055] 3. Indicator Extraction and 3D Driving Table Construction (Steps 4-5)

[0056] Key indicators are uniformly extracted from the output file of database D. The sources and column names of each indicator are shown in Table 3. Among them, SWC_0-30 is obtained by weighting the volumetric water content of the 0–30 cm soil layer in SoilWat.OUT according to the soil layer thickness. The extracted scenario long table data and the measured phenotypic data are aligned and merged according to the number of days after sowing (DAS) to construct a three-dimensional driving table T. In this embodiment, the three-dimensional driving table T uses the DAS output daily by the RZWQM2 model as a unified time index, with a DAS step size of 1 day, covering all simulated DAS from sowing to maturity. Records with the same DAS as the measured phenotypic data are directly written to the measured values. DAS located between adjacent measured periods and lacking measured phenotypic values ​​are filled in using piecewise linear interpolation. The three-dimensional images of the key growth periods used for display can be generated by extracting from the complete daily three-dimensional driving table T according to the preset DAS.

[0057]

[0058] 4. .sci parameter sequence generation and 3D visualization

[0059] Using the params_template.sci file exported from the GroIMP 3D modeling project as a template, a stable replacement strategy of "replacing key external driving fields and freezing internal structural parameters of the template" is adopted (see...). Figure 2 (and Table 4), only the key fields in the 3D driving table T are mapped and replaced with the corresponding fields in the template: the time control field Chr_Age(N) is determined by DAS, and the density field Projected_area(Sp) is converted from the planting density. In this embodiment, the planting density is 113,100 plants per hectare. -2 , Projected_area(Sp) is taken as 884.2 cm²·plant -1The upper limit field Macro_Num (Nu_Ma) is determined by the measured maximum number of leaves. The organogenesis sequence, connectivity, basic structural template, and related matrix parameters used to maintain the basic topological structure of the silage maize 3D plant architecture remain unchanged in the template. Based on this, 25 scenario-specific .sci parameter sequence files are automatically generated. After batch importing into the GroIMP project, comparable 3D plant architecture / canopy image sequences V for key periods are output. RZWQM2 is used to simulate crop growth and soil moisture processes under different water and nitrogen scenarios. The GroIMP 3D modeling project is mainly used to convert the RZWQM2 output results and measured phenotypic data into visualized 3D plant architecture / canopy evidence. The 3D images are used to assist in interpreting morphological differences and recommended intervals under different scenarios, and do not replace the water and nitrogen process simulation of RZWQM2.

[0060]

[0061] 5. Scenario Evaluation and Recommendation Interval Output

[0062] Based on simulation results of 25 irrigation-nitrogen application scenarios, TopDM, SWC_0-30, cumulative irrigation amount, and cumulative nitrogen application amount at maturity were extracted for each scenario to construct a scenario evaluation data table. During the evaluation process, TopDM was used to characterize the aboveground biomass formation capacity of silage maize, and SWC_0-30 was used to characterize the soil moisture retention status in the 0-30 cm soil layer at maturity. The rationality of resource input was judged in conjunction with changes in irrigation amount and nitrogen application amount. The higher level of TopDM was determined based on the TopDM of the baseline scenario W1N2 or a preset proportion of the maximum TopDM value among the 25 scenarios. In this embodiment, a two-year average relative dry matter mass of not less than 0.99 and a two-year minimum relative dry matter mass of not less than 0.98 were used as candidate conditions for high dry matter mass values. The preset lower limit of SWC_0-30 was determined based on the SWC_0-30 at maturity in the baseline scenario, the lower limit of suitable soil moisture content in the study area, or field trial observation results. During partitioning, scenarios with a high TopDM level and SWC_0-30 not lower than a preset lower limit are assigned to the priority recommendation zone. Scenarios partially meeting the above conditions are assigned to the candidate zone. Scenarios with low TopDM, insufficient SWC_0-30, or where increased input does not significantly improve TopDM are assigned to the non-recommended zone. Based on this, the system outputs a priority recommendation interval R consisting of the irrigation scaling factor range and the nitrogen application scaling factor range. Representative scenarios are selected from the priority recommendation zone to generate a three-dimensional plant type / canopy image sequence V for key growth stages, forming a continuous output of "scenario simulation results - recommendation interval - three-dimensional evidence chain". Using this process, simulation, analysis, three-dimensional parameter conversion, and visualization output of 25 water and nitrogen scenarios can be completed in batches under the same parameter set, the same management date, and a unified output caliber. Compared with methods that only use two-dimensional tables or single curves for scenario comparison, this invention can simultaneously provide numerical evaluation basis and three-dimensional morphological evidence, reducing the workload of manually compiling model output and manually adjusting three-dimensional parameters, and improving the consistency, traceability, and intuitiveness of scenario selection.

Claims

1. A three-dimensional visualization decision-making method for water and nitrogen management scenarios in silage maize, characterized in that, Includes the following steps: Step 1: Collect and organize field test data of silage maize and the input data required for model operation. The field test data shall include at least the number of leaves, plant height, stem diameter, leaf length and width, and leaf area index (LAI). The input data shall include at least meteorological data, soil profile parameters, and irrigation and fertilization management records. Step 2: Construct the RZWQM2 model based on the input data described in Step 1, and use a step-by-step calibration strategy to obtain the validation parameter set: First, calibrate the soil moisture process parameters so that the simulated and measured values ​​of soil moisture content in the 0-30 cm layer at maturity meet the preset threshold; after the moisture process parameters stabilize, adjust the crop growth parameters so that the simulated and measured values ​​of aboveground dry matter meet the preset threshold; and validate the parameters on independent annual experimental data to obtain the parameter set P that can be used for scenario expansion. Step 3: Under the condition that the parameter set P remains unchanged, construct and simulate 25 complete irrigation-nitrogen application scenarios. The application dates and frequency of irrigation and nitrogen application are consistent with the baseline scenario. Only the amount of irrigation and nitrogen application for each application are scaled proportionally according to the irrigation scaling factor and the nitrogen application scaling factor, thereby forming the scenario simulation result database D. Step 4: Extract key indicators with a unified caliber from the output file of the scenario simulation results database D. The key indicators include at least: Days after sowing (DAS), Leaf area index (LAI), Top DM (top dry matter), Plant height or developmental progress indicators, and Soil moisture content (SWC_0-30) at maturity (0-30 cm); and form a structured scenario long table dataset. Step 5: Align and merge the scenario long table dataset described in Step 4 with the measured phenotypic data described in Step 1 according to the number of days after broadcast to construct a three-dimensional driving table T; Step 6: Based on the template parameter file params_template.sci exported from the GroIMP 3D modeling project, according to the field mapping relationship and stable replacement strategy, automatically convert the 3D driving table T into an importable .sci parameter sequence file set F, and batch import the .sci parameter sequences into the GroIMP project to generate key period 3D plant type / canopy image sequence V; Step 7: Construct scenario evaluation and partitioning rules based on the scenario long table dataset, and output the priority recommendation interval R and its corresponding three-dimensional visualization evidence chain image sequence V; the partitioning rules use the aboveground dry matter mass (TopDM) at maturity to represent yield formation capacity, and the soil moisture content (SWC_0-30) at maturity (0-30 cm) to represent soil moisture retention status, and combine the cumulative irrigation amount and cumulative nitrogen application amount to judge the rationality of input; scenarios with TopDM reaching the benchmark scenario level or the preset ratio requirement, and SWC_0-30 not lower than the preset lower limit are classified into the priority recommendation area, scenarios that partially meet the above conditions are classified into the candidate area, and scenarios with low TopDM, insufficient SWC_0-30, or increased input but no significant improvement in TopDM are classified into the non-recommended area; the priority recommendation interval R is determined by the irrigation scaling factor range and nitrogen scaling factor range corresponding to the scenarios in the priority recommendation area.

2. The method according to claim 1, characterized in that, The 25 scenarios described in step 3 are formed by a complete combination of 5 irrigation scaling levels and 5 nitrogen application scaling levels, where the set of irrigation scaling coefficients is {0.8, 0.9, 1.0, 1.1, 1.2} and the set of nitrogen application scaling coefficients is {0.8, 0.9, 1.0, 1.1, 1.2}. The field trial treatment W1N2 is used as the baseline scenario, where W1 corresponds to an irrigation amount of 553.5 mm and N2 corresponds to a nitrogen application rate of 280 kg N·hm². -2 When the irrigation scaling factor is 1.0 and the nitrogen application scaling factor is 1.0, the absolute irrigation amount and nitrogen application amount are the same as in W1N2. In other scenarios, the irrigation amount and nitrogen application amount for each irrigation and nitrogen application are scaled proportionally based on W1N2.

3. The method according to claim 1, characterized in that, The source files for the key indicators in step 4 should include at least: PlantGro.OUT, whose column names should include at least DAS, LAID, CWAD, and CHTD; SoilWat.OUT, whose column names should include at least DAS, SW1D, SW2D, and SW3D; and ACCWAT.OUT, whose column names should include at least IRR, RAIN, and AET; where SWC_0-30 is obtained by weighting the volumetric water content of the soil layer corresponding to 0-30 cm in SoilWat.OUT according to the soil layer thickness.

4. The method according to claim 1, characterized in that, The preset threshold mentioned in step 2 is evaluated using root mean square error (RMSE) and normalized root mean square error (nRMSE). RMSE is calculated based on the root mean square error of the observed and simulated value sequences, and nRMSE is the ratio of RMSE to the observed mean. The nRMSE of TopDM is no greater than 15%, and the nRMSE of SWC_0-30 is no greater than 10%.

5. The method according to claim 1, characterized in that, In step 5, the three-dimensional driving table T uses the daily output DAS of the RZWQM2 model as a unified time index, with a DAS step size of 1 day, covering all simulated DAS from sowing to maturity. The three-dimensional driving table T contains at least the following fields: step, DAS, leaf_no_obs, height_m, stem_diam, leaf_len, leaf_wid, LAI, and density_plants_m2. Records that match the measured phenotypic data DAS are directly merged. For DAS lacking measured phenotypic values ​​between adjacent measured periods, piecewise linear interpolation is used to fill in the gaps. The key period image sequences used for three-dimensional display are extracted from the complete three-dimensional driving table T according to the preset DAS. The number of leaves is corrected to an integer value as a discrete variable, while plant height, stem diameter, leaf length, leaf width, and LAI are treated as continuous variables.

6. The method according to claim 1, characterized in that, The stable replacement strategy described in step 6 includes replacing key external driving fields and freezing template internal structural parameters. The key external driving fields include at least the time control field Chr_Age(N), the density-related field Projected_area(Sp), and the structural upper limit field Macro_Num(Nu_Ma). The template internal structural parameters include the organogenesis order, leaf-internode connection relationship, basic structural template, and related matrix parameters used to maintain the basic topological structure of the silage maize plant. These parameters remain unchanged during the batch generation of .sci parameter sequences to ensure the stability of the batch generation of 3D images and the comparability between scenarios.

7. The method according to claim 6, characterized in that, The density-related field Projected_area(Sp) is calculated from the planting density using the formula Projected_area(Sp) = 10000 / density, where density is measured in plants per meter (m²). -2 Projected_area(Sp) is in cm²·plants -1 When the planting density is 113,100 plants per hectare -2 At that time, Projected_area(Sp) was taken as 884.2 cm²·plant. -1 .

8. A three-dimensional visualization decision-making system for implementing the method of any one of claims 1-7, characterized in that, include: The system includes modules for data acquisition and management, RZWQM2 simulation engine, automated batch processing control, scenario library generation, indicator extraction and standardization, phenotypic fusion and 3D-driven table construction, .sci parameter sequence generation, 3D visualization, and scenario evaluation and recommendation.

9. A three-dimensional visualization decision-making system for water and nitrogen management scenarios in silage maize, characterized in that, include: One or more processors; Memory, used to store one or more programs; The processor is configured to execute program instructions stored in the memory, which, when executed, perform the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program, when executed by one or more processors, implements the method of any one of claims 1 to 7.

Images