Search

CN-121981602-A - Method and device for determining soil salinity stable yield threshold value of cotton root zone

CN121981602ACN 121981602 ACN121981602 ACN 121981602ACN-121981602-A

Abstract

The invention provides a method and a device for determining a soil salinity stable yield threshold value of a cotton root zone. The method comprises the steps of constructing a structured salinity-yield database through literature integration, uniformly extracting site scale stable yield threshold values through index standardization and relative yield, generating comparable mechanism characteristics through combining crop models, quantifying uncertainty through light sampling, and utilizing an interpretable machine learning training threshold value mapping model to realize threshold extrapolation of a cotton crossing region/mode crossing. The method can output regional stable yield salinity threshold, uncertainty interval and evidence intensity, solves the threshold migration problem caused by the difference of root zone water-salt process under different modes of drip irrigation film covering, non-drip irrigation and the like, and provides scientific and interpretable decision basis for national regional mode cotton salinization regulation and control.

Inventors

  • HAO XINMEI
  • YAO XINMING
  • DING RISHENG

Assignees

  • 中国农业大学

Dates

Publication Date
20260505
Application Date
20260119

Claims (10)

  1. 1. The method for determining the soil salinity stable yield threshold value of the cotton root zone is characterized by comprising the following steps of: S10, constructing a structured cotton salinity-yield information base containing multi-source document data, and eliminating the data isomerism influence by unifying yield calculation calibers and marking salinity index types, measurement depths, time points and management modes; s20, extracting a stable yield salinity threshold value of a site-year scale based on a unified stable yield criterion, and adding a quality grade mark according to literature salinity gradient coverage conditions; s30, simulating a root region salinity dynamic process by using a crop model, extracting mechanism characteristics with physical semantics and quantifying uncertainty caused by the loss of key input parameters; S40, mapping the regional, mode, soil climate covariates and mechanism characteristics into stable yield thresholds by adopting an interpretable machine learning model, outputting regional thresholds and uncertainty intervals thereof, and analyzing characteristic contributions to support cross-regional extrapolation.
  2. 2. The method of claim 1, wherein S10 comprises: S101, collecting and recording literature meta information, test site information, test year information, planting/irrigation mode information, soil and climate background information, root zone salinity index information and yield information, and establishing a site-year-processing structured mapping relation; S102, defining SITEYEARID is used for uniquely identifying a site-year, defining TREATMENTID is used for uniquely identifying different processes in the same SiteYear, and recording a data source mode and an extraction error level.
  3. 3. The method of claim 1, wherein S20 comprises: S201, when other salinity conductivity indexes are given in the literature, converting to ECe according to the measurement conditions and conversion relation, retaining the original indexes and marking as non-ECe when the original indexes cannot be converted reliably, and taking the original indexes as covariates or removing the covariates when the model is trained; S202, adopting a formula The relative yield is calculated, and a yield stabilizing criterion is set to be Y r to be more than or equal to 0.90.
  4. 4. The method of claim 1, wherein S30 comprises: s301, operating a crop model to obtain root zone salinity dynamic ECroot (t) and a salt stress coefficient Ks (t), and calculating a stage salt stress integral: As a mechanism feature; S302, extracting the number of days of super-threshold exposure: root zone salt peak: As a salt exposure feature.
  5. 5. The method of claim 1, wherein S40 comprises: S401, constructing an input feature vector X, wherein the feature vector X at least comprises cotton distinguishing areas, irrigation/planting modes, soil textures, drought indexes and mechanism features; S402, adopting a cross-document verification strategy to evaluate extrapolation capability, preferably leaving document cross-verification, and outputting model error statistics.
  6. 6. A cotton root zone soil salinity stable yield threshold determining device, comprising: The structured information base construction module is used for constructing a structured cotton salinity-yield information base containing multi-source document data, calculating calibers through unified yield, marking salinity index types, measuring depths, time points and management modes, and eliminating data isomerism influence; The stable yield threshold extraction and quality identification module is used for extracting a stable yield salinity threshold of a site-year scale based on a unified stable yield criterion and adding quality grade identification according to literature salinity gradient coverage conditions; the crop model simulation and uncertainty quantization module is used for simulating a root region salinity dynamic process by using a crop model, extracting mechanism characteristics with physical semantics and quantifying uncertainty caused by the loss of key input parameters; the machine learning mapping and threshold outputting module is used for mapping the regional, mode, soil climate covariates and mechanism characteristics into stable yield thresholds by adopting an interpretable machine learning model, outputting regional thresholds and uncertainty intervals thereof, and analyzing characteristic contributions to support cross-regional extrapolation.
  7. 7. The apparatus of claim 6, wherein the structured information store construction module is further to: collecting and recording literature meta information, test site information, test year information, planting/irrigation mode information, soil and climate background information, root zone salinity index information and yield information, and establishing a site-year-processing structured mapping relation; Definition SITEYEARID is used to uniquely identify the site-year, definition TREATMENTID is used to uniquely identify the different processes within the same SiteYear, and record the data source mode and extraction error level.
  8. 8. The apparatus of claim 6, wherein the stable yield threshold extraction and quality identification module is further to: When other salinity conductivity indexes are given in the literature, converting to ECe according to the measurement conditions and conversion relation, retaining the original indexes and marking as non-ECe when the original indexes cannot be converted reliably, and taking the original indexes as covariates or removing the covariates when the model is trained; Using the formula The relative yield is calculated, and a yield stabilizing criterion is set to be Y r to be more than or equal to 0.90.
  9. 9. A computer device comprising a processor and a memory; Wherein the processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory for implementing a cotton root zone soil salinity stable production threshold determination method according to any one of claims 1-5.
  10. 10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the program when executed by a processor implements a cotton root zone soil salinity stable production threshold determination method according to any one of claims 1-5.

Description

Method and device for determining soil salinity stable yield threshold value of cotton root zone Technical Field The invention relates to the technical field of agricultural water and soil resource management, salinization farmland regulation and control and crop stable yield threshold identification, in particular to a cotton root zone soil salinity stable yield threshold determination method and device. Background Cotton is used as an important economic crop in China and is widely planted in arid-semiarid irrigation areas such as northwest inland and the like. The regional precipitation is rare, the evaporation and transpiration intensity is high, the agricultural production has large irrigation dependence, under the combined action of underground water burial depth, soil texture and irrigation system difference, salt is easy to accumulate and migrate seasonally in the root zone of crops, and the key growth period (such as the full bloom-boll bearing period) of cotton, biomass formation and stable yield are obviously limited. Therefore, the reasonable determination of the salinity stable yield threshold value of the cotton in different areas and management modes is an important basic problem in the salinization irrigation area stable yield enhancement and irrigation management decision. Existing engineering and agronomic researches generally adopt conductivity (ECe) of a root zone soil saturated paste extract to represent the salinity stress level, and fit a yield-salinity relationship through a 'threshold-yield reduction rate' type response relationship (such as a broken line function or an S-shaped function) so as to determine a salt stress initial threshold and yield reduction parameters. In crop model systems, salt stress generally reflects its inhibition of transpiration, dry matter accumulation and yield formation processes by stress coefficient Ks and corresponding ECe threshold parameters. In recent years, with the development of remote sensing technology and machine learning method, research and patent practice for carrying out soil salinity inversion, yield prediction or irrigation management optimization by utilizing multi-source data have also appeared so as to improve space coverage capability and prediction accuracy. However, the prior art has certain applicability under local research or single management conditions, but has obvious limitations in forming cotton salt stability thresholds for "regional scale, cotton division, or planting division modes" that can be used for engineering decisions. Firstly, soil layer depth, sampling time point (before sowing, during key growth period or at the end of season), salinity index type and conversion path of salinity measurement in different researches are inconsistent, the differences of yield caliber, variety background and water and fertilizer management conditions are obvious, and under the condition of lacking a unified stable yield criterion and data tracing mechanism, direct integration of literature data easily misjudges the difference of the measurement method as regional difference, so that systematic deviation of threshold value estimation is caused. Secondly, different irrigation and planting modes obviously change the water-salt migration process, for example, the drip irrigation tectorial membrane has essential differences with the root zone wetting body form, the leaching path and the salt space distribution under the condition of furrow irrigation or furrow irrigation, so that the salt exposure degree in the key growth period can be decoupled from the characterization result of the quaternary end ECe, the meaning of the same ECe value on stable yield under different management modes is not equivalent, and the cross-mode migration of the threshold value is obviously reduced. Again, traditional piecewise regression or empirical threshold fitting methods are highly sensitive to sample structure and salinity gradient coverage, threshold points are prone to fluctuation under conditions of sparse samples or large noise, confidence intervals, evidence strength assessment and uncertainty transfer mechanisms are usually lacking, and support engineering and regional application are difficult. Meanwhile, although the pure data driving method has advantages in the aspect of salinity or yield prediction precision, the error is mostly minimized as an objective function, and the threshold value physical semantics, the consistency of stable yield criteria and the lack of endophytic constraint on the trans-regional extrapolation boundary are difficult to directly output the salinity stable yield threshold value result with definite interpretation meaning and mobility. In the above background, providing an alternative technical route has clear practical motivation and application requirements. The field salinity threshold test has high cost and long period, is difficult to be developed systematically under the conditions of multiple cotton areas and m