Intelligent processing method and system for agrometeorological observation data

Abstract

The invention discloses an intelligent processing method and system for agrometeorological observation data, and relates to the technical field of agrometeorological data processing, and the method comprises the following steps: S1, receiving the agrometeorological observation data, including manual input data, Internet of Things equipment data and satellite remote sensing data; s2, performing block chain evidence storage operation on the agrometeorological observation data, and generating a non-tampering record including an operator signature, a data hash chain and a timestamp; s3, constructing a three-level cleaning mechanism to verify the agrometeorological observation data; s4, mapping the verified agrometeorological observation data to a geographic grid space, and fusing multi-source data to generate a space-time continuous data set; and S5, carrying out crop growth stage analysis, disaster influence evaluation and yield prediction based on the fused data, and outputting a result. According to the method, human tampering is completely eradicated from the source through full-link trace leaving, and meanwhile, data is scientific and reliable through data cleaning and verification.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of agricultural data processing, in particular to an intelligent processing method and system for agricultural meteorological observation data. BACKGROUND

[0002] Agricultural meteorological observation is an important basic work to support the development of modern agriculture and ensure food security. It systematically observes the growth and development of crops, field meteorological elements, agricultural disasters, and field work, etc., to provide key data support for the arrangement of agricultural activities, disaster warning and prevention, yield situation assessment, and climate change research.

[0003] At present, China's agricultural meteorological observation business mainly relies on the combination of manual observation and automatic instrument monitoring. The observation data is recorded in traditional paper observation books or simple electronic input systems, and then sorted and reported to the business platform by manual or semi-automatic methods. However, the existing technical system has obvious deficiencies in data processing efficiency, quality control, depth analysis and application, mainly in the following aspects: data quality is difficult to guarantee: manual observation records are easily affected by subjective factors, especially in the aspects of crop development period determination and disaster identification which require experience, and the records of different observers may be biased; multi-source data fusion and utilization is insufficient: the current business system usually stores and manages sensor data, satellite inversion data, etc. in isolation, and lacks effective technical means to integrate and cooperatively analyze them with traditional manual observation data; the current data processing process mainly focuses on "recording" and "reporting", and fails to fully utilize historical data and modern algorithm models to establish growth prediction, disaster loss assessment, yield prediction, etc.

[0004] A Chinese patent with publication number CN115619583A discloses a construction method of a composite agricultural meteorological disaster monitoring index system, including the following steps: S1, determining a rice agricultural meteorological observation station, using continuous 5-6 years of rice observation data to determine the sensitivity parameters of process variables such as LAI, aboveground biomass and nitrogen accumulation, and realizing the simulation of yield and grain protein content by CERES-RICE model; S2, constructing a rice quality prediction model based on spectrum-statistics; S3, after localizing the parameters of DSSAT model and assimilating remote sensing data and crop model, introducing time series crop growth remote sensing data based on GIS technology, and calling crop model by Matlab to realize fine simulation at regional scale; S4, based on step S3, according to the spatial and temporal distribution law of major composite disasters, establishing a composite agricultural meteorological disaster monitoring index system; providing technical support for stabilizing the position of national important grain production base, ensuring stable and high yield of agriculture, and guaranteeing food security.

[0005] Chinese Patent CN105303063B, for example, discloses a method and system for leaf area index (LAI) inversion that integrates phenological data and remote sensing data. The method includes establishing several sampling points in a target area; measuring the LAI of plants at each sampling point in the target area under diffuse light meteorological conditions using a plant canopy analyzer; recording the vegetation phenological period of the target area at the time of measurement; averaging the LAI of each sampling point to obtain its true LAI; acquiring multispectral remote sensing images of the same target area observed at the same time, preprocessing them to obtain true reflectance images, and calculating the vegetation index of each sampling point; performing correlation analysis using the vegetation index and the true LAI to obtain a quantitative relationship model corresponding to the vegetation phenological period of the target area at the time of measurement; and inverting and analyzing the plant growth status of the target area at the corresponding vegetation phenological period based on the model. This invention can meet the dynamic monitoring needs of large-area, long-term vegetation LAI, solves the problem of field measurement, and meets the needs of agricultural and forestry applications. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to address the shortcomings of the existing technology by providing an intelligent processing method and system for agricultural meteorological observation data.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] Intelligent processing methods for agricultural meteorological observation data include the following steps:

[0009] Step S1: Receive agricultural meteorological observation data, including manually entered data, IoT device data, and satellite remote sensing data;

[0010] Step S2: Perform blockchain notarization on agricultural meteorological observation data to generate an immutable record containing the operator's signature, data hash chain, and timestamp;

[0011] Step S3: Construct a three-level cleaning mechanism to verify agricultural meteorological observation data. The three-level cleaning mechanism includes: a basic layer to verify the compliance of numerical range, decimal places, and enumerated values; a rule layer to verify the consistency of logic by applying crop-specific business rules; and an AI layer to call a pre-trained LSTM development prediction model to identify data anomalies.

[0012] Step S4: Map the verified agricultural meteorological observation data to a geographic grid space and fuse multi-source data to generate a spatiotemporal continuous dataset;

[0013] Step S5: Analyze crop growth stages, assess disaster impacts, and predict yields based on the fused data, and output the results.

[0014] Furthermore, in step S3, the consistency verification logic of crop-specific business rules applied at the rule layer specifically includes:

[0015] Developmental period time sequence conflict detection: Construct a finite state machine to enforce a strict temporal relationship between the developmental period date sequence: sowing date is earlier than emergence date, which is earlier than heading date, which is earlier than maturity date. If a reverse order is detected, it is marked as a logical error.

[0016] Disaster meteorological correlation verification: When disaster data is entered, the system automatically retrieves and verifies the meteorological data of the same period: For freezing damage events, it verifies whether the lowest temperature on the day of occurrence is less than or equal to 0℃; for drought events, it verifies whether the cumulative precipitation anomaly during the occurrence period is less than or equal to -50%.

[0017] Density-Yield Coupling Verification: Ensuring that plant density meets theoretical yield requirements.

[0018] Y∈[0.8·k·D,1.2·k·D]

[0019] Where Y represents theoretical yield, k represents the empirical coefficient of crop variety, which is calibrated through historical experimental data, and D represents plant density.

[0020] Furthermore, in step S3, the anomaly detection model of the AI ​​layer is a crop-specific LSTM development prediction model. The model is input with a continuous sequence of environmental parameters, including daily minimum temperature, daily maximum temperature, precipitation, sunshine duration and development stage encoding, and outputs the predicted date of the target development period.

[0021] Among them, a credible weighted arbitration mechanism is triggered when the deviation between the actual observed development date and the predicted development date exceeds twice the historical standard deviation of the same crop development period in the past 30 days.

[0022] Furthermore, the credible weighted arbitration mechanism specifically includes the following steps:

[0023] Retrieve multi-source verification data, including manual observation data, field data collected in real time by IoT sensors, and phenological data retrieved from satellites;

[0024] Construct a reliable weighted algorithm to calculate the final arbitration value;

[0025] When the absolute error between the final arbitration value and the actual observed development date is less than or equal to the historical standard deviation, the actual observed development date is determined to be uploaded incorrectly; when the absolute error between the final arbitration value and the actual observed development date is greater than the historical standard deviation, the prediction model is determined to be invalid, triggering an online model update that uses the final arbitration value as a new training sample.

[0026] Furthermore, in step S5, the crop growth stage analysis includes;

[0027] Developmental anomalies were calculated based on the average values ​​of the same period over the past 10 years.

[0028] Growth rate modeling was completed by fitting plant height changes using the Logistic function.

[0029] The leaf area index is calculated by combining dry matter weight and specific leaf area.

[0030] Furthermore, in step S5, the disaster impact assessment is achieved by establishing a nonlinear response function between disaster intensity and production reduction rate.

[0031] Furthermore, in step S5, the production forecast specifically includes:

[0032] Yield prediction is made by integrating and verifying density, number of ears, number of grains, and thousand-grain weight data;

[0033] Temperature stress coefficient and moisture stress coefficient are introduced to correct the yield forecast for environmental stress.

[0034] An intelligent processing system for agricultural meteorological observation data includes:

[0035] The multi-source access module is used to collect data from manual observation, IoT sensor data, and satellite data.

[0036] The data verification module is used to clean and verify the collected multi-source data through a three-level cleaning mechanism;

[0037] The intelligent processing output module is used to generate development period corrections, disaster warnings, and intelligent yield forecasts based on the verified data.

[0038] A blockchain-based evidence storage network is used to record hash values ​​of data operations and arbitration processes.

[0039] The spatiotemporal grid fusion module is used to transform discrete agricultural meteorological observation data into spatiotemporally continuous high-resolution grid datasets.

[0040] Furthermore, the multi-source access module specifically includes:

[0041] Manual observation terminals are used for structured recording of crop phenological stages, disaster events, and agricultural operations, with data formats conforming to agricultural meteorological observation standards;

[0042] Satellite data interface, used to retrieve phenological periods from satellite data in real time:

[0043] Sensor gateway, used to parse soil temperature and humidity array data transmitted via wireless protocol.

[0044] Furthermore, the blockchain-based evidence storage network includes:

[0045] The data chain is used to store the original observation values, arbitration results, and operator digital signatures;

[0046] An audit chain is used to record user login, data modification, and verification operation logs.

[0047] Cross-chain validators are used to associate two chains through timestamp hashes and support tracing the entire lifecycle of operation records by inputting data IDs.

[0048] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0049] 1. This invention achieves multi-source cross-validation in the determination of abnormalities in the development period through a trusted weighted arbitration mechanism, which can effectively reduce the error rate of crop development period recording. When the heading date is entered incorrectly by manual input, the system automatically retrieves satellite data and sensor growth parameters, calculates the arbitration value by weighting, corrects the deviation, and ensures that the data is scientific and reliable.

[0050] 2. This invention constructs a blockchain evidence storage network to record the entire chain of observation operations and arbitration processes. The operator's signature is bound to the timestamp hash, and any data modification must be verified by multi-node consensus, thus preventing human tampering from the source.

[0051] 3. This invention adopts a nonlinear disaster response model and combines multi-source data after credible arbitration to quantify the relationship between disaster intensity and loss. For example, the introduction of arbitration-corrected temperature data in the frost damage early warning helps to improve the accuracy of the early warning.

[0052] 4. This invention utilizes a spatiotemporal grid fusion engine to transform discrete site data into a continuous grid dataset, and combines it with an LSTM model for real-time prediction, thereby improving the timeliness of decision-making. Attached Figure Description

[0053] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0054] Figure 1 This is a flowchart illustrating an embodiment of the present invention;

[0055] Figure 2 This is a schematic diagram of the system architecture according to an embodiment of the present invention;

[0056] Figure 3 This is a schematic diagram of the system visualization interface according to an embodiment of the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] like Figure 1 As shown, the intelligent processing method for agricultural meteorological observation data includes the following steps:

[0059] Step S1: Receive agricultural meteorological observation data, including manually entered data, IoT device data, and satellite remote sensing data;

[0060] Manually entered crop growth period records (including dates of sowing, emergence, heading, maturity, and other developmental stages);

[0061] Field environmental data collected by IoT devices (including real-time parameters such as air temperature, soil moisture, and light intensity);

[0062] Vegetation index data acquired by satellite remote sensing (including Normalized Difference Vegetation Index (NDVI) data retrieved from Sentinel1-2 or Landsat-8 satellites).

[0063] Step S2: Perform blockchain notarization on the agricultural meteorological observation data to generate an immutable record containing the operator's signature, data hash chain, and timestamp, specifically including:

[0064] The digital signature generated by the operator based on the national cryptographic algorithm SM2;

[0065] A hash chain formed by data version changes (each new version is associated with the hash value of the previous version);

[0066] A timestamp in UTC (Coordinated Universal Time) format.

[0067] Step S3: Construct a three-level cleaning mechanism to verify agricultural meteorological observation data. The three-level cleaning mechanism includes: a basic layer to verify the compliance of numerical range, decimal places, and enumerated values; a rule layer to verify the consistency of logic by applying crop-specific business rules; and an AI layer to call a pre-trained LSTM development prediction model to identify data anomalies.

[0068] Basic layer verification: Verify whether the values ​​are within the preset physical range (e.g., temperature -50℃ to 60℃), whether the number of decimal places meets the precision requirements (e.g., soil moisture retains 1 decimal place), and whether the enumerated values ​​are compliant (e.g., the developmental stage name is in the preset dictionary).

[0069] Step S4: Map the verified agricultural meteorological observation data to a geographic grid space and fuse multi-source data to generate a spatiotemporal continuous dataset;

[0070] The validated data is mapped to a spatial grid in the WGS-84 geographic coordinate system, and a spatiotemporally continuous dataset with a resolution of 1 km × 1 km is generated using a spatiotemporal fusion algorithm, specifically including:

[0071] Kriging interpolation was used to convert discrete station data into a regular grid.

[0072] Resample satellite data to the target grid resolution;

[0073] It integrates three types of data sources: human, sensor, and satellite.

[0074] Step S5: Analyze crop growth stages, assess disaster impacts, and predict yields based on the fused data, and output the results.

[0075] In step S3, the application of crop-specific business rules to verify the consistency of the rules layer specifically includes:

[0076] Developmental period time sequence conflict detection: Construct a finite state machine to enforce a strict temporal relationship between the developmental period date sequence: sowing date is earlier than emergence date, which is earlier than heading date, which is earlier than maturity date. If a reverse order is detected, it is marked as a logical error.

[0077] Disaster meteorological correlation verification: When disaster data is entered, the system automatically retrieves and verifies the meteorological data of the same period: For freezing damage events, it verifies whether the lowest temperature on the day of occurrence is less than or equal to 0℃; for drought events, it verifies whether the cumulative precipitation anomaly during the occurrence period is less than or equal to -50%.

[0078] Density-Yield Coupling Verification: Ensuring that plant density meets theoretical yield requirements.

[0079] Y∈[0.8·k·D,1.2·k·D]

[0080] Where Y represents theoretical yield, k represents the empirical coefficient of crop variety, which is calibrated through historical experimental data, and D represents plant density.

[0081] In step S3, the anomaly detection model of the AI ​​layer is a crop-specific LSTM development prediction model. The model is input with a continuous sequence of environmental parameters, including daily minimum temperature, daily maximum temperature, precipitation, sunshine duration and development stage coding, and outputs the date prediction value of the target development period.

[0082] Among them, a credible weighted arbitration mechanism is triggered when the deviation between the actual observed development date and the predicted development date exceeds twice the historical standard deviation of the same crop development period in the past 30 days.

[0083] The credible weighted arbitration mechanism specifically includes the following steps:

[0084] Retrieve multi-source verification data, including manual observation data, field data collected in real time by IoT sensors, and phenological data retrieved from satellites;

[0085] A trusted weighted algorithm is constructed to calculate the final arbitration value. The specific formula is as follows:

[0086]

[0087] Among them, W i fThe final arbitration value represents the anomaly determination result after weighted fusion of multi-source data. `i` represents the data source index, indicating the i-th type of data source participating in the arbitration. `i=1` represents manual observation, `i=2` represents IoT sensors, and `i=3` represents satellite remote sensing. `W` represents the final arbitration value. i C represents the original observation value. i R represents the blockchain evidence credibility of data source i (0-1). i The historical accuracy coefficient of data source i is determined by calibration experiments;

[0088] Manual observation: R = 1.0 (default baseline);

[0089] Sensor: R = 0.8 (affected by equipment error);

[0090] Satellite remote sensing: R = 0.7 (affected by cloud cover);

[0091] The formula for calculating the credibility of blockchain evidence is:

[0092] C i =x·BS+z·DI

[0093] Where BS represents the proportion of operation records uploaded to the chain, DI represents the data field completeness, and x and z represent empirical coefficients, which are usually set to x = 0.7 and z = 0.3.

[0094] When the absolute error between the final arbitration value and the actual observed development date is less than or equal to the historical standard deviation, the actual observed development date is determined to be uploaded incorrectly; when the absolute error between the final arbitration value and the actual observed development date is greater than the historical standard deviation, the prediction model is determined to be invalid, triggering an online model update that uses the final arbitration value as a new training sample.

[0095] During the winter wheat heading stage observation in 2024, a pilot area manually recorded the date as April 18th, but sensor data indicated that the accumulated temperature had already reached the heading threshold (April 12th), and the inflection point of the satellite NDVI curve appeared on April 14th. The system arbitration finalized the date as April 14th, and subsequent verification confirmed the accuracy of this date. Statistically, this type of arbitration reduced the error rate of developmental stage recordings by 42% (from an average error of 4 days to 2.3 days).

[0096] In step S5, the crop growth stage analysis includes;

[0097] The developmental anomaly is calculated based on the average of the same period over the past 10 years, using the following formula:

[0098] ΔT=T c -T h

[0099] Where ΔT represents the developmental anomaly, which is the difference between the current developmental date and the historical average developmental date. cT represents the actual date of a currently observed developmental stage (usually expressed as the day of the year). h This indicates the average date (day of the year) for this developmental period within the same historical period (usually the past 10 years);

[0100] Growth rate modeling is achieved by fitting plant height changes using the Logistic function. The specific formula is as follows:

[0101]

[0102] Where H(t) represents the plant height at time t, K represents the plant's maximum height, determined by the crop's genetic characteristics, r represents the growth rate parameter, controlling the steepness of the curve's ascent, t represents the current time (usually expressed as the day of the year), and T... m This indicates the point in time when the plant grows fastest (unit: days), i.e., the inflection point of the growth curve;

[0103] The leaf area index is calculated by combining dry matter weight and specific leaf area. The specific formula is as follows:

[0104] L=W d ·SLA·D

[0105] Where L represents the leaf area index, which represents the total leaf area per unit land area, and W... d SLA represents the dry matter weight per plant, while SLA represents the specific leaf area, which is the leaf area per unit dry weight. A typical value is 100-200 cm² for rice. 2 / g, where D represents plant density.

[0106] In step S5, the disaster impact assessment is achieved by establishing a nonlinear response function between disaster intensity and yield reduction rate, specifically using the following formula:

[0107] η=α·I β +γ

[0108] Where η represents the rate of yield reduction caused by the disaster (a decimal between 0 and 1, e.g., 0.1 represents a 10% reduction), α, β, and γ represent model parameters obtained through regression analysis of historical disaster data, and I represents the disaster intensity. For example, for frost damage: α = 0.15, β = 1.2, γ = 0.03.

[0109] Traditional early warning accuracy: Frost damage early warning based on single meteorological data (such as minimum temperature ≤ 0℃) has a high false alarm rate because it does not take into account factors such as crop cold resistance and field microclimate.

[0110] The accuracy of early warning in this invention is improved to 92% by using a nonlinear response function and incorporating multi-source data (such as satellite inversion data and sensor data) after arbitration correction.

[0111] For example, in a certain late spring cold snap, a traditional weather station reported a minimum temperature of -1.5℃ (which should trigger a frost damage warning), but the system of this invention retrieves data from multiple sources:

[0112] Satellite inversion shows that the actual temperature only dropped to 0.8℃;

[0113] The lowest canopy temperature recorded by field sensors was 0.5℃;

[0114] After arbitration calculations, it was determined that the actual damage threshold had not been reached, and no warning was issued.

[0115] In step S5, the production forecast specifically includes:

[0116] Yield prediction is performed by integrating and validating density, ear number, grain number, and thousand-grain weight data. The specific formula is as follows:

[0117] Y pred =D·N e ·N g ·W g ·10 -6

[0118] Among them, Y pred N represents the predicted base output. e N represents the average number of ears per plant. g W represents the average number of grains per ear. g Indicates the weight of 1000 grains;

[0119] Environmental stress corrections are applied to yield forecasts using temperature and moisture stress coefficients, as specified in the formula:

[0120] Y f =Y pred ·(1-max(S T S W ))

[0121] Among them, Y f S represents the final predicted yield after adjustments for environmental stress. T The temperature stress coefficient (between 0 and 1) represents the degree of impact of high or low temperatures on yield, S. W The water stress coefficient (between 0 and 1) indicates the degree of impact of drought or waterlogging on yield.

[0122] The temperature stress coefficient is calculated as |Daily average temperature - Crop optimum temperature / Tolerance threshold;

[0123] The water stress coefficient is calculated as |soil moisture - field capacity / wilting coefficient.

[0124] Traditional analysis time-consuming: From data collection, manual entry, and processing to the generation of regional yield assessment reports, it usually takes 3-5 hours.

[0125] The invention improves analysis efficiency: through spatiotemporal grid fusion and real-time prediction using LSTM models, and with fully automated processing, the time to generate a production assessment report can be reduced to within 10 minutes.

[0126] Specific example: In a certain production assessment, the system accessed manual observation data from 125 stations, 300+ IoT sensors, and daily updated satellite remote sensing data.

[0127] Data cleaning and fusion time: 2 minutes; L-model prediction and stress correction time: 1 minute; Generation of visualized grid output distribution map and report: 2 minutes; Total time: 5 minutes.

[0128] like Figure 2 As shown, the intelligent processing system for agricultural meteorological observation data includes:

[0129] The multi-source access module is used to collect data from manual observation, IoT sensor data, and satellite data.

[0130] The data verification module is used to clean and verify the collected multi-source data through a three-level cleaning mechanism;

[0131] The intelligent processing output module is used to generate development period corrections, disaster warnings, and intelligent yield forecasts based on the verified data.

[0132] A blockchain-based evidence storage network is used to record hash values ​​of data operations and arbitration processes.

[0133] The spatiotemporal grid fusion module is used to transform discrete agricultural meteorological observation data into spatiotemporally continuous high-resolution grid datasets.

[0134] The multi-source access module specifically includes:

[0135] Manual observation terminals are used for structured recording of crop phenological stages, disaster events, and agricultural operations, with data formats conforming to agricultural meteorological observation standards;

[0136] Satellite data interface, used to retrieve phenological periods from satellite data in real time:

[0137] Sensor gateway, used to parse soil temperature and humidity array data transmitted via wireless protocol.

[0138] The blockchain-based evidence storage network includes:

[0139] The data chain is used to store the original observation values, arbitration results, and operator digital signatures;

[0140] An audit chain is used to record user login, data modification, and verification operation logs.

[0141] Cross-chain validators are used to associate two chains through timestamp hashes and support tracing the entire lifecycle of operation records by inputting data IDs.

[0142] The spatiotemporal grid fusion module performs the following:

[0143] Spatial interpolation: The Kriging algorithm is used to convert site data into a regular grid;

[0144] Spatiotemporal alignment: resampling satellite data to the target grid resolution;

[0145] Multi-source fusion: Integrate human, sensor, and satellite data sources according to trusted weights to generate a spatiotemporally continuous data cube.

[0146] The system deployment adopts an edge-cloud collaborative architecture:

[0147] Edge nodes: Deployed at field weather stations to perform data cleaning and real-time arbitration (response time <200ms);

[0148] Cloud Center: Runs LSTM prediction models and blockchain consensus mechanisms (PBFT algorithm).

[0149] like Figure 3 As shown, the system parameter configuration includes four modules: station information, user configuration, role configuration, and personal center. Station information allows maintenance of station information; user configuration allows adding station users; role configuration allows managing user permissions; and the personal center allows maintenance of the user's information. Selecting the province, station, and station type (including agricultural gas stations and agricultural test stations) and clicking the search button will display data matching the criteria in a list.

[0150] This system also includes:

[0151] Real-time validation rules for manual data entry:

[0152] When you enter numerical data, the system automatically compares it to a preset reasonable range. If the data exceeds the range, a red warning box will immediately pop up to indicate that the data is abnormal, and you must confirm or correct it before submitting.

[0153] IoT data transmission guarantee mechanism:

[0154] Resuming interrupted transmission: When the device detects a network interruption, it automatically caches unsent data packets and prioritizes retransmitting the data during the interruption after the connection is restored. It also generates a transmission interruption event log in the management backend for verification.

[0155] Signal interference handling: When the receiving end detects a data packet that fails to verify three times in a row, it immediately triggers a retransmission request; if no valid data is received within 5 minutes, the time period is marked as a data missing state and an alarm is pushed to the operation and maintenance terminal.

[0156] Integrity verification: All transmitted data must include a CRC checksum. If the receiving end fails to verify the data, it will discard the data packet and notify the device to resend it.

[0157] The examples described herein are merely preferred embodiments of the invention and are not intended to limit the concept and scope of the invention. Any modifications and improvements made by those skilled in the art to the technical solutions of the invention without departing from the design concept of the invention should fall within the protection scope of the invention.

Claims

1. An intelligent processing method for agricultural meteorological observation data, characterized in that, Includes the following steps: Step S1: Receive agricultural meteorological observation data, including manually entered data, IoT device data, and satellite remote sensing data; Step S2: Perform blockchain notarization on agricultural meteorological observation data to generate an immutable record containing the operator's signature, data hash chain, and timestamp; Step S3: Construct a three-level cleaning mechanism to verify agricultural meteorological observation data. The three-level cleaning mechanism includes: a basic layer to verify the compliance of numerical range, decimal places, and enumerated values; a rule layer to verify the consistency of logic by applying crop-specific business rules; and an AI layer to call a pre-trained LSTM development prediction model to identify data anomalies. Step S4: Map the verified agricultural meteorological observation data to a geographic grid space and fuse multi-source data to generate a spatiotemporal continuous dataset; Step S5: Analyze crop growth stages, assess disaster impacts, and predict yields based on the fused data, and output the results.

2. The method according to claim 1, characterized in that, In step S3, the application of crop-specific business rules to verify the consistency of the rules layer specifically includes: Developmental period time sequence conflict detection: Construct a finite state machine to enforce a strict temporal relationship between the developmental period date sequence: sowing date is earlier than emergence date, which is earlier than heading date, which is earlier than maturity date. If a reverse order is detected, it is marked as a logical error. Disaster meteorological correlation verification: When disaster data is entered, the system automatically retrieves and verifies the meteorological data of the same period: For freezing damage events, it verifies whether the lowest temperature on the day of occurrence is less than or equal to 0℃; for drought events, it verifies whether the cumulative precipitation anomaly during the occurrence period is less than or equal to -50%. Density-Yield Coupling Verification: Ensuring that plant density meets theoretical yield requirements. Y∈[0.8·k·D,1.2·k·D] Where Y represents theoretical yield, k represents the empirical coefficient of crop variety, which is calibrated through historical experimental data, and D represents plant density.

3. The method according to claim 2, characterized in that, In step S3, the anomaly detection model of the AI ​​layer is a crop-specific LSTM development prediction model. The model is input with a continuous sequence of environmental parameters, including daily minimum temperature, daily maximum temperature, precipitation, sunshine duration and development stage coding, and outputs the date prediction value of the target development period. Among them, a credible weighted arbitration mechanism is triggered when the deviation between the actual observed development date and the predicted development date exceeds twice the historical standard deviation of the same crop development period in the past 30 days.

4. The method according to claim 3, characterized in that, The credible weighted arbitration mechanism specifically includes the following steps: Retrieve multi-source verification data, including manual observation data, field data collected in real time by IoT sensors, and phenological data retrieved from satellites; Construct a reliable weighted algorithm to calculate the final arbitration value; When the absolute error between the final arbitration value and the actual observed development date is less than or equal to the historical standard deviation, the actual observed development date is determined to be uploaded incorrectly; when the absolute error between the final arbitration value and the actual observed development date is greater than the historical standard deviation, the prediction model is determined to be invalid, triggering an online model update that uses the final arbitration value as a new training sample.

5. The method according to claim 4, characterized in that, In step S5, the crop growth stage analysis includes; Developmental anomalies were calculated based on the average values ​​of the same period over the past 10 years. Growth rate modeling was completed by fitting plant height changes using the Logistic function. The leaf area index is calculated by combining dry matter weight and specific leaf area.

6. The method according to claim 5, characterized in that, In step S5, the disaster impact assessment is achieved by establishing a nonlinear response function between disaster intensity and production reduction rate.

7. The method according to claim 6, characterized in that, In step S5, the production forecast specifically includes: Yield prediction is made by integrating and verifying density, number of ears, number of grains, and thousand-grain weight data; Temperature stress coefficient and moisture stress coefficient are introduced to correct the yield forecast for environmental stress.

8. An intelligent processing system for agricultural meteorological observation data, characterized in that: include: The multi-source access module is used to collect data from manual observation, IoT sensor data, and satellite data. The data verification module is used to clean and verify the collected multi-source data through a three-level cleaning mechanism; The intelligent processing output module is used to generate development period corrections, disaster warnings, and intelligent yield forecasts based on the verified data. A blockchain-based evidence storage network is used to record hash values ​​of data operations and arbitration processes. The spatiotemporal grid fusion module is used to transform discrete agricultural meteorological observation data into spatiotemporally continuous high-resolution grid datasets.

9. The system according to claim 8, characterized in that, The multi-source access module specifically includes: Manual observation terminals are used for structured recording of crop phenological stages, disaster events, and agricultural operations, with data formats conforming to agricultural meteorological observation standards; Satellite data interface, used for real-time retrieval of phenological periods from satellite data: Sensor gateway, used to parse soil temperature and humidity array data transmitted via wireless protocol.

10. The system according to claim 9, characterized in that, The blockchain-based evidence storage network includes: The data chain is used to store the original observation values, arbitration results, and operator digital signatures; An audit chain is used to record user login, data modification, and verification operation logs. Cross-chain validators are used to associate two chains through timestamp hashes and support tracing the entire lifecycle of operation records by inputting data IDs.

Images