A Method for Assessing the Climate Risk of Spring Low Temperature Disasters for Winter Wheat

Abstract

This application relates to a method for assessing the climate risk of spring low-temperature disasters for winter wheat, comprising: acquiring multi-source data for a target time period within a target area; dividing the spring development period of winter wheat into different developmental stages based on the multi-source data, and determining the hourly minimum temperature and accumulated chilling amount for each developmental stage based on meteorological data; determining the degree of damage to winter wheat at different developmental stages based on the hourly minimum temperature and accumulated chilling amount, and determining the probability of occurrence of spring low-temperature disasters for winter wheat based on the degree of damage; determining the average yield reduction rate and the coefficient of variation of the yield reduction rate of winter wheat based on crop data, and constructing a climate risk index for spring low-temperature disasters for winter wheat based on the probability of occurrence of spring low-temperature disasters for winter wheat, the average yield reduction rate of winter wheat, and the coefficient of variation of the yield reduction rate of winter wheat; and classifying the risk level based on the climate risk index for spring low-temperature disasters for winter wheat. This application improves the accuracy of quantitative assessment of the climate risk of spring low-temperature disasters for winter wheat.

Description

Technical Field

[0001] This application relates to the field of agricultural technology, and in particular to a method for assessing the climate risk of low-temperature disasters in winter wheat during spring, a computer device, and a computer-readable storage medium. Background Technology

[0002] Low-temperature disasters are sudden agricultural meteorological disasters characterized by frequent occurrences, wide-ranging impacts, delayed onset, significant temperature drops, and prolonged durations. While climate warming has significantly reduced the frequency and intensity of low-temperature disasters in my country, the risk remains high in actual production due to earlier crop growth periods, greater fluctuations in climate factors, and drastic temperature changes, posing a serious challenge to the sustainable development of agriculture in my country.

[0003] In my country, wheat is the second largest grain crop, divided into spring wheat and winter wheat. Winter wheat accounts for approximately 95% of China's total wheat production and is crucial to the country's food security. Therefore, timely and accurate assessment of the climate risk of spring low-temperature disasters for winter wheat is of significant practical importance and social value for ensuring national food security, comprehensively promoting rural revitalization, and further enhancing my country's agricultural capacity to cope with climate change.

[0004] In recent decades, my country has made great progress in agricultural meteorological disaster risk assessment research in terms of methodological improvement, application of new technologies, and system construction. However, the severity of low-temperature disasters affecting winter wheat is influenced by a combination of factors, including atmospheric, natural environmental, and physiological factors. The relationships between these factors are complex, and existing low-temperature disaster indicators based on meteorological elements are mostly diurnal in scale, making it difficult to accurately assess climate risk and meet the actual needs of local agricultural disaster prevention and mitigation and food security.

[0005] Therefore, there is an urgent need for a refined assessment method for the climate risk of low-temperature disasters in winter wheat during the spring. Summary of the Invention

[0006] (a) Technical problems to be solved

[0007] In view of the above-mentioned shortcomings and deficiencies of the prior art, this application provides a method for assessing the climate risk of low temperature disasters in winter wheat in spring. This method solves the technical problem that the prior art lacks a refined assessment technology for the risk of low temperature disasters in winter wheat in spring, making it difficult to accurately assess its climate risk and failing to meet the actual needs of local agricultural disaster prevention and mitigation and food security.

[0008] (II) Technical Solution

[0009] To achieve the above objectives, the main technical solutions adopted in this application include:

[0010] In a first aspect, embodiments of this application provide a method for assessing the climate risk of spring low-temperature disasters for winter wheat, including:

[0011] Acquire multi-source data for a target time period within a target area; wherein the multi-source data includes meteorological data, crop data, winter wheat planting distribution data, soil moisture data, and vegetation growth data; divide the spring development period of winter wheat into different developmental periods based on the multi-source data, and determine the hourly minimum temperature and accumulated chilling amount for each developmental period based on the meteorological data; determine the degree of damage to winter wheat in different developmental periods based on the hourly minimum temperature and accumulated chilling amount, and determine the probability of occurrence of spring low-temperature disasters for winter wheat based on the degree of damage to winter wheat in different developmental periods; determine the average yield reduction rate and the coefficient of variation of winter wheat yield reduction rate based on the crop data, and construct a climate risk index for spring low-temperature disasters for winter wheat based on the probability of occurrence of spring low-temperature disasters for winter wheat, the average yield reduction rate, and the coefficient of variation of winter wheat yield reduction rate; classify the risk level based on the climate risk index for spring low-temperature disasters for winter wheat.

[0012] Optionally, in some embodiments of this application, the spring development period of winter wheat is divided into different developmental stages based on the multi-source data, including: extracting cumulative accumulated temperature features, low-temperature stress intensity index features, soil moisture status features, and vegetation growth status index features that are dynamically related to the development process of winter wheat based on the multi-source data, and performing spatiotemporal alignment processing to obtain a temporal feature sequence; inputting the temporal feature sequence into a pre-trained dynamic prediction model for the development period of winter wheat to obtain the prediction results of the dynamic start and end times of each developmental stage; and dividing the spring development period of winter wheat into the greening stage, jointing stage, booting stage, and heading-flowering stage based on the prediction results of the dynamic start and end times of each developmental stage.

[0013] Optionally, in some embodiments of this application, the dynamic prediction model for winter wheat development is trained based on a multimodal long short-term memory neural network, wherein the multimodal long short-term memory neural network includes: a multi-channel feature embedding layer, used to normalize and embed key features in the input temporal feature sequence to generate feature vectors with temporal correlation; a spatiotemporal feature fusion layer, used to extract long-term temporal dependencies between the feature vectors using a bidirectional LSTM layer, and to capture the dynamic interaction effect between the mutation point of the low-temperature stress intensity index and the cumulative temperature trend, so as to obtain a temporal feature sequence after fusing spatiotemporal features; a stress response attention layer, used to introduce a variety-specific adjustment factor to weight the temporal feature sequence after fusing spatiotemporal features according to the variety resistance coefficient and the temperature sensitivity threshold of the key development period in the crop data, so as to obtain a temporal feature sequence after attention weighting; and a development period transition decision layer, used to output the starting probability distribution of each development period using a fully connected layer and a temporal softmax layer, and to determine the dynamic start and end time prediction results of each development period according to the starting probability distribution.

[0014] Optionally, in some embodiments of this application, determining the hourly minimum temperature and process cold accumulation for different developmental periods based on the meteorological data includes: determining the hourly minimum temperature and the duration of hourly low temperature based on the meteorological data; and determining the process cold accumulation based on the hourly minimum temperature and the duration of hourly low temperature.

[0015] Optionally, in some embodiments of this application, the process cooling accumulation is determined according to the following formula:

[0016]

[0017] Among them, X H T represents the accumulated cooling capacity of the process. min For the hourly minimum temperature, T c H represents the lower limit temperature for growth at different developmental stages of winter wheat, and H represents the duration of low temperature in hours.

[0018] Optionally, in some embodiments of this application, determining the degree of damage to winter wheat at different developmental stages based on the minimum hourly temperature and accumulated cooling during the different developmental stages includes: obtaining the number of times winter wheat was damaged at different developmental stages based on the minimum hourly temperature and accumulated cooling during the different developmental stages; obtaining the degree of damage for each damage based on the number of times winter wheat was damaged at different developmental stages; determining the degree of damage to winter wheat as severe when at least one of the minimum hourly temperature and the accumulated cooling during the different developmental stages is in the severe damage range of the corresponding developmental stage; determining the degree of damage to winter wheat as moderate when both the minimum hourly temperature and the accumulated cooling during the different developmental stages are in the moderate damage range of the corresponding developmental stage; determining the degree of damage to winter wheat as mild when both the minimum hourly temperature and the accumulated cooling during the different developmental stages are in the mild damage range of the corresponding developmental stage; and determining the degree of damage to winter wheat as mild when one of the minimum hourly temperature and the accumulated cooling during the different developmental stages is in the mild damage range of the corresponding developmental stage and the other is in the moderate damage range of the corresponding developmental stage.

[0019] Optionally, in some embodiments of this application, determining the probability of spring low-temperature disasters occurring in winter wheat based on the degree of damage suffered by the winter wheat at different developmental stages includes: obtaining the number of times the winter wheat was lightly damaged, moderately damaged, and severely damaged, respectively, based on the degree of damage suffered by the winter wheat at different developmental stages; and determining the probability of spring low-temperature disasters occurring in winter wheat based on the number of times the winter wheat was lightly damaged, moderately damaged, and severely damaged.

[0020] Optionally, in some embodiments of this application, determining the average yield reduction rate and the coefficient of variation of the winter wheat yield reduction rate based on the crop data includes: obtaining the actual yield and the maximum possible yield of the target area based on the crop data; determining the trend yield based on the maximum possible yield, and determining the meteorological yield based on the actual yield and the trend yield, and determining the relative meteorological yield based on the trend yield and the meteorological yield; determining the average yield reduction rate of winter wheat based on the relative meteorological yield, and determining the coefficient of variation of the winter wheat yield reduction rate based on the relative meteorological yield and the average yield reduction rate of winter wheat.

[0021] Secondly, embodiments of this application provide a computer device, the computer device including a processor and a memory, the processor being used to execute instructions stored in the memory, so that the computer device executes the winter wheat spring low temperature disaster climate risk assessment method described in the above embodiments.

[0022] Thirdly, embodiments of this application provide a computer-readable storage medium including computer program instructions, which, when executed by a processor, implement the method for assessing the climate risk of spring low-temperature disasters for winter wheat described in the above embodiments.

[0023] (III) Beneficial Effects

[0024] This application provides a method for assessing the climate risk of spring low-temperature disasters in winter wheat. It employs a time-scale index for spring low-temperature disasters in winter wheat, enabling the identification of the hourly disaster occurrence time and the assessment of the disaster level, thus improving the accuracy of climate risk assessment. Furthermore, this application determines the probability of spring low-temperature disasters in winter wheat by different developmental stages and intensities, and constructs a climate risk index for spring low-temperature disasters in winter wheat based on the average yield reduction rate, the coefficient of variation of yield reduction, and the probability of low-temperature disaster occurrence. Using the quantile method, the climate risk level of spring low-temperature disasters in winter wheat is divided into five levels: low risk, low-medium risk, medium risk, medium-high risk, and high risk. This further improves the accuracy of quantitative assessment of the climate risk of spring low-temperature disasters in winter wheat, providing scientific support for disaster prevention and mitigation and safe production in my country's agriculture, and ensuring food security. Attached Figure Description

[0025] Figure 1 This is a flowchart illustrating a method for assessing the climate risk of spring low-temperature disasters for winter wheat according to an embodiment of this application;

[0026] Figure 2 This is a distribution map of the average yield reduction rate of winter wheat in various regions of the Huang-Huai-Hai Plain from 1981 to 2020, according to an embodiment of this application.

[0027] Figure 3 This is a distribution map of the coefficient of variation of winter wheat yield reduction rate in various regions of the Huang-Huai-Hai Plain from 1981 to 2020, according to an embodiment of this application.

[0028] Figure 4 A distribution map of the climate risk index for low-temperature disasters of winter wheat in the Huang-Huai-Hai region from 1981 to 2020, according to an embodiment of this application;

[0029] Figure 5 This is a climate risk distribution map of winter wheat low-temperature disasters in the Huang-Huai-Hai region from 1981 to 2020, according to an embodiment of this application. Detailed Implementation

[0030] To better explain and facilitate understanding of this application, the following detailed description of the application is provided in conjunction with the accompanying drawings and specific embodiments.

[0031] Winter wheat, as one of the most important grain crops in northern my country, has a yield stability that directly affects national food security and agricultural economic development. Spring is a critical developmental stage for winter wheat as it transitions from vegetative to reproductive growth. The greening, jointing, booting, and heading-flowering stages are extremely sensitive to low-temperature stress. If sustained low temperatures or sudden, severe temperature drops occur during this time, it can easily lead to frost damage to wheat cells and physiological metabolic disorders, resulting in reduced tillering and a decrease in the number of grains per ear. In severe cases, this can cause significant yield reductions, making it one of the main meteorological disasters restricting high and stable yields of winter wheat.

[0032] While some progress has been made in climate risk assessment research on spring low-temperature disasters for winter wheat, several technical bottlenecks remain. For example, traditional methods for dividing the developmental stages of winter wheat often rely on fixed phenological periods or empirical dates, neglecting the dynamic regulatory effects of multiple factors such as variety characteristics, soil moisture, and accumulated temperature on the developmental process. This leads to biases in judging the start and end times of the developmental stages, directly affecting the accurate identification of the disaster-causing periods of low-temperature disasters. Furthermore, in characterizing the intensity of low-temperature disasters, most methods rely solely on static indicators such as daily minimum temperatures, failing to effectively quantify the cumulative stress effect of "hourly minimum temperature - accumulated chilling" during sustained low temperatures. This makes it difficult to comprehensively reflect the actual degree of harm caused by low temperatures to wheat at different developmental stages. At the same time, existing risk assessment models often lack a systematic integration of the probability of disaster occurrence, the degree of yield loss, and the variability of loss. They fail to fully incorporate the differences in crop variety resistance for refined risk assessment, resulting in discrepancies between the assessment results and actual production needs, making it difficult to effectively support agricultural disaster prevention and mitigation decisions.

[0033] Therefore, this application provides a method for assessing the climate risk of winter wheat spring low-temperature disasters. It employs a time-scale winter wheat spring low-temperature disaster index to identify the hourly disaster occurrence time and disaster severity level of the low-temperature disaster process. Furthermore, this application determines the probability of winter wheat spring low-temperature disasters by different developmental stages and intensities, and constructs a winter wheat spring low-temperature disaster climate risk index based on the average yield reduction rate, yield reduction variation coefficient, and the probability of low-temperature disaster occurrence. Using the quantile method, the climate risk level of winter wheat spring low-temperature disasters is divided into five levels: low risk, low-medium risk, medium risk, medium-high risk, and high risk. This further improves the accuracy of the quantitative assessment of winter wheat spring low-temperature disaster climate risk, providing scientific support for disaster prevention and mitigation and safe production in my country's agriculture, and ensuring food security.

[0034] To better understand the above technical solutions, exemplary embodiments of this application will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this application can be understood more clearly and thoroughly, and that the scope of this application can be fully conveyed to those skilled in the art.

[0035] Figure 1 This is a flowchart illustrating a method for assessing the climate risk of spring low-temperature disasters for winter wheat according to an embodiment of this application. Figure 1 As shown, the climate risk assessment method for spring low-temperature disasters in winter wheat includes:

[0036] Step S1: Obtain multi-source data for the target time period within the target area; the multi-source data includes meteorological data, crop data, winter wheat planting distribution data, soil moisture data, and vegetation growth data.

[0037] In the specific implementation process, meteorological data can come from the daily and hourly meteorological observation data of the China Meteorological Administration's meteorological big data cloud platform "Tianqing", including parameters such as precipitation, maximum temperature, average temperature and minimum temperature, wind speed, relative humidity, and sunshine duration.

[0038] Crop data can be obtained from agricultural meteorological observation data of the China Meteorological Administration's meteorological big data cloud platform "Tianqing", including agricultural meteorological observation data such as winter wheat development period, yield, and production management.

[0039] Winter wheat planting distribution data can be obtained by extracting remote sensing data, such as using the multi-temporal phase and high-resolution remote sensing data products built into the Google Earth Engine platform to extract winter wheat growth and development information and obtain the winter wheat planting distribution area.

[0040] Soil moisture data can be collected in real time through soil moisture monitoring stations and automatic soil moisture observation stations deployed by the China Meteorological Administration and local meteorological bureaus.

[0041] Vegetation growth data can be obtained through remote sensing monitoring and ground observation. Remote sensing monitoring data can be downloaded from NASAEarthData, while ground observation data is collected from field observation stations of relevant domestic departments.

[0042] After obtaining the above multi-source data, a database of winter wheat spring low-temperature disasters can be constructed. Based on this database, climate correlation data for the target time period within the target area can be obtained as multi-source data for the climate risk assessment of winter wheat spring low-temperature disasters.

[0043] Step S2: Divide the spring development period of winter wheat into different developmental periods based on multi-source data, and determine the hourly minimum temperature and accumulated chilling amount for each developmental period based on meteorological data.

[0044] Optionally, in some embodiments of this application, the spring development period of winter wheat is divided into different developmental stages based on multi-source data, including:

[0045] Step S21: Based on multi-source data, extract the cumulative accumulated temperature characteristics, low temperature stress intensity index characteristics, soil moisture status characteristics, and vegetation growth status index characteristics that are dynamically related to the development process of winter wheat, and perform spatiotemporal alignment processing to obtain a time-series feature sequence.

[0046] Optionally, in one example of this application, the extraction of accumulated temperature characteristics, low-temperature stress intensity index characteristics, soil moisture status characteristics, and vegetation growth status index characteristics that are dynamically related to the development process of winter wheat specifically includes:

[0047] Cumulative accumulated temperature feature extraction: Based on the daily average temperature in meteorological data, the cumulative accumulated temperature is calculated from the end of the winter dormancy period. The cumulative accumulated temperature is the sum of the differences between the daily average temperature and the biological zero point, where the biological zero point for winter wheat is set to 3℃. The end of the winter dormancy period is defined as the point at which the daily average temperature consistently exceeds 3℃.

[0048] Low-temperature stress intensity index feature extraction: Combining hourly minimum temperature data, the low-temperature stress intensity index is calculated when the temperature is below the temperature sensitivity threshold for different developmental stages of winter wheat. For example, the temperature sensitivity threshold during the greening stage is 2℃. When the hourly minimum temperature is below 2℃, the low-temperature stress intensity index for that hour is (2 - hourly minimum temperature) / 2. The higher the index, the stronger the stress. This is used to generate an hourly time series sequence of low-temperature stress intensity.

[0049] Soil moisture status feature extraction: The volumetric water content of the 0-20cm soil layer was extracted from the soil moisture data, and the daily average was taken to form daily-scale soil moisture time-series data. At the same time, according to the soil moisture level classification standard (e.g., water content <12% is drought, 12%~22% is suitable, >22% is excessively wet), it was converted into a moisture status code (drought=1, suitable=2, excessively wet=3) to enhance the identifiability of the features.

[0050] Vegetation growth status indicator feature extraction: Based on NDVI (Normalized Difference Vegetation Index) and LAI (Leaf Area Index) in the vegetation growth data, the average value for every 10 days was obtained to form a time series sequence of vegetation growth status within the target time period. Among them, NDVI reflects the intensity of vegetation photosynthesis, and LAI reflects the degree of vegetation leaf coverage. Both serve as important indicators of winter wheat growth status.

[0051] After obtaining the above features, to address the issue of differences in temporal resolution among different features, all types of features are uniformly converted into time-scale time-series features. The specific processing method is as follows:

[0052] The low-temperature stress intensity index is already hourly data and requires no additional conversion; only standardization is needed. The original values ​​of the low-temperature stress intensity index for each hour are retained to form a continuous hourly time series, ensuring that there is a corresponding index record for each hour, laying the foundation for subsequent fusion with other features.

[0053] The cumulative accumulated temperature is first calculated by evenly distributing the daily cumulative accumulated temperature increment (the difference between the current day's and the previous day's) over the 24 hours of the day, resulting in the hourly accumulated temperature increment. Then, it is accumulated hourly, meaning the hourly accumulated temperature is the sum of the previous hour's accumulated temperature and the current hour's increment. For example, if the daily cumulative accumulated temperature increment is 10℃·d, then the hourly accumulated temperature increment is approximately 10 / 24 ≈ 0.417℃·h. This hourly accumulation forms an hourly cumulative accumulated temperature time series.

[0054] Soil moisture data, consisting of daily volumetric water content and moisture status codes, is converted to hourly data using a constant value filling method. The daily average soil volumetric water content and moisture status code are assigned to the 24 hours of that day. That is, the soil volumetric water content for each hour equals the daily average, and the moisture status code is consistent with the daily value. For example, if the average soil volumetric water content for a certain day is 18% and the moisture status code is 2 (suitable), then the soil volumetric water content for all 24 hours of that day will be recorded as 18%, and the moisture status code will be 2 for all 24 hours.

[0055] The NDVI and LAI indices for vegetation growth status are 10-day averages, converted to hourly data using linear interpolation. First, the difference between NDVI and LAI for two adjacent 10-day nodes is calculated. Then, this difference is evenly distributed across the hours between these two nodes (240 hours in total over 10 days), yielding the hourly increments of NDVI and LAI. Starting with the NDVI and LAI values ​​of the first 10-day node, the increments are accumulated hourly to obtain the hourly NDVI and LAI values. For example, the average NDVI for days 1-10 is 0.3, and the average NDVI for days 11-20 is 0.5, with a difference of 0.2. The hourly NDVI increment is 0.2 / 240 ≈ 0.00083. Starting from the first hour of day 1, this increment is accumulated hourly to form an hourly NDVI time series. The conversion method for LAI is the same as for NDVI.

[0056] Based on winter wheat planting distribution data, the converted hourly accumulated temperature, hourly low-temperature stress intensity index, hourly soil moisture status, and hourly vegetation growth status indicators are spatially mapped to each grid cell in the target area. This ensures that each grid cell has corresponding accumulated temperature, low-temperature stress intensity index, soil volumetric water content, moisture status coding, NDVI, and LAI data for each hour, ultimately forming an hourly temporal feature sequence containing the above four types of features, achieving precise spatiotemporal alignment.

[0057] Step S22: Input the time-series feature sequence into the pre-trained dynamic prediction model for winter wheat development period to obtain the prediction results of the dynamic start and end times of each development period.

[0058] The dynamic prediction model for winter wheat development is based on a multimodal long short-term memory (LSTM) neural network, which includes:

[0059] The multi-channel feature embedding layer is used to normalize and embed key features in the input temporal feature sequence to generate feature vectors with temporal correlation.

[0060] Specifically, key features such as accumulated temperature, low-temperature stress intensity index, soil moisture status, and vegetation growth status indicators are first normalized (the feature values ​​are scaled to the range of 0-1) to eliminate the influence of differences in the dimensions of different features. Then, through embedded encoding, each feature is converted into a 64-dimensional feature vector, making the features more consistent with the input requirements of the neural network while preserving the temporal correlation of the features.

[0061] The spatiotemporal feature fusion layer is used to extract long-term temporal dependencies between feature vectors using a bidirectional LSTM layer, and to capture the dynamic interaction effect between the abrupt change point of the low temperature stress intensity index and the accumulated temperature trend, so as to obtain the temporal feature sequence after fusing spatiotemporal features.

[0062] In the specific implementation process, a bidirectional LSTM (Long Short-Term Memory) layer is used to process the feature vectors. The forward LSTM extracts features from the initial time step backward, while the backward LSTM extracts features from the last time step forward, effectively capturing long-term temporal dependencies between feature vectors. For example, it can identify the temporal pattern of a significant increase in vegetation growth status indicators after accumulated temperature reaches a certain level. At the same time, this layer uses a special gating mechanism to focus on capturing the dynamic interaction effect between abrupt changes in the low-temperature stress intensity index (such as a sudden increase of more than 50% in the low-temperature stress index on a single day) and the trend of accumulated temperature. When a low-temperature abrupt change occurs, the contribution weight of accumulated temperature to the advancement of the development period is dynamically adjusted, thereby obtaining a temporal feature sequence after fusing spatiotemporal features.

[0063] The stress response attention layer is used to introduce variety-specific regulatory factors to weight and regulate the temporal feature sequence after fusing spatiotemporal features based on the variety resistance coefficient and temperature sensitivity threshold of key development stages in crop data, so as to obtain the attention-weighted temporal feature sequence.

[0064] In the stress response attention layer, variety-specific regulatory factors are introduced based on the variety resistance coefficients (e.g., 0.8 for variety A and 0.6 for variety B) and temperature sensitivity thresholds for key developmental stages (2℃ for greening, 5℃ for jointing, 8℃ for booting, and 10℃ for heading-flowering) from the crop data. For the temporal feature sequence after integrating spatiotemporal features, when features involve low-temperature stress, the feature values ​​are weighted according to the variety resistance coefficient; the higher the resistance coefficient, the lower the weight of the low-temperature stress feature. Simultaneously, feature points close to the temperature sensitivity threshold are assigned higher weights, resulting in an attention-weighted temporal feature sequence that allows the model to focus more on key features that significantly influence developmental stage transitions.

[0065] The developmental transition decision layer is used to output the start probability distribution of each developmental period using the fully connected layer and the temporal Softmax layer, and to determine the dynamic start and end time prediction results of each developmental period based on the start probability distribution.

[0066] The developmental stage transition decision layer uses a fully connected layer to map the attention-weighted temporal feature sequence to a high-dimensional feature space, and then outputs the start probability distribution of each developmental stage through a temporal softmax layer. For example, on a certain date, it outputs the probability that the date is the start date of the greening stage, the probability that the jointing stage is the start date, etc. A probability threshold is set (e.g., start probability > 0.5). When the start probability of a certain developmental stage exceeds the threshold for three consecutive days, that date is determined as the dynamic start time of that developmental stage; when the start probability of a certain developmental stage is less than 0.1, and the start probability of the next developmental stage exceeds 0.5, it is determined as the dynamic end time of that developmental stage. Finally, the dynamic start and end time prediction results of each developmental stage are obtained.

[0067] Step S23: Based on the predicted dynamic start and end times of each developmental stage, the spring development period of winter wheat is divided into the greening stage, jointing stage, booting stage, and heading-flowering stage.

[0068] The above approach, by accurately extracting multi-source features, unifying the hourly time scale and achieving spatiotemporal alignment, combined with the synergistic effect of each layer of the multimodal long short-term memory neural network, not only ensures the comprehensiveness of feature extraction and data consistency, but also improves the accuracy of developmental period prediction through dynamic weight adjustment and variety-specific regulation. Ultimately, it achieves the scientific division of the spring development period of winter wheat, laying a precise temporal foundation for subsequent low-temperature disaster assessment.

[0069] Furthermore, in step S2, the minimum hourly temperature and accumulated cooling amount for different developmental stages are determined based on meteorological data, specifically including:

[0070] Determine the hourly minimum temperature and the duration of hourly low temperatures based on meteorological data;

[0071] The amount of accumulated cold during the process is determined based on the hourly minimum temperature and the duration of the hourly low temperature.

[0072] The process cooling accumulation is determined according to the following formula:

[0073]

[0074] Among them, X H T represents the accumulated cooling capacity during the process. min For the hourly minimum temperature, T c H represents the lower limit temperature for growth at different developmental stages of winter wheat, and H represents the duration of low temperature in hours.

[0075] In the specific implementation process, the lower limit temperatures for growth during the greening stage, jointing stage, booting stage, and heading-flowering stage can be 0℃, 3℃, 7℃, and 9℃, respectively.

[0076] Step S3: Determine the degree of damage to winter wheat at different developmental stages based on the minimum hourly temperature and accumulated chilling during different developmental stages, and determine the probability of low-temperature disasters occurring in winter wheat during spring based on the degree of damage to winter wheat at different developmental stages.

[0077] Optionally, in some embodiments of this application, determining the degree of damage to winter wheat at different developmental stages based on the minimum hourly temperature and accumulated chilling during different developmental periods includes:

[0078] The number of disasters suffered by winter wheat at different developmental stages was obtained based on the lowest temperature and the accumulated chill during the process.

[0079] For a complete disaster event, if, within a certain developmental period, the hourly disaster status record is not zero for ≥24 consecutive hours (i.e., low-temperature stress exists), and there is no consecutive "no disaster" interval of more than 12 hours during this period (a continuous 12-hour disaster status of 0 is considered a disaster interruption), then it is determined to be a complete disaster event. For example, if there are continuous disaster records from 00:00 on March 5th to 12:00 on March 7th during the greening period, and there is no consecutive 12-hour no-disaster interval, it is considered a single disaster event.

[0080] If the low-temperature process spans two developmental periods (e.g., from 18:00 on March 20 to 6:00 on March 21, covering the end of the greening period and the beginning of the jointing period), the statistics are based on the "disaster subject attribution principle": if more than 70% of the total duration of the disaster occurs in the previous developmental period, it is counted in the previous developmental period; if more than 70% occurs in the subsequent developmental period, it is counted in the subsequent developmental period; if the proportion is ≤50% for both periods, it is counted as one mild disaster in each of the two developmental periods (each is converted into one mild disaster based on 50% of the duration).

[0081] Specifically, the degree of damage to winter wheat at different developmental stages was determined based on the hourly minimum temperature and accumulated chilling during different developmental periods, as shown in the table below:

[0082] Table 1. Indicators of different severity levels of spring low-temperature disasters at various developmental stages of winter wheat.

[0083]

[0084] Optionally, in one specific embodiment of this application, if at least one of the minimum hourly temperature and the accumulated cooling during the process is in the severely affected range of the corresponding developmental period, the degree of damage to winter wheat is determined to be severe.

[0085] Given that both the minimum temperature and the accumulated chilling amount are within the moderately affected range for the corresponding developmental stage, the degree of damage to winter wheat is determined to be moderate.

[0086] When both the minimum temperature and the accumulated chilling amount are within the mildly affected range of the corresponding developmental stage, the degree of damage to winter wheat is determined to be mild.

[0087] If either the minimum temperature or the accumulated chilling during the process falls within the mildly affected range of the corresponding developmental stage, and the other falls within the moderately affected range of the corresponding developmental stage, the degree of damage to winter wheat is determined to be mild.

[0088] The above steps, by defining threshold ranges for mild, moderate, and severe damage based on the minimum temperature and accumulated chilling during the developmental stage, and establishing a "dual-indicator combination judgment rule"—that is, at least one severe indicator qualifies as severe, all moderate indicators as moderate, and all mild indicators or one mild and one moderate indicator as mild—quantify and standardize the assessment of damage severity. This approach fully considers the differences in the sensitivity of winter wheat to low temperatures at different developmental stages, and avoids the one-sidedness of single-indicator assessments through the synergistic judgment of temperature and accumulated chilling, ensuring the scientific and targeted nature of the damage severity classification. Simultaneously, the clear interval standards and judgment logic reduce subjective errors, making the damage severity assessment results consistent and comparable across different developmental stages, and providing a reliable grading basis for accurately determining the probability of spring low-temperature disasters in winter wheat.

[0089] Furthermore, in a specific embodiment of this application, determining the probability of spring low-temperature disasters occurring in winter wheat based on the degree of damage suffered by winter wheat at different developmental stages includes:

[0090] The degree of damage to winter wheat was determined based on the extent of damage at different developmental stages, with the number of times it was lightly damaged, moderately damaged, and severely damaged being recorded.

[0091] The probability of spring low-temperature disasters occurring in winter wheat is determined based on the number of times of mild, moderate, and severe disasters.

[0092] Specifically, the calculation of the probability of spring low-temperature disasters occurring in winter wheat is based on the following formula:

[0093]

[0094] Wherein, F represents the probability of low-temperature disasters occurring in winter wheat in spring; P1 represents the number of times mild low-temperature disasters occur in winter wheat in spring, and the indicators of mild low-temperature disasters at different developmental stages of winter wheat are shown in Table 1; W1 represents the influence coefficient of mild low-temperature disasters on the growth and development of winter wheat, with a value of 0.1; P2 represents the number of times moderate low-temperature disasters occur in winter wheat in spring, and the indicators of moderate low-temperature disasters at different developmental stages of winter wheat are shown in Table 1; W2 represents the influence coefficient of moderate low-temperature disasters on the growth and development of winter wheat, with a value of 0.3; P3 represents the number of times severe low-temperature disasters occur in winter wheat in spring, and the indicators of severe low-temperature disasters at different developmental stages of winter wheat are shown in Table 1; W3 represents the influence coefficient of severe low-temperature disasters on the growth and development of winter wheat, with a value of 0.6; S represents the total number of samples in the entire target time interval.

[0095] This step first accurately counts the number of mild, severe, and extreme disasters according to different developmental stages, and then introduces differentiated impact coefficients to construct a weighted probability formula, thus achieving a quantitative calculation of the probability of disaster occurrence. By counting the number of disasters at different developmental stages, it preserves the detailed characteristics of disasters at different stages, and by assigning higher weights to severe disasters through coefficients, it highlights their crucial impact on yield, avoiding the underestimation of risk caused by simple averaging. Simultaneously, the calculation logic based on the sample size of the entire target time interval, combined with clear indicator correspondences, ensures the objectivity and comparability of the probability results. This allows the final probability of spring low-temperature disasters to more accurately reflect the level of disaster risk, providing a scientifically quantitative probabilistic basis for subsequent risk index construction and classification.

[0096] Step S4: Determine the average yield reduction rate and the coefficient of variation of the winter wheat yield reduction rate based on crop data, and construct the winter wheat spring low temperature disaster climate risk index based on the probability of occurrence of winter wheat spring low temperature disaster, the average yield reduction rate of winter wheat, and the coefficient of variation of the winter wheat yield reduction rate.

[0097] Among them, the average yield reduction rate and the coefficient of variation of the winter wheat yield reduction rate were determined based on crop data, including:

[0098] Step S41: Obtain the actual yield and maximum possible yield of the target area based on crop data;

[0099] Step S42: Determine the trend output based on the maximum possible output, determine the meteorological output based on the actual output and the trend output, and determine the relative meteorological output based on the trend output and the meteorological output.

[0100] Step S43: Determine the average yield reduction rate of winter wheat based on the relative meteorological yield, and determine the coefficient of variation of the winter wheat yield reduction rate based on the relative meteorological yield and the average yield reduction rate of winter wheat.

[0101] In the specific implementation process, the trend output can be calculated using the following formula:

[0102]

[0103] Among them, Y t Let K be the trend output, K be the maximum possible output, a and b be the regression coefficients, and t be the target time series number.

[0104] Meteorological yield can be calculated using the following formula:

[0105] Y m =YY t (4)

[0106] Among them, Y m Y represents meteorological yield, and Y represents actual yield. t This represents the trend output.

[0107] Based on the above trend yield and meteorological yield, calculate the relative meteorological yield:

[0108]

[0109] Y r For relative meteorological yield, Y m For meteorological yield, Y t This represents the trend output.

[0110] Based on relative meteorological yields, the average yield reduction rate for winter wheat was calculated as follows:

[0111]

[0112] D represents the average yield reduction rate of winter wheat, Yr represents the relative meteorological yield, and N represents the total number of samples whose relative meteorological yield sequence is less than 0.

[0113] Furthermore, the coefficient of variation of winter wheat yield reduction rate can be calculated based on the following formula:

[0114]

[0115] Where V is the coefficient of variation of winter wheat yield reduction rate, Yr is the relative meteorological yield, D is the average yield reduction rate of winter wheat, and N is the total number of samples with a relative meteorological yield sequence less than 0.

[0116] Therefore, based on the above-mentioned probability of winter wheat spring low-temperature disasters, average yield reduction rate of winter wheat, and coefficient of variation of winter wheat yield reduction rate, the climate risk index for winter wheat spring low-temperature disasters is obtained as follows:

[0117] C = F × D × V (8)

[0118] Wherein, C is the climate risk index for low-temperature disasters in spring for winter wheat; F is the probability of low-temperature disasters in spring for winter wheat; D is the average yield reduction rate of winter wheat; and V is the coefficient of variation of the yield reduction rate of winter wheat.

[0119] Step S5: Classify the risk level based on the winter wheat spring low temperature disaster climate risk index.

[0120] Specifically, based on the climate risk index for spring low-temperature disasters in winter wheat, the quantile method can be used to classify the climate risk level of spring low-temperature disasters in winter wheat into five levels: low risk, low-to-medium risk, medium risk, medium-to-high risk, and high risk. The classification criteria are shown in the table below:

[0121] Table 2. Climate Risk Levels for Low Temperature Disasters in Spring for Winter Wheat

[0122]

[0123]

[0124] In summary, this embodiment of the method for assessing the climate risk of low-temperature disasters in winter wheat during spring adopts a time-scale index for low-temperature disasters in winter wheat during spring, enabling the identification of the hourly disaster occurrence time and the assessment of the disaster level. It determines the probability of occurrence of low-temperature disasters in winter wheat during spring by time period (greening stage, jointing stage, booting stage, and heading-flowering stage) and by intensity (mild, moderate, and severe). By employing yield separation technology to obtain trend yield and meteorological yield, a climate risk index for low-temperature disasters in winter wheat during spring is constructed based on the average yield reduction rate, yield reduction coefficient of variation, and the probability of low-temperature disaster occurrence. The quantile method is used to classify the climate risk level of low-temperature disasters in winter wheat during spring into five levels: low risk, low-medium risk, medium risk, medium-high risk, and high risk. This further improves the accuracy of the quantitative assessment of the climate risk of low-temperature disasters in winter wheat during spring and can provide scientific support for disaster prevention, mitigation, and safe production in my country's agriculture.

[0125] To better understand the implementation scheme described in this application, the following detailed explanation uses the Huang-Huai-Hai region of China as the target area and the period from 1981 to 2020 as the target time interval, combined with specific steps.

[0126] It should be noted that the Huang-Huai-Hai region of China in this embodiment includes seven provinces and municipalities: Beijing, Tianjin, Shandong, Hebei, Henan, Anhui, and Jiangsu, which belong to the warm continental monsoon climate zone (29°39′–42°62′N, 110°36′–122°75′E).

[0127] This embodiment of a method for assessing the climate risk of low-temperature disasters in winter wheat during spring includes the following steps:

[0128] Step A1: Determine the target area and target time, and obtain multi-source data for the target time period; multi-source data includes meteorological data, crop data, winter wheat planting distribution data, soil moisture data, and vegetation growth data;

[0129] The target region is the winter wheat producing area of ​​the Huang-Huai-Hai Plain in China, and the target time period is 1981-2020;

[0130] The meteorological data comes from the daily and hourly meteorological observation data of the China Meteorological Administration's meteorological big data cloud platform "Tianqing". It includes the daily maximum temperature (°C), daily minimum temperature (°C), daily average temperature (°C), daily precipitation (mm) from 20:00 to 20:00, daily average wind speed (m / s), daily average relative humidity (%), daily average air pressure (hPa), sunshine duration (0.1h), and hourly meteorological observation data of 535 meteorological stations in the Huang-Huai-Hai region (Beijing, Tianjin, Hebei, Henan, Shandong, Anhui, and Jiangsu) from 1981 to 2020.

[0131] The crop data comes from agricultural meteorological observation data of the China Meteorological Administration's meteorological big data cloud platform "Tianqing", including agricultural meteorological observation data such as winter wheat development period, yield, and production management; specifically, the yield data includes winter wheat yield in Beijing from 2000 to 2022, winter wheat yield in Tianjin, Hebei, Jiangsu, and Anhui from 1978 to 2022, and winter wheat yield in Shandong and Henan from 1949 to 2022.

[0132] The winter wheat planting distribution data was obtained by extracting data based on remote sensing data. Using the multi-temporal phase, high-resolution remote sensing data product provided by the Google Earth Engine platform (https: / / code.earthengine.google.com), the data was extracted from the bottom-of-atmosphere corrected reflectance data of the 10m resolution reflectance image of Sentinel-2L2A from 2020 to 2021. The growth and development information of winter wheat was extracted, and the planting distribution area of ​​winter wheat was obtained.

[0133] Furthermore, the multi-source data in this embodiment also includes disaster data, which comes from the "Meteorological Disaster Management System" developed by the China Meteorological Administration, the Disaster Encyclopedia, and the "China Statistical Yearbook". It contains a large amount of meteorological disaster event data across the country, and monitors and assesses meteorological disaster events in real time.

[0134] Step A2: Divide the spring development period of winter wheat into different developmental periods based on the multi-source data, and determine the hourly minimum temperature and accumulated chilling amount for each of the different developmental periods based on the meteorological data.

[0135] Step A3: Determine the degree of damage to winter wheat at different developmental stages based on the minimum hourly temperature and accumulated chilling during the process, and determine the probability of low-temperature disasters occurring in winter wheat during spring based on the degree of damage to winter wheat at different developmental stages.

[0136] Specifically, based on the aforementioned multi-source data, the number of mild, moderate, and severe low-temperature disasters occurring in winter wheat at 535 meteorological stations in the Huang-Huai-Hai region from 1981 to 2020 was calculated, and the probability of low-temperature disasters occurring throughout the entire spring development period of winter wheat was determined. The results show that Hebei, Anhui, and Shandong provinces have a relatively high probability of low-temperature disasters occurring in winter wheat. Specifically, the highest probability of low-temperature disasters occurring in winter wheat is 7.44 in Hebei Province, 4.68 in Anhui Province, 4.60 in Shandong Province, 2.51 in Beijing, 1.00 in Henan Province, 0.71 in Jiangsu Province, and 0.52 in Tianjin.

[0137] Step A4: Determine the average yield reduction rate and the coefficient of variation of the winter wheat yield reduction rate based on the crop data, and construct the winter wheat spring low temperature disaster climate risk index based on the probability of occurrence of winter wheat spring low temperature disaster, the average yield reduction rate of winter wheat, and the coefficient of variation of the winter wheat yield reduction rate.

[0138] The average yield reduction rate and the coefficient of variation of the winter wheat yield reduction rate were determined based on crop data, including:

[0139] Step A41: Obtain the actual yield and maximum possible yield of the target area based on crop data;

[0140] Step A42: Determine the trend output based on the maximum possible output, determine the meteorological output based on the actual output and the trend output, and determine the relative meteorological output based on the trend output and the meteorological output.

[0141] Step A43: Determine the average yield reduction rate of winter wheat based on the relative meteorological yield, and determine the coefficient of variation of the winter wheat yield reduction rate based on the relative meteorological yield and the average yield reduction rate of winter wheat.

[0142] Specifically, analysis of the meteorological yield of winter wheat in various provinces and cities in the Huang-Huai-Hai region reveals that Beijing's highest meteorological yield of winter wheat was in 2000 (17,500 tons / 10,000 mu), and the lowest was in 2003 (-10,300 tons / 10,000 mu); Tianjin's highest meteorological yield of winter wheat was in 1979 (1,000 tons / 10,000 mu), and the lowest was in 1982 (-1,100 tons / 10,000 mu); Hebei's highest meteorological yield of winter wheat was in 1997 (600 tons / 10,000 mu), and the lowest was in 1980 (-500 tons / 10,000 mu); Jiangsu's highest meteorological yield of winter wheat was in 1999 (2,100 tons / 10,000 mu), and the lowest was in 2003 (-1,000 tons / 10,000 mu). The highest meteorological yield of winter wheat in Anhui Province was in 1997 (0.07 million tons / 10,000 mu), and the lowest meteorological yield was in 1991 (-0.09 million tons / 10,000 mu). The highest meteorological yield of winter wheat in Shandong Province was in 1997 (0.05 million tons / 10,000 mu), and the lowest was in 2002 (-0.05 million tons / 10,000 mu). The highest meteorological yield of winter wheat in Henan Province was in 1997 (0.04 million tons / 10,000 mu), and the lowest was in 2002 (-0.02 million tons / 10,000 mu).

[0143] The average yield reduction rate of winter wheat in various regions of the Huang-Huai-Hai Plain was calculated based on trend yield and meteorological yield. Figure 2 ).Depend on Figure 2 It can be seen that Beijing had the largest average yield reduction rate for winter wheat (27.19%), followed by Tianjin (15.63%), Shandong ranked third with 10.38%, and Anhui and Henan ranked fourth and fifth with yield reduction rates of 8.89% and 8.01% respectively; Jiangsu and Hebei had relatively low average yield reduction rates of 7.68% and 6.47% respectively.

[0144] The coefficient of variation of winter wheat yield reduction rate in various regions of the Huang-Huai-Hai Plain was calculated based on the average winter wheat yield reduction rate and relative meteorological yield, such as... Figure 3 As shown. By Figure 3It can be seen that Hebei Province has the highest coefficient of variation for winter wheat yield reduction at 1.27; Henan Province ranks second at 1.24, while the coefficients of variation for winter wheat yield reduction rates in Anhui, Shandong, and Jiangsu Provinces are 1.21, 0.99, and 0.93, respectively. The coefficients of variation for winter wheat yield reduction rates in Tianjin and Beijing are relatively low, at 0.83 and 0.41, respectively, still showing varying degrees of yield reduction volatility.

[0145] Based on the probability of low-temperature disasters occurring in spring for winter wheat, the average yield reduction rate, and the coefficient of variation of the yield reduction rate, the climate risk index for low-temperature disasters of winter wheat in the Huang-Huai-Hai region was calculated, and its spatial distribution map is shown below. Figure 4 As shown. From Figure 4 It can be seen that Hebei Province has the highest climate risk index for low-temperature disasters in spring for winter wheat, with a maximum value of 0.61; followed by Beijing, where the climate risk index for low-temperature disasters in spring for winter wheat is also relatively high, with a maximum value of 0.28.

[0146] Based on the climate risk index for spring low-temperature disasters in winter wheat, the climate risk level for spring low-temperature disasters in winter wheat was divided into five levels—low risk, low-to-medium risk, medium risk, medium-to-high risk, and high risk—using the quantile method. The spatial distribution is shown in the figure. Figure 5 As shown.

[0147] Table 3 shows the area distribution of winter wheat in the Huang-Huai-Hai region based on the risk level of spring low-temperature disasters. As can be seen from Table 3, the area with the largest proportion of medium-risk areas (26.41%) of winter wheat planting area is the medium-low risk area (25.20%), followed by medium-high risk areas (23.38%) and low risk areas (17.51%), while the high-risk area is the smallest (7.50%). Shandong Province is mainly at medium-high and high risk, making it the province with the highest risk of spring low-temperature disasters for winter wheat in the Huang-Huai-Hai region; followed by Henan Province, which is mainly at medium and medium-high risk.

[0148] Table 3. Percentage of area at risk of spring low-temperature disasters for winter wheat in the Huang-Huai-Hai Plain, 1981-2020

[0149]

[0150]

[0151] This embodiment of the method for assessing the climate risk of low-temperature disasters in winter wheat during spring further improves the accuracy of quantitative assessment of the climate risk of low-temperature disasters in winter wheat during spring, and can provide scientific support for disaster prevention and mitigation and safe production in my country's agriculture.

[0152] In addition, this application also proposes a computer device, which includes a processor and a memory. The processor is used to execute instructions stored in the memory so that the computer device can perform the winter wheat spring low temperature disaster climate risk assessment method described in the above embodiments.

[0153] Finally, this application also proposes a computer-readable storage medium including computer program instructions, which, when executed by a processor, implement the method for assessing the climate risk of winter wheat low-temperature disasters in spring as described in the above embodiments.

[0154] The processor can be a central processing unit, a network processor, or a combination thereof. The processor may further include hardware chips. These hardware chips can be application-specific integrated circuits (ASICs), programmable logic devices (PLDs), or combinations thereof. The programmable logic devices can be complex programmable logic devices (CLPs), field-programmable gate arrays (FPGAs), general-purpose array logic (GDAs), or any combination thereof.

[0155] The memory stores instructions executable by at least one processor to enable the at least one processor to perform the winter wheat low-temperature disaster climate risk assessment method shown in the above embodiments.

[0156] The memory may include a stored program area and a stored data area, wherein the stored program area may store the operating system and application programs required for at least one function; the stored data area may store data created based on the use of the computer device, etc. Furthermore, the memory may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory may optionally include memory remotely located relative to the processor, which can be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0157] The memory may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory may also include a combination of the above types of memory.

[0158] In summary, the climate risk assessment method for low-temperature disasters in winter wheat in this application adopts a time-scale low-temperature disaster level index based on a combination of process accumulated chilling and minimum temperature. This breaks through the limitations of traditional daily-scale meteorological element identification and achieves accurate identification of the hourly disaster occurrence time and scientific assessment of the disaster level, laying a more refined foundation for subsequent risk assessment. Furthermore, this application subdivides the spring development period of winter wheat into the greening stage, jointing stage, booting stage, and heading-flowering stage, and classifies it into mild, moderate, and severe according to the degree of disaster. It specifically determines the probability of disaster occurrence under different time periods and intensities, fully considering the sensitivity differences of winter wheat at different growth stages to low-temperature disasters, making the assessment results more consistent with the actual growth of the plant. In addition, this application extracts trend yield and meteorological yield through yield separation technology, and constructs a multi-dimensional climate risk index by combining the average yield reduction rate, the coefficient of variation of the yield reduction rate (reflecting the volatility of yield reduction), and the probability of disaster occurrence. This comprehensively considers the probability of disaster occurrence, the average level of loss, and the stability of loss, achieving a quantitative and comprehensive risk assessment.

[0159] The climate risk assessment method for low-temperature disasters in winter wheat during spring, as proposed in this application, significantly improves the accuracy and relevance of climate risk assessment for low-temperature disasters in winter wheat during spring. It can provide precise risk information for agricultural producers and operators, as well as relevant government departments, helping to enhance my country's agricultural capacity to cope with climate change, ensuring food security, and serving the implementation of rural revitalization.

[0160] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0161] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0162] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0163] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0164] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0165] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0166] In the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0167] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make modifications, alterations, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for assessing the climate risk of spring low-temperature disasters for winter wheat, characterized in that, include: Acquire multi-source data for a target time period within a target area; wherein, the multi-source data includes meteorological data, crop data, winter wheat planting distribution data, soil moisture data, and vegetation growth data; Based on the multi-source data, the spring development period of winter wheat is divided into different developmental periods, and the hourly minimum temperature and accumulated chilling amount for each developmental period are determined based on the meteorological data. Based on the aforementioned multi-source data, the spring development period of winter wheat is divided into different developmental stages, including: Based on the multi-source data, the cumulative accumulated temperature characteristics, low temperature stress intensity index characteristics, soil moisture status characteristics, and vegetation growth status index characteristics that are dynamically related to the development process of winter wheat are extracted, and spatiotemporal alignment processing is performed to obtain a time-series feature sequence. The time-series feature sequence is input into a pre-trained dynamic prediction model for the development period of winter wheat to obtain the prediction results of the dynamic start and end times of each development period. Based on the prediction results of the dynamic start and end times of each developmental stage, the spring development period of winter wheat is divided into the greening stage, jointing stage, booting stage, and heading-flowering stage. The degree of damage to winter wheat at different developmental stages is determined based on the minimum hourly temperature and accumulated chilling during the process, and the probability of low-temperature disasters occurring in spring for winter wheat is determined based on the degree of damage to winter wheat at different developmental stages. The average yield reduction rate and the coefficient of variation of the winter wheat yield reduction rate are determined based on the crop data. A climate risk index for the winter wheat spring low temperature disaster is constructed based on the probability of occurrence of the winter wheat spring low temperature disaster, the average yield reduction rate of winter wheat, and the coefficient of variation of the winter wheat yield reduction rate. Risk levels are classified based on the aforementioned winter wheat spring low-temperature disaster climate risk index.

2. The method for assessing the climate risk of spring low-temperature disasters for winter wheat according to claim 1, characterized in that, The dynamic prediction model for the developmental stage of winter wheat is obtained by training a multimodal long short-term memory neural network, wherein the multimodal long short-term memory neural network includes: A multi-channel feature embedding layer is used to normalize and embed key features in the input temporal feature sequence to generate feature vectors with temporal correlation. The spatiotemporal feature fusion layer is used to extract the long-term temporal dependencies between the feature vectors using a bidirectional LSTM layer, and to capture the dynamic interaction effect between the abrupt change point of the low temperature stress intensity index and the accumulated temperature trend, so as to obtain the temporal feature sequence after fusing spatiotemporal features. The stress response attention layer is used to introduce variety-specific adjustment factors to weight the temporal feature sequence after the fusion of spatiotemporal features based on the variety resistance coefficient and temperature sensitivity threshold of key development period in the crop data, so as to obtain the attention-weighted temporal feature sequence. The developmental stage transition decision layer is used to output the start probability distribution of each developmental stage using the fully connected layer and the temporal Softmax layer, and to determine the dynamic start and end time prediction results of each developmental stage based on the start probability distribution.

3. The method for assessing the climate risk of spring low-temperature disasters for winter wheat according to any one of claims 1-2, characterized in that, Based on the meteorological data, the hourly minimum temperature and accumulated cooling amount for the different developmental stages are determined, including: The hourly minimum temperature and the duration of the hourly low temperature were determined based on the meteorological data. The amount of cold accumulated during the process is determined based on the hourly minimum temperature and the duration of the hourly low temperature.

4. The method for assessing the climate risk of spring low-temperature disasters for winter wheat according to claim 3, characterized in that, The amount of cold accumulated during the process is determined according to the following formula: ； Among them, X H T represents the accumulated cooling capacity of the process. min For the hourly minimum temperature, T c H represents the lower limit temperature for growth at different developmental stages of winter wheat, and H represents the duration of low temperature in hours.

5. The method for assessing the climate risk of spring low-temperature disasters for winter wheat according to claim 3, characterized in that, The degree of damage to winter wheat at different developmental stages is determined based on the hourly minimum temperature and accumulated chilling during those stages, including: The number of disasters suffered by winter wheat at different developmental stages was obtained based on the minimum temperature and accumulated cooling during the process. The degree of damage for each disaster was obtained based on the number of times the winter wheat was affected at different developmental stages. If at least one of the minimum temperature and the accumulated cooling during the process falls within the severely affected range of the corresponding developmental period, the degree of damage to the winter wheat is determined to be severe. When both the minimum temperature and the accumulated chilling during the process are within the moderately affected range of the corresponding developmental period, the degree of damage to the winter wheat is determined to be moderate. When both the minimum temperature and the accumulated chilling during the process are within the mildly affected range of the corresponding developmental period, the degree of damage to the winter wheat is determined to be mild. If either the minimum temperature or the accumulated cooling during the process falls within the mildly affected range of the corresponding developmental stage, and the other falls within the moderately affected range of the corresponding developmental stage, then the degree of damage to the winter wheat is determined to be mild.

6. The method for assessing the climate risk of spring low-temperature disasters for winter wheat according to claim 5, characterized in that, The probability of spring low-temperature disasters occurring in winter wheat is determined based on the degree of damage suffered by winter wheat at different developmental stages, including: The degree of damage to winter wheat at different developmental stages was determined by the number of times it was lightly damaged, moderately damaged, and severely damaged. The probability of winter wheat experiencing spring low-temperature disasters is determined based on the number of times the disaster was mild, moderate, and severe.

7. The method for assessing the climate risk of spring low-temperature disasters for winter wheat according to claim 1, characterized in that, The average yield reduction rate and the coefficient of variation of the winter wheat yield reduction rate were determined based on the crop data, including: Based on the crop data, obtain the actual yield and maximum possible yield of the target area; The trend output is determined based on the maximum possible output, the meteorological output is determined based on the actual output and the trend output, and the relative meteorological output is determined based on the trend output and the meteorological output. The average yield reduction rate of winter wheat is determined based on the relative meteorological yield, and the coefficient of variation of the winter wheat yield reduction rate is determined based on the relative meteorological yield and the average yield reduction rate of winter wheat.

8. A computer device, characterized in that, include: The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform the method for assessing the climate risk of spring low-temperature disasters for winter wheat as described in any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the method for assessing the climate risk of winter wheat low-temperature disasters in spring, as described in any one of claims 1-8.

Images