Temperature forecast error evaluation method under extreme weather condition
By identifying and quantifying the core influencing factors of extreme weather temperature forecasts, a global error assessment and bias attribution model is constructed, solving the problem that existing technologies cannot deeply analyze the sources of error, and realizing accurate and efficient temperature forecasts under extreme weather conditions.
Patent Information
- Application Number
- CN202511255517.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-04
- Publication Date
- 2025-12-16
AI Technical Summary
Existing methods for assessing temperature forecast errors lack in-depth analysis of error sources under extreme weather conditions, fail to accurately reflect the impact of complex factors on forecast errors, and are difficult to effectively guide the improvement and optimization of forecast models.
By identifying the core influencing factors of extreme weather temperature forecasts, quantifying the dimensions and corresponding quantitative indicators of each factor, constructing a global error assessment model and optimizing its solution, a global error assessment result is obtained. Based on this, a deviation attribution model is constructed and optimized to achieve error feedback.
It improves the accuracy and reliability of temperature forecasts, provides scientific error assessment methods, and can provide scientific basis for meteorological agencies and related departments, reduce economic losses and social impacts, and enhance the adaptability of meteorological services and public trust.
Smart Images
Figure CN121145170A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the technical field of forecast error analysis, and in particular to a method for assessing temperature forecast errors under extreme weather conditions. Background Technology
[0002] With the intensification of global climate change, the frequency, intensity, and impact range of extreme weather events such as heat waves, severe cold waves, and extreme low / high temperatures have significantly increased. These extreme weather events severely disrupt socio-economic activities such as agricultural production, transportation, and energy supply. The accuracy of temperature forecasts during extreme weather events directly determines the effectiveness of various response decisions; therefore, a scientific and accurate assessment of extreme weather temperature forecast errors is necessary.
[0003] Existing methods for assessing temperature forecast errors mainly focus on simple deviations between forecast and observation values, lacking in-depth analysis and attribution of error sources. This results in an inability to accurately reflect the impact of various complex factors on forecast errors under extreme weather conditions, making it difficult to effectively guide the improvement and optimization of forecast models.
[0004] Therefore, there is an urgent need for an extreme weather temperature forecast error assessment method that can comprehensively consider multiple core influencing factors, quantify the contribution of each factor to the forecast error, and accurately attribute the deviation. Summary of the Invention
[0005] This invention provides a method for evaluating temperature forecast errors under extreme weather conditions, which can provide more accurate, scientific and efficient temperature forecasts under complex extreme weather conditions and can effectively solve the problems in the background technology.
[0006] To achieve the above objectives, in a first aspect, the present invention provides a method for evaluating temperature forecast errors under extreme weather conditions, comprising:
[0007] Identify the core influencing factors of extreme weather temperature forecasts;
[0008] The dimensions and corresponding quantitative indicators of each of the core influencing factors are quantified, and the influence weight of each of the core influencing factors on the forecast error is calculated.
[0009] A global error evaluation model is constructed and optimized to obtain the global error evaluation results;
[0010] Based on the global error assessment results, a deviation attribution model is constructed and optimized to obtain a deviation attribution scheme.
[0011] Based on the global error assessment results and the deviation attribution scheme, error feedback is provided for extreme weather temperature forecasts.
[0012] In conjunction with the first aspect, in one possible design, the core influencing factors include at least the intensity of the weather system, the characteristics of the underlying surface, and the quality of the observation data.
[0013] In conjunction with the first aspect, in one possible design, a global error evaluation model is constructed and optimized to obtain the global error evaluation results, including:
[0014] The primary optimization objective is to minimize the overall global error in extreme weather temperature forecasts.
[0015] The first constraint is the spatiotemporal steady-state equilibrium of extreme weather forecasting, which includes at least time constraints, spatial constraints, and extreme sample constraints.
[0016] The first optimization algorithm is used to optimize the first optimization objective under the first constraint, and outputs the specific values of the global comprehensive error, the error of each component, and the distribution of key evaluation indicators.
[0017] In conjunction with the first aspect, in one possible design, the global integrated error is calculated by weighting at least the extreme specific error, the systematic deviation, and the local deviation.
[0018] In conjunction with the first aspect, in one possible design, based on the global error assessment results, a deviation attribution model is constructed and optimized to obtain a deviation attribution scheme, including:
[0019] The second optimization objective is to maximize the accuracy of attribution of biases in extreme weather temperature forecasts.
[0020] The second constraint is the logical consistency of bias attribution, which includes at least causal constraints, data constraints, and spatiotemporal constraints.
[0021] The second optimization algorithm is used to optimize the second optimization objective under the second constraint, and outputs the specific sources of deviation and the distribution statistics of the sources of deviation for each high error sample.
[0022] In conjunction with the first aspect, in one possible design, the specific dimensions of the weather system intensity include the intensity of extreme events and the duration of extreme events.
[0023] In conjunction with the first aspect, in one possible design, the specific dimensions of the underlying surface characteristics include regional cover type and local effects.
[0024] In conjunction with the first aspect, in one possible design, the first optimization algorithm is a particle swarm optimization algorithm, wherein the particle swarm optimization algorithm uses the error contribution value of each extreme event sample as a particle, and the termination condition for iterative optimization is that the number of iterations reaches a preset number and the change in the global comprehensive error is less than or equal to a preset threshold.
[0025] In conjunction with the first aspect, in one possible design, the second optimization algorithm is a decision tree classification algorithm. The decision tree classification algorithm takes the core influencing factor characteristics of high error samples as input and the type of deviation source as output. The type of deviation source includes at least the model physical process deviation, the underlying surface parameter error, and the observation data error.
[0026] Secondly, the present invention also provides a temperature forecast error assessment system under extreme weather conditions, comprising:
[0027] The core influencing factor identification module identifies the core influencing factors for extreme weather temperature forecasts.
[0028] The module for quantifying and weighting influencing factors quantifies the dimensions and corresponding quantitative indicators of each core influencing factor and calculates the weight of each core influencing factor on the forecast error.
[0029] The global error assessment model construction and solution module constructs the global error assessment model and performs optimization to obtain the global error assessment results.
[0030] The deviation attribution model construction and solution module constructs a deviation attribution model and optimizes the solution based on the global error assessment results to obtain a deviation attribution scheme.
[0031] The error feedback module provides error feedback for extreme weather temperature forecasts based on the global error assessment results and the deviation attribution scheme.
[0032] The technical solution of this invention achieves the following technical effects: by identifying and integrating multiple core influencing factors, it can comprehensively assess the factors affecting temperature forecast errors, thereby improving forecast accuracy; the method not only identifies influencing factors but also quantifies and calculates their influence weights, enabling decision-makers to more intuitively understand the importance of which factors in specific situations, thus enabling effective adjustment of response strategies; constructing a global error assessment model and a deviation attribution model enables a dynamic feedback mechanism; by providing error feedback based on assessment results, this method makes the adjustment of forecast models more timely and effective, improving the reliability and adaptability of meteorological services; the comprehensive assessment method considering multiple factors can provide scientific basis for meteorological agencies, governments, and relevant departments in responding to extreme weather events; and effective forecasting... Accurate error feedback can help stakeholders develop more reasonable and efficient response measures, reducing economic losses and social impact. Multi-dimensional factor analysis makes the assessment model highly adaptable, enabling optimization and adjustment for different extreme weather conditions and providing customized forecast services. The systematic analysis and feedback mechanism makes the assessment results more scientific and reliable, thereby increasing public trust in the forecasts. Because of the comprehensive data and research results it provides, it can promote cooperation between meteorological departments and other sectors such as agriculture and transportation, forming a more efficient response mechanism. This extreme weather temperature forecast error assessment method, through comprehensive consideration of multiple factors, quantitative analysis, and optimized feedback mechanisms, can provide more accurate, scientific, and efficient temperature forecasts under complex extreme weather conditions. Attached Figure Description
[0033] Figure 1 This is a flowchart of the present invention;
[0034] Figure 2 This is a structural diagram for assessing temperature forecast errors under extreme weather conditions. Detailed Implementation
[0035] This application will now be described with reference to the accompanying drawings.
[0036] like Figure 1 As shown, the method for evaluating temperature forecast errors under extreme weather conditions according to the present invention specifically includes the following steps:
[0037] S1. Identify the core influencing factors of extreme weather temperature forecasts;
[0038] S2. Quantify the dimensions and corresponding quantitative indicators of each of the core influencing factors, and calculate the weight of each of the core influencing factors on the forecast error;
[0039] S3. Construct a global error evaluation model and find the optimal solution to obtain the global error evaluation result;
[0040] S4. Based on the global error assessment results, construct a deviation attribution model and optimize the solution to obtain a deviation attribution scheme;
[0041] S5. Based on the global error assessment results and the deviation attribution scheme, error feedback is provided for extreme weather temperature forecasts.
[0042] In this embodiment, by identifying and integrating multiple core influencing factors, the factors affecting temperature forecast errors can be comprehensively assessed, thereby improving forecast accuracy. The method not only identifies influencing factors but also quantifies them and calculates their influence weights, enabling decision-makers to more intuitively understand the importance of which factors in specific situations, thus allowing for effective adjustments to response strategies. Constructing a global error assessment model and a bias attribution model enables a dynamic feedback mechanism. By providing error feedback based on the assessment results, this method makes adjustments to the forecast model more timely and effective, improving the reliability and adaptability of meteorological services. The comprehensive assessment method, considering multiple factors, can provide a scientific basis for meteorological agencies, governments, and relevant departments in responding to extreme weather events, resulting in effective forecasts and accurate error prediction. Feedback helps stakeholders develop more reasonable and efficient response measures, reducing economic losses and social impact. Multi-dimensional factor analysis enhances the adaptability of the assessment model, enabling optimization and adjustment for different extreme weather conditions and providing customized forecast services. The systematic analysis and feedback mechanism makes the assessment results more scientific and reliable, thereby increasing public trust in the forecasts. The comprehensive data and research results it provides can promote cooperation between meteorological departments and other sectors such as agriculture and transportation, forming a more efficient response mechanism. This extreme weather temperature forecast error assessment method, through comprehensive consideration of multiple factors, quantitative analysis, and optimized feedback mechanisms, can provide more accurate, scientific, and efficient temperature forecasts under complex extreme weather conditions.
[0043] In some embodiments of the present invention, for step S1, the core influencing factors of extreme weather temperature forecast are identified;
[0044] Collect historical temperature forecast data and actual observation data under different types of extreme weather events in different regions;
[0045] At the same time, other data related to extreme weather are collected, including meteorological monitoring data, geographic information data, and equipment information and data quality records of observation stations;
[0046] Based on existing research findings and practical experience, a preliminary list of factors affecting temperature forecast errors in extreme weather has been compiled.
[0047] Using statistical methods, a correlation analysis was conducted between the initially screened list of influencing factors and temperature forecast errors. The correlation coefficient between each influencing factor and the forecast error was calculated, and factors with significant correlation to the forecast error were screened based on the magnitude and significance level of the correlation coefficient. For example, for weather system intensity, the average forecast error under different intensities of weather systems can be calculated, and the correlation between the two can be analyzed. If the forecast error shows a significant trend with the increase of weather system intensity, and the correlation coefficient reaches a significant level, it indicates that weather system intensity is an influencing factor closely related to the forecast error.
[0048] To further determine the causal relationship between the selected factors and the forecast error, rather than just the correlation, causal inference methods can be used. Taking the Granger causality test as an example, for two time series variables, we can determine whether there is a causal relationship between them by testing whether the past value of one variable has the predictive power for the current value of the other variable. If the past value of a certain influencing factor can significantly improve the prediction of the forecast error, it indicates that the factor may be one of the causes of the forecast error.
[0049] Based on the results of correlation analysis and causal relationship verification, the core influencing factors that significantly affect the temperature forecast error of extreme weather were identified. These core influencing factors include at least the intensity of the weather system, the characteristics of the underlying surface, and the quality of the observational data. The specific dimensions of the intensity of the weather system include the intensity and duration of the extreme event. The specific dimensions of the characteristics of the underlying surface include regional cover type and local effects.
[0050] In this embodiment, by collecting historical temperature forecast data and actual observation data from different regions and types, and combining multi-dimensional data such as meteorological monitoring, geographic information, and equipment information, the breadth and accuracy of the data foundation are ensured, thus improving the credibility of the research results. A preliminary list of factors affecting temperature forecast errors in extreme weather is compiled, and correlation analysis is performed using statistical methods. Significantly correlated factors are screened based on significance levels, clarifying which factors significantly affect forecast accuracy under complex extreme weather conditions, avoiding biases caused by empirical judgments. A causal inference method is used to conduct in-depth analysis of the relationship between each screened influencing factor and forecast errors, distinguishing between correlation and causation. This helps identify which factors are the true causes of forecast errors, allowing for more targeted model optimization. By identifying core influencing factors, more accurate input information can be provided to the temperature forecast model. Identifying these key factors helps meteorological agencies make more effective adjustments and predictions under extreme weather conditions, thereby improving the accuracy and reliability of forecasts.
[0051] In some embodiments of the present invention, for step S2, the dimensions and corresponding quantitative indicators of each of the core influencing factors are quantified, and the influence weight of each of the core influencing factors on the forecast error is calculated.
[0052] The dimensions and corresponding quantitative indicators for each of the core influencing factors are quantified, specifically including:
[0053] To assess the intensity of weather systems, the difference between the central pressure of an extreme weather system and the average pressure within a certain surrounding area is calculated as a quantitative indicator of the pressure gradient. The average and maximum wind speeds at different altitudes during extreme weather events are statistically analyzed. The average wind speed reflects the overall movement speed and intensity of the weather system, while the maximum wind speed reflects the extreme nature of the weather system. Temperature changes over a certain period before and after the occurrence of extreme weather events are analyzed, and the rate of temperature change is calculated.
[0054] Based on the characteristics of the underlying surface, lidar or ground measurement equipment is used to measure the height undulation of different underlying surfaces and calculate the surface roughness parameters; the vegetation cover of different areas is calculated using satellite remote sensing images and methods such as the normalized vegetation index; soil moisture is measured at different depths using soil moisture sensors, and the average value is taken as the quantitative index of soil moisture in the area.
[0055] To ensure the quality of observational data, consult the instruction manual of the observation instrument to obtain its nominal accuracy. Simultaneously, regularly calibrate and compare the instruments, correcting the instrument accuracy based on the calibration results and comparison errors, using this as a quantitative indicator of the instrument's precision. Record the time point of each observation and calculate the average interval between adjacent observation time points. The shorter the observation interval, the more details of temperature changes can be captured, resulting in higher data quality. Statistically analyze the number or proportion of missing data within a certain time period. Missing data affects the continuous monitoring and analysis of temperature changes; a higher proportion of missing data indicates poorer data quality.
[0056] Calculate the weight of each of the core influencing factors on the forecast error, specifically including:
[0057] Standardize the quantified core influencing factor indicators and temperature forecast error values to eliminate the dimensional influence between different indicators;
[0058] Statistical methods were used to calculate the correlation between each core influencing factor index and temperature forecast error;
[0059] To reduce data dimensionality and eliminate collinearity among indicators, principal component analysis is used. The indicators of each core influencing factor are linearly combined to obtain a comprehensive indicator of the relationship between the principal components and the original indicators. The principal components can retain most of the information of the original data.
[0060] A multiple linear regression model was established using temperature forecast error as the dependent variable and principal components as independent variables.
[0061] Based on the regression model and the relationship between principal components and original indicators, the weights of each core influencing factor on forecast error are calculated.
[0062] In this embodiment, by quantifying the core influencing factors, scientific analysis can be conducted in a data-driven manner, thereby clarifying the specific contribution of each factor to the forecast results. The principal component analysis method effectively reduces collinearity between data dimensions and indicators, making complex multidimensional data easier to process and understand, and comprehensively reflecting the overall characteristics of the influencing factors. By establishing a multiple linear regression model, the influence weight of each core influencing factor on forecast error can be accurately calculated, helping forecasters understand the key factors affecting forecast accuracy, thus enabling targeted improvements in data acquisition and forecast models. In-depth analysis of the influencing factors can identify and correct potential biases, thereby reducing errors in temperature forecasts and improving forecast accuracy.
[0063] In some embodiments of the present invention, for step S3, a global error evaluation model is constructed and optimized to obtain a global error evaluation result;
[0064] Extreme weather is complex and changeable. Minimizing the overall global error as the primary optimization objective can help forecast models reduce various biases and improve accuracy and reliability.
[0065] The overall global error is calculated by weighting at least the extreme specific error, systematic bias, and local bias.
[0066] Extreme specific error is calculated based on extreme threshold error and extreme event hit rate. Extreme threshold error reflects the deviation between the forecast and the actual situation within the extreme threshold range, while extreme event hit rate measures the accuracy of the model's judgment on extreme events. The combination of the two can comprehensively evaluate the model's performance under extreme weather conditions.
[0067] Systematic bias is calculated based on average bias and reflects the overall error trend of the model, thus pointing the way for model improvement;
[0068] Local deviation is calculated based on the root mean square error between points, taking into account the difference between the forecast and the actual situation at each observation point, and accurately reflecting the difference in forecast accuracy at different geographical locations;
[0069] The spatiotemporal steady-state equilibrium of extreme weather forecasting is taken as the first constraint condition. The spatiotemporal steady-state equilibrium includes at least time constraints, spatial constraints and extreme sample constraints.
[0070] Time constraints can prevent unreasonable jumps or abrupt changes in forecasts. Extreme weather has a time evolution pattern. Setting a threshold for the correlation coefficient between the forecast and the actual time series, or limiting the rate of temperature change within the time interval;
[0071] Spatial constraints utilize GIS technology to consider the impact of factors such as geographical location and topography on temperature forecasts; spatial correlation coefficients are set to correlate and correct forecast results for different locations.
[0072] Extreme sample constraints are determined based on the definition and characteristics of extreme weather to set thresholds and screening conditions, ensuring that the samples can reflect the characteristics and are sufficient in number and representative, thereby improving the model's ability to predict extreme weather.
[0073] The first optimization algorithm is adopted to optimize the first optimization objective under the first constraint, and outputs the specific values of the global comprehensive error, the error of each component, and the distribution of key evaluation indicators. The first optimization algorithm is a particle swarm optimization algorithm, in which the error contribution value of each extreme event sample is a particle, and the termination condition of the iterative optimization is that the number of iterations reaches a preset number and the change in the global comprehensive error is less than or equal to a preset threshold.
[0074] In this embodiment, minimizing the global comprehensive error is used as the optimization objective to reduce bias in the forecast model across multiple dimensions, thereby improving the overall accuracy of the forecast. The weighted calculation of extreme-specific errors, systematic biases, and local biases enhances the comprehensiveness and accuracy of the evaluation results. Introducing spatiotemporal steady-state equilibrium as a constraint ensures the rationality of the forecast model over time, avoiding unrealistic forecast results. Simultaneously, the introduction of spatial constraints utilizes GIS technology to consider geographical and environmental factors, improving the local accuracy of the forecast. Setting extreme sample constraints ensures that the selected samples effectively reflect the characteristics and patterns of extreme weather, thereby enhancing the model's ability to cope with extreme weather events. By employing the first optimization algorithm, the optimization objective is pursued under constraints, effectively outputting the specific values of the global comprehensive error and each component error.
[0075] In some embodiments of the present invention, for step S4, a deviation attribution model is constructed and optimized based on the global error evaluation result to obtain a deviation attribution scheme;
[0076] The second optimization objective is to maximize the attribution accuracy of extreme weather temperature forecast deviations. The attribution accuracy is calculated by weighting the attribution accuracy of weather system intensity and underlying surface attribution accuracy of high-error samples. The high-error samples are the samples with the highest predetermined proportion of global comprehensive error in the global error assessment results.
[0077] The intensity of weather systems has a significant impact on temperature forecasts. By analyzing the difference between the forecast and actual weather system intensities in high-error samples, the attribution accuracy is calculated. The physical characteristics of the underlying surface also affect temperature forecasts. The urban heat island effect causes urban areas to be warmer than surrounding rural areas. The contribution of underlying surface factors to forecast errors is analyzed in high-error samples, and the attribution accuracy is calculated.
[0078] The second constraint is the logical consistency of bias attribution. The logical consistency includes at least the following constraints: causal constraint, bias attribution must conform to the basic principles and causal relationships of meteorology; data constraint, the data used for bias attribution must be accurate, complete and representative; and spatiotemporal constraint, considering the influence of time and space factors on bias attribution. In terms of time, the meteorological conditions and forecasting difficulty are different in different seasons and time periods, and bias attribution should take into account these time differences.
[0079] The second optimization algorithm is used to optimize the second optimization objective under the second constraint, and outputs the specific sources of deviation and the distribution statistics of the sources of deviation for each high error sample.
[0080] The second optimization algorithm is the decision tree classification algorithm. The decision tree classification algorithm takes the core influencing factor features of high error samples as input and the type of bias source as output. The type of bias source includes at least the bias of the physical process of the model, the error of the underlying surface parameter, and the error of the observation data. The decision tree performs classification prediction by recursively dividing the data features and constructing a tree structure.
[0081] The second optimization algorithm is used to optimize the second optimization objective under the second constraint, and outputs the specific sources of deviation and the distribution statistics of the deviation sources for each high error sample, including:
[0082] Extracting the core influencing factors of high-error samples;
[0083] Determine the type of bias source corresponding to each high-error sample and use it as the classification label;
[0084] Using criteria such as information gain, information gain ratio, or Gini index, the attribute that can effectively classify the sample is selected from all core influencing factor features as the splitting attribute of the current node;
[0085] Based on the selected splitting attribute and its value, the samples contained in the current node are divided into several subsets, and child nodes are generated.
[0086] The process of selecting the splitting attribute and splitting the node is repeated for each child node until the stopping condition is met, eventually forming a complete decision tree.
[0087] To prevent decision trees from overfitting and improve the model's generalization ability, the constructed decision trees are pruned. Pre-pruning or post-pruning methods can be used, and the optimal pruning strategy can be determined through methods such as cross-validation.
[0088] Input the core influencing factors of new high-error samples into the trained decision tree model, and output the specific source of deviation for each sample based on the model classification results;
[0089] Statistically analyze the frequency of different sources of deviation in all high-error samples and create bar charts or pie charts for visual representation.
[0090] In this embodiment, by weighting the attribution accuracy of weather system intensity and underlying surface characteristics for high-error samples, the accuracy of temperature forecast bias attribution can be improved; ensuring that bias attribution conforms to basic meteorological principles and causal relationships, avoiding attribution results from deviating from the actual situation, and ensuring that the data is accurate, complete, and highly representative; classifying high-error samples using decision trees to accurately identify and statistically analyze different sources of bias; employing criteria such as information gain, information gain ratio, or Gini index for feature selection to effectively improve attribution accuracy and avoid overfitting, ensuring the model's generalization ability; and visually displaying the frequency of different types of bias sources through bar charts or pie charts for easy understanding and analysis; this step, through efficient optimization algorithms and strict constraints, can improve the accuracy of temperature forecasts and the reliability of bias attribution, enhancing the scientific rigor and applicability of meteorological data analysis.
[0091] In some embodiments of the present invention, for step S5, error feedback is performed on extreme weather temperature forecasts based on the global error assessment results and the deviation attribution scheme.
[0092] The feedback information is integrated, and the data from the global error assessment results and the precise deviation attribution scheme are converted into a format suitable for feedback. For the global error assessment results, they are organized into error statistics tables containing indicators such as average error, maximum error, and error distribution range under different regions, time periods, and extreme weather types. For the precise deviation attribution scheme, information such as the type of deviation source and the quantitative values of relevant influencing factors for each high-error sample are organized into structured data. Data visualization tools are used to present the integrated data in an intuitive chart format.
[0093] Based on the main sources of bias identified in the bias attribution scheme, adjust the relevant parameters of the forecast model;
[0094] If the bias attribution results show that the underlying surface parameter error has a significant impact on the forecast error, more detailed and accurate underlying surface data should be introduced into the forecast model.
[0095] The forecast model was retrained using the adjusted model parameters and the introduced external data. During the training process, methods such as cross-validation were used to evaluate the model's performance to ensure that the forecast accuracy under extreme weather conditions was improved. After training, the model was validated using an independent test dataset to check its performance on the new data.
[0096] Prepare a detailed error feedback report, including the overall error assessment results, precise bias attribution scheme, model adjustment suggestions, etc. The report should be described in clear and concise language, avoiding overly technical terms, to ensure that decision-makers can understand it. At the same time, the report should include visual charts to intuitively show the error situation and attribution results.
[0097] Integrating error feedback information into the decision support system provides decision-makers with real-time, accurate error forecasts and improvement suggestions.
[0098] In this embodiment, by providing feedback based on global error assessment results and a precise bias attribution scheme, targeted adjustments can be made to errors in extreme weather forecasts, improving forecast accuracy and reliability and effectively enhancing forecasting capabilities under extreme weather conditions. The error feedback system not only provides error statistics tables listing errors for different regions, time periods, and weather types, but also presents in detail the sources and influencing factors of each high-error sample, thus clearly identifying the root causes of errors. Based on the error analysis results from the feedback, the relevant parameters of the forecast model can be dynamically adjusted, and more accurate external data can be introduced in a timely manner to optimize model training and performance. By integrating error feedback information and transforming it into easy-to-understand reports and visualizations, intuitive and clear feedback can be provided to decision-makers.
[0099] like Figure 2 As shown, the present invention also provides a temperature forecast error assessment system under extreme weather conditions, specifically including the following modules;
[0100] The core influencing factor identification module identifies the core influencing factors for extreme weather temperature forecasts.
[0101] The module for quantifying and weighting influencing factors quantifies the dimensions and corresponding quantitative indicators of each core influencing factor and calculates the weight of each core influencing factor on the forecast error.
[0102] The global error assessment model construction and solution module constructs the global error assessment model and performs optimization to obtain the global error assessment results.
[0103] The deviation attribution model construction and solution module constructs a deviation attribution model and optimizes the solution based on the global error assessment results to obtain a deviation attribution scheme.
[0104] The error feedback module provides error feedback for extreme weather temperature forecasts based on the global error assessment results and the deviation attribution scheme.
[0105] In this embodiment, by identifying and integrating multiple core influencing factors, the system can comprehensively assess the factors affecting temperature forecast errors, thereby improving forecast accuracy. The system not only identifies influencing factors but also quantifies them and calculates their influence weights, enabling decision-makers to more intuitively understand the importance of which factors in specific situations, thus allowing for effective adjustments to response strategies. The construction of a global error assessment model and a deviation attribution model enables a dynamic feedback mechanism. By providing error feedback based on the assessment results, the system makes adjustments to the forecast model more timely and effective, improving the reliability and adaptability of meteorological services. This comprehensive assessment system, considering multiple factors, can provide scientific evidence for meteorological agencies, governments, and relevant departments in responding to extreme weather events, resulting in effective forecasts and accurate error prediction. Feedback helps stakeholders develop more reasonable and efficient response measures, reducing economic losses and social impact. Multi-dimensional factor analysis enhances the adaptability of the assessment model, enabling optimization and adjustment for different extreme weather conditions and providing customized forecast services. The systematic analysis and feedback mechanism makes the assessment results more scientific and reliable, thereby increasing public trust in the forecasts. The comprehensive data and research results it provides can promote cooperation between meteorological departments and other sectors such as agriculture and transportation, forming a more efficient response mechanism. This extreme weather temperature forecast error assessment system, through comprehensive consideration of multiple factors, quantitative analysis, and optimized feedback mechanisms, can provide more accurate, scientific, and efficient temperature forecasts under complex extreme weather conditions.
[0106] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.
Claims
1. A method for evaluating temperature forecast errors under extreme weather conditions, characterized in that, include: Identify the core influencing factors of extreme weather temperature forecasts; The dimensions and corresponding quantitative indicators of each of the core influencing factors are quantified, and the influence weight of each of the core influencing factors on the forecast error is calculated. A global error evaluation model is constructed and optimized to obtain the global error evaluation results; Based on the global error assessment results, a deviation attribution model is constructed and optimized to obtain a deviation attribution scheme. Based on the global error assessment results and the deviation attribution scheme, error feedback is provided for extreme weather temperature forecasts.
2. The method for evaluating temperature forecast errors under extreme weather conditions according to claim 1, characterized in that, The core influencing factors include at least the intensity of the weather system, the characteristics of the underlying surface, and the quality of the observation data.
3. The method for evaluating temperature forecast errors under extreme weather conditions according to claim 1, characterized in that, A global error evaluation model is constructed and optimized to obtain the global error evaluation results, including: The primary optimization objective is to minimize the overall global error in extreme weather temperature forecasts. The first constraint is the spatiotemporal steady-state equilibrium of extreme weather forecasting, which includes at least time constraints, spatial constraints, and extreme sample constraints. The first optimization algorithm is used to optimize the first optimization objective under the first constraint, and outputs the specific values of the global comprehensive error, the error of each component, and the distribution of key evaluation indicators.
4. The method for evaluating temperature forecast errors under extreme weather conditions according to claim 3, characterized in that, The global comprehensive error is calculated by weighting at least the extreme specific error, systematic deviation, and local deviation.
5. The method for evaluating temperature forecast errors under extreme weather conditions according to claim 1, characterized in that, Based on the global error assessment results, a deviation attribution model is constructed and optimized to obtain a deviation attribution scheme, including: The second optimization objective is to maximize the accuracy of attribution of biases in extreme weather temperature forecasts. The second constraint is the logical consistency of bias attribution, which includes at least causal constraints, data constraints, and spatiotemporal constraints. The second optimization algorithm is used to optimize the second optimization objective under the second constraint, and outputs the specific sources of deviation and the distribution statistics of the sources of deviation for each high error sample.
6. The method for evaluating temperature forecast errors under extreme weather conditions according to claim 2, characterized in that, The specific dimensions of the intensity of the weather system include the intensity of extreme events and the duration of extreme events.
7. The method for evaluating temperature forecast errors under extreme weather conditions according to claim 2, characterized in that, The specific dimensions of the underlying surface characteristics include regional cover type and local effects.
8. The method for evaluating temperature forecast errors under extreme weather conditions according to claim 3, characterized in that, The first optimization algorithm is a particle swarm optimization algorithm. The particle swarm optimization algorithm uses the error contribution value of each extreme event sample as a particle. The termination condition for iterative optimization is that the number of iterations reaches a preset number and the change in the global comprehensive error is less than or equal to a preset threshold.
9. The method for evaluating temperature forecast errors under extreme weather conditions according to claim 5, characterized in that, The second optimization algorithm is a decision tree classification algorithm. The decision tree classification algorithm takes the core influencing factor characteristics of high error samples as input and the deviation source type as output. The deviation source type includes at least the model physical process deviation, underlying surface parameter error and observation data error.
10. A temperature forecast error assessment system under extreme weather conditions, characterized in that, include: The core influencing factor identification module identifies the core influencing factors for extreme weather temperature forecasts. The module for quantifying and weighting influencing factors quantifies the dimensions and corresponding quantitative indicators of each core influencing factor and calculates the weight of each core influencing factor on the forecast error. The global error assessment model construction and solution module constructs the global error assessment model and performs optimization to obtain the global error assessment results. The deviation attribution model construction and solution module constructs a deviation attribution model and optimizes the solution based on the global error assessment results to obtain a deviation attribution scheme. The error feedback module provides error feedback for extreme weather temperature forecasts based on the global error assessment results and the deviation attribution scheme.