Quantitative rainfall estimation method based on Hurdle-IMDL framework

Abstract

The invention discloses a quantitative rainfall estimation method based on a Hurdle-IMDL framework. The method comprises the following steps: acquiring historical rainfall measurement data and meteorological satellite observation data; constructing a biased quantitative rainfall estimation probability model based on a Hurdle model, correcting a biased rainfall estimation probability sub-model according to an IMDL method, substituting into empirical distribution to construct an empirical quantitative rainfall estimation probability model, and deriving a negative logarithm likelihood function of the empirical quantitative rainfall estimation probability model; constructing an AI model, and optimizing the AI model by taking a negative log-likelihood function as a loss function; and inputting meteorological satellite observation data of a to-be-inverted region into the trained AI model to obtain an estimated value of a parameter of the empirical quantitative precipitation estimation probability model, and estimating the precipitation according to conditional expectation. According to the invention, the Hurdle model is used to solve the zero expansion problem, the IMDL learning method is used to cope with the long tail problem, and the inversion accuracy of extreme rainfall is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of quantitative remote sensing, quantitative precipitation estimation and artificial intelligence, and in particular to a quantitative precipitation estimation method based on a Hurdle-IMDL artificial intelligence learning framework. BACKGROUND

[0002] In recent years, artificial intelligence (AI) has been widely applied in quantitative remote sensing (QRS), significantly promoting the development of QRS. However, when the label distribution is imbalanced, conventional learning focuses on common samples, while underfitting occurs on rare samples, and the obtained inversion model misestimates rare values, leading to poor performance on extreme events. Among numerous environmental variables, precipitation distribution is the most imbalanced and obvious, and AI-based quantitative precipitation estimation (QPE) models significantly underestimate heavy precipitation.

[0003] In related research, existing methods can be roughly divided into four categories: the first category-decomposition learning. Under this learning paradigm, precipitation distribution imbalance is decomposed into zero inflation (precipitation samples are significantly less than non-precipitation samples) and long tail (strong precipitation samples are much less than weak precipitation samples), and the two parts are treated respectively. Zero inflation is usually solved by two-step modeling: first, detect whether precipitation occurs, then estimate precipitation intensity. The long tail is handled by data resampling, ensemble learning or introducing specific constraints to the objective function. This method still has significant shortcomings in strong precipitation inversion and cannot effectively solve the problem. The second category-cost sensitive learning. The mean square error (MSE) is decomposed into a non-precipitation term (observation = 0) and a precipitation term (observation > 0), the latter is further divided into an underestimation term (inversion < observation) and an overestimation term (inversion > observation), and each term is assigned a weight to adjust the model bias. This method improves the consistency of inversion and observation value distribution and improves the performance of strong precipitation detection; however, when the precipitation is greater than 15mm·h -1 , the model still shows significant shortcomings. The third category-generative learning. Some scholars have proposed a generative model for QPE, and the evaluation shows that the model has excellent ability to reconstruct the fine structure of precipitation space, which is closely related to extreme precipitation. At the same time, frequency distribution analysis shows that the number of strong precipitation samples inverted by generative learning is significantly more than that by conventional learning. The generative model shows the potential to enhance precipitation inversion, but its effectiveness still needs further verification. The fourth category-multitask learning. A collaborative multitask learning method enables detection and estimation tasks to promote each other. For the case where the precipitation rate is higher than 10mm·h -1 , the detection performance is improved by 4%. However, the analysis is limited to the case where the precipitation rate is higher than 10mm·h -1 , and lacks evaluation of strong to extreme precipitation inversion error, so the effectiveness of this method is not fully clear.

[0004] In summary, there is no clear and effective method to deal with the precipitation imbalance in QPE, so as to improve the inversion of heavy to extreme precipitation. SUMMARY

[0005] The present application aims to at least partially solve one of the technical problems existing in the related art.

[0006] One object of the present application is to provide a quantitative precipitation estimation method based on a Hurdle-IMDL framework, which uses a Hurdle model to solve the zero inflation problem, and uses an inversion model debiasing learning (IMDL) to deal with the problem of long-tailed distribution, so as to improve the inversion effect of heavy precipitation and even extreme precipitation.

[0007] In order to achieve the above-mentioned purpose, one aspect of the present application provides a quantitative precipitation estimation method based on a Hurdle-IMDL framework, comprising the following steps: Obtaining historical precipitation measurement data and meteorological satellite observation data to form a data set, wherein the precipitation labels of the data set are characterized by zero inflation and non-zero part of long-tailed distribution; Based on the Hurdle model, a biased quantitative precipitation estimation probability model is constructed, which consists of two parts: a rain area detection probability model and a biased rainfall estimation probability model. Based on the IMDL, the biased rainfall estimation probability model is converted into an ideal rainfall estimation probability model to obtain an ideal quantitative precipitation estimation probability model. Then, an empirical distribution is introduced into the ideal quantitative precipitation estimation probability model to construct an empirical quantitative precipitation estimation probability model and derive its negative log-likelihood function; An AI model for fitting the empirical quantitative precipitation estimation probability model is constructed. The negative log-likelihood function is used as a loss function to train the AI model, so that the AI model can achieve optimal estimation of the parameters of the empirical quantitative precipitation estimation probability model; The meteorological satellite observation data of the area to be inverted is input into the trained AI model to obtain the parameter estimation value of the corresponding empirical quantitative precipitation estimation probability model, and then the precipitation is estimated according to the conditional expectation.

[0008] The further preferred technical solution of the present application is to construct an empirical quantitative precipitation estimation probability model under the Hurdle-IMDL framework. Specifically: Based on the Hurdle model, a biased quantitative precipitation estimation probability model is constructed, which consists of two parts: a rain area detection probability model and a biased rainfall estimation probability model, which are used to model precipitation occurrence and precipitation amount respectively, to deal with the zero inflation data characteristics of the data set; For the biased rainfall estimation probability model, the IMDL method is used to establish a transformation relationship between the ideal rainfall estimation probability model and the biased rainfall estimation probability model. Based on the transformation relationship, the biased rainfall estimation probability model is converted into the ideal rainfall estimation probability model, and the biased quantitative precipitation estimation probability model is corrected into the ideal quantitative precipitation estimation probability model. The ideal quantitative precipitation estimation probability model is learned from the unprocessed long-tailed distribution data. The biased quantitative precipitation estimation probability model is replaced with the biased rainfall estimation model in the biased quantitative precipitation estimation probability model, and the ideal model is obtained.

[0009] Preferably, the Hurdle model, which constructs a biased quantitative precipitation estimation probability model, consists of two parts: a rain zone detection model and a biased precipitation estimation model, which respectively model precipitation occurrence and precipitation amount, as follows: Regarding precipitation and remote sensing signals Its conditional probability density function is expressed as: ; in, Indicates that in a given remote sensing signal The probability that precipitation will not occur under certain conditions; Indicates that in a given remote sensing signal Under these conditions, precipitation Take a specific positive value The probability of biased rainfall estimation model; The modeling process consists of two parts. The first part is modeling the occurrence of precipitation, corresponding to the estimation... The second part involves modeling precipitation and fitting the data. .

[0010] As a preferred option, for the precipitation modeling part, the established transformation relationship between the ideal rainfall estimation model and the biased rainfall estimation model is expressed as follows: ; in, It is a biased rainfall estimation probability model. For the probability model of ideal rainfall estimation, The probability distribution of historical precipitation measurement data in the dataset; Replacing the biased rainfall estimation model with an ideal rainfall estimation model, the probabilistic model for ideal quantitative precipitation estimation is expressed as: .

[0011] Preferably, the step of introducing an empirical distribution into the ideal quantitative precipitation estimation probability model based on practical scenarios, constructing an empirical quantitative precipitation estimation probability model, and deriving its negative log-likelihood function is as follows: Replacing the biased rainfall estimation model in the biased quantitative precipitation estimation probability model with an ideal rainfall estimation model, an ideal quantitative precipitation estimation probability model is obtained, and its expression is: ; Wherein, The parameter of the ideal rainfall estimation probability model is represented by The parameter of is represented by It is estimated from the data set by statistical method; The negative log-likelihood of the empirical quantitative precipitation estimation probability model served for its fitting, is represented by ; Each term is respectively ; ; ; ; Wherein, is the number of labels in the data set, is represented by the indicator function.

[0012] As preferred, based on the constructed ideal quantitative precipitation estimation probability model, the three parameters to be estimated are determined as , wherein is regarded as a hyperparameter and is given a fixed value; The parameter is estimated by constructing an AI model, which is an improved version of the ordinary U-Net network, with an additional output module added at the end of the network to estimate and at the same time; the improved network extracts deep features of meteorological satellite observation data , which are then processed by two branches, the first branch consisting of stacked 1x1 convolutional layers, with a ReLU activation function immediately following the first layer of convolutional layers, to estimate the parameter ; the second branch estimates by a parallel sequence ending with a Sigmoid activation function. The above U-Net network is trained with the negative log-likelihood function as the loss function.

[0013] As preferred, the precipitation is estimated according to the conditional expectation, specifically: .

[0014] Another aspect of the present application provides a non-transitory computer readable storage medium having stored thereon computer instructions, which cause a computer to execute the Hurdle-IMDL framework-based quantitative precipitation estimation method described above.

[0015] Still another aspect of the present application provides an electronic device, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus, and the processor invokes the logic instructions in the memory to execute the Hurdle-IMDL framework-based quantitative precipitation estimation method described above.

[0016] Yet another aspect of the present application provides a computer program product, which comprises a computer program stored on a non-transitory computer readable storage medium, and the computer program is executed by a processor to cause a computer to execute the Hurdle-IMDL framework-based quantitative precipitation estimation method described above.

[0017] Advantages: The Hurdle-IMDL framework-based quantitative precipitation estimation method of the present application proposes a learning framework-Hurdle-IMDL, wherein the Hurdle branch constructs a biased quantitative precipitation estimation probability model, and the IMDL corrects the biased rainfall estimation model in the biased quantitative precipitation estimation probability model, so that the AI model for estimating the parameters of the quantitative precipitation estimation probability model automatically corrects the tendency of heavy rain underestimation in the training stage, and outputs the already-debiased pixel-level precipitation intensity once in the forward reasoning, without additional resampling, weighting or post-processing, so as to obtain quantitative precipitation estimation results with complete spatial details, strong real-time performance and good generalization ability, and significantly improve the reliability of extreme precipitation detection and intensity estimation. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 Figure 1 is a neural network structure diagram in Embodiment 1 of the present application; Figure 2 Figure 2 is the inversion result of each model of the first precipitation event in Embodiment 1 of the present application; Figure 3 Figure 3 is the inversion result of each model of the second precipitation event in Embodiment 1 of the present application; Figure 4 Figure 4 is the comparative evaluation result of Hurdle-IMDL and the benchmark model in Embodiment 1 of the present application. DETAILED DESCRIPTION

[0019] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described clearly and completely below in conjunction with the drawings in the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, and should not be construed as limiting the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application. In the description of the present application, it should be understood that the terms used are only for the purpose of description and should not be construed as indicating or implying relative importance.

[0020] The following will be described in conjunction with Figures 1-4 The present application provides a Hurdle-IMDL framework-based quantitative precipitation estimation method.

[0021] Embodiment 1: The present embodiment provides a Hurdle-IMDL framework-based quantitative precipitation estimation method, which specifically comprises the following steps: S1, obtaining historical precipitation measurement data and meteorological satellite observation data to form a data set.

[0022] In the present embodiment, a specific region (115.75°E–120.75°E, 28.0°N–33.0°N) is selected, and precipitation measurement data and observation data of the Japanese new-generation geostationary meteorological satellite Himawari-8 in each summer (May–August) from 2016 to 2021 are collected. The region contains 5,407 precipitation measurement stations, and the dense distribution ensures that reliable gridded precipitation products can be obtained by interpolation.

[0023] In the standard mode, the Advanced Himawari-8 Imaging Instrument (AHI) carried on the Himawari-8 provides full-disk data every 10 minutes. The AHI contains 16 channels (band01–band16) covering visible light, near-infrared and far-infrared wavelengths. One infrared channel and five brightness temperature differences are selected as input features. A total of 4,883 time step precipitation event samples are selected. Through the nearest neighbor interpolation method, the half-hour satellite data and the hourly cumulative precipitation data are interpolated to a 0.05°×0.05° grid (100×100 grid points), and finally 4,883 spatio-temporal matching samples are obtained. The samples in 2016–2019, 2020 and 2021 are used for training, verification and testing, respectively. The statistical results of the training samples show that the ratio of precipitation to non-precipitation grid points is about 3.6:1, in which ≈0.46 and , showing significant zero inflation and long-tail characteristics.

[0024] S2, constructing an ideal quantitative precipitation estimation probability model according to the Hurdle model and the IMDL.

[0025] The Hurdle model is a statistical model specifically designed to handle zero inflation problems. It is widely used in fields such as healthcare spending, species abundance counting, and insurance claims, and has achieved successful applications in precipitation forecasting. (Regarding precipitation...) and remote sensing signals Its conditional probability density function (PDF) is expressed as: (1); in, Indicates that in a given remote sensing signal The probability that precipitation will not occur under certain conditions; Indicates that in a given remote sensing signal Under these conditions, precipitation Take a specific positive value The probability of; The biased quantitative precipitation estimation probabilistic model consists of two parts: a rain zone detection model and a biased precipitation estimation model. Correspondingly, the modeling process is divided into two parts: the first part models precipitation occurrence and the corresponding estimation... The second part involves modeling precipitation and fitting the data. .

[0026] Although the Hurdle model addresses the zero-inflation problem by separating zero values ​​from positive values, conventional learning methods may still underestimate heavy rainfall due to long-tailed distributions in its precipitation modeling. To address this issue, this embodiment employs the IMDL learning method to establish a transformation relationship linking the ideal rainfall estimation model and the biased rainfall estimation model, learning the ideal rainfall estimation model from unprocessed long-tailed data.

[0027] When trained on datasets with long-tailed distributions, conventional learning methods result in biased rainfall estimation models. This phenomenon can be expressed as: (2); A more in-depth analysis of formula (2) is needed. Forward modeling describes how to extract values ​​from environmental variables (such as precipitation). Generate remote sensing signals ,use This refers to the forward model. The inversion aims to... Precipitation was calculated from the middle ,use This refers to the inversion model. According to Bayes' theorem, forward and inversion models are related to long-tail datasets (...). and The relationship is as follows: (3); As shown in formula (3), It not only relies on long-tail datasets, but also on forward modeling. Therefore, formula (2) can be further extended to: (4); Now consider an ideal scenario: an ideal balanced dataset, where... and The distribution is uniform, and they are denoted as follows: and At the same time, there is also an ideal forward model. Under these ideal conditions, the ideal rainfall estimation model obtained through conventional learning... There is no deviation, which is expressed as: (5); According to Bayes' theorem, the following equation holds: (6); In practice, perfectly balanced datasets are unavailable due to theoretical and technological limitations. However, a key invariant exists: the forward model. Primarily determined by physical processes, it is independent of data acquisition and modeling, and therefore unaffected by data and algorithms. This invariance can be expressed as: (7); Based on formulas (3), (6) and (7), it can be derived that: (8); Considering From formula (8), we further obtain: (9); Based on the above derivation steps, we can obtain: (10); because For uniform distribution, the numerator and denominator in You can make an appointment to go, and you will eventually get: (11); Formula (11) establishes a model for estimating ideal rainfall. And biased rainfall estimation model The transformation relationship that connects them.

[0028] The goal of this embodiment is to learn from the long-tail dataset using formula (11). Assuming Depend on Parameterization, denoted as Given a containing Independent and identically distributed samples For long-tailed datasets, the likelihood function Defined as: (12); Substituting formula (11) into formula (12), we get: (13); parameter Estimate using the following equation: (14); here, This represents the parameters of the ideal rainfall estimation model that maximizes the likelihood function given a long-tailed dataset. This process allows for the direct fitting of the ideal rainfall estimation model to unprocessed long-tailed data. This learning method is known as Inversion Model Debiasing Learning (IMDL).

[0029] The Hurdle model and IMDL together constitute the complete Hurdle-IMDL framework. The ideal quantitative precipitation estimation probability model obtained after modifying the biased quantitative precipitation estimation probability model is expressed as follows: (15).

[0030] S4. Based on practical scenarios, introduce empirical distribution into the ideal quantitative precipitation estimation probability model, construct the empirical quantitative precipitation estimation probability model, and derive its negative log-likelihood function.

[0031] To enable the practical application of the ideal quantitative precipitation estimation probability model, an empirical distribution is introduced, and its negative log-likelihood function is derived. The log-normal distribution is widely used to characterize... The distribution has also been successfully applied to modeling given satellite observation conditions. The conditional distribution. According to formulas (1) and (11), It can be represented as: (16); in, These represent the parameters of the ideal rainfall estimation model; express The parameters, It is estimated from the dataset using statistical methods.

[0032] Negative log-likelihood of empirical quantitative precipitation estimation probability model To achieve its fitted objective function, which has an analytical solution, it can be decomposed into the sum of the following terms: (17); ; ; ; ; in, The number of the label in the dataset. Indicates an indicator function. It is a correction item that embodies IMDL's adjustments and guidance to conventional learning.

[0033] S5. Construct an AI model to fit the above empirical quantitative precipitation estimation probability model; train the AI ​​model using the negative log-likelihood function.

[0034] Based on the above formula, the three parameters to be estimated are as follows: Based on the observed phenomena, this embodiment employs a hybrid estimation scheme. For It can be observed that without range constraints on the AI ​​model output, its estimates often diverge to infinity, causing the model to fail; on the other hand, when range constraints are applied, all estimates converge uniformly to the set upper limit. This phenomenon suggests that the current IMDL is insufficient to guide the development of AI models. Differential dynamic estimation. To address this problem, The parameter is treated as a hyperparameter and assigned a fixed value, which is selected from the candidate set through an external optimization method such as grid search. Although this hyperparameterization is a compromise, subsequent quantitative precipitation estimation results indicate that it is an effective strategy.

[0035] For parameters This embodiment employs an improved U-Net network for simultaneous estimation. U-Net was originally developed for medical image segmentation and has since been widely applied in computer vision, weather forecasting, and remote sensing. Inspired by Mixture Density Networks, this embodiment extends U-Net by adding an additional output module. Figure 1 This embodiment demonstrates the overall design of the U-Net network. The improved U-Net network extracts depth features from meteorological satellite observation data. Then it is processed by two branches. The first branch consists of stacked 1×1 convolutional layers, where the ReLU activation function is used after the first convolutional layer to estimate the parameters. The second branch estimates the value of a parallel sequence ending with a sigmoid activation function. .

[0036] S6. Input the meteorological satellite observation data of the area to be inverted into the trained network to obtain the corresponding data. and The estimated value, combined with the given Value, through conditional expectation The specific amount of precipitation is determined by the following expression: (18).

[0037] To evaluate the effectiveness and superiority of Hurdle-IMDL, this embodiment selected five benchmarks, namely: The original mean squared error (OMSE) represents a learning method with mean squared error (MSE) as the objective function. Nonlinear weighted mean squared error (NWMSE) corresponds to cost-sensitive learning methods. Linear weighted mean squared error (LWMSE) is a cost-sensitive learning method designed for imbalanced label distribution in computer vision tasks. Diffusion is a typical generative model. Multi-task collaborative deep learning framework (MTCF) corresponds to multi-task learning methods.

[0038] A tiered evaluation method was adopted, with 12 threshold values ​​set (0, 0.1, 0.5, 1, 2, 3, 5, 7, 10, 15, 20, 30 mm·h). -1 The model defines 12 levels, each corresponding to whether an observation is greater than or equal to a given threshold. Inversion error is measured by root mean square error (RMSE) and mean error (ME). RMSE ranges from 0 to +∞, with smaller values ​​indicating lower error (higher accuracy), while ME reflects systematic bias: ME < 0 indicates underestimation, ME > 0 indicates overestimation, and larger absolute values ​​indicate stronger bias. Detection performance is measured by probability of detection (POD), false alarm rate (FAR), and fair threat score (ETS). POD ranges from 0 to 1, with higher values ​​indicating stronger detection capability and fewer missed detections. FAR also ranges from 0 to 1, with smaller values ​​indicating fewer false detections. The fair threat score (ETS) is commonly used to evaluate extreme precipitation forecasts and provides a more comprehensive measure of detection performance compared to POD or FAR. Its value ranges from -1 / 3 to 1, with higher values ​​indicating better model detection performance, while ETS ≤ 0 indicates no predictive ability.

[0039] Two typical precipitation events were used as case studies to evaluate the advantages of Hurdle-IMDL.

[0040] The first case example Figure 2 As shown, the rainfall occurred at 04:00 UTC on July 2, 2021, when a Meiyu front triggered continuous precipitation in East China. The precipitation band exhibited a zonal distribution, with a peak intensity of 48.9 mm·h⁻¹. -1The precipitation intensity retrieved by OMSE and Diffusion is generally weak, with local maximum values ​​below 30 mm·h. -1 Although LWMSE and NWMSE generated yields higher than 30 mm·h -1 The values ​​were calculated, but areas of heavy precipitation were significantly underestimated. In contrast, OMSE, LWMSE, NWMSE, and Diffusion significantly overestimated values ​​above 1 mm·h⁻¹. -1 The Hurdle-IMDL method misclassifies many samples with no precipitation or light rain as moderate rain due to its inaccurate prediction of precipitation range. Overall, the Hurdle-IMDL provides the most reasonable estimate, capturing both the extent of the precipitation band and accurately estimating the precipitation intensity.

[0041] The second precipitation case is as follows Figure 3 As shown, this occurred at 06:00 UTC on July 7, 2021, by which time the Meiyu front had dissipated. Under conditions of high temperature and humidity, convective cells developed in multiple locations, producing sporadic precipitation, with localized intensity reaching 30 mm·h. -1 In this context, MTCF and Diffusion significantly overestimated the precipitation area, while OMSE, LWMSE, and NWMSE underestimated the precipitation intensity. This again demonstrates that Hurdle-IMDL best approximates the observed results in both spatial extent and precipitation intensity.

[0042] Based on the entire test set, a quantitative comparison was conducted between Hurdle-IMDL and various benchmark models, and the results are as follows: Figure 4 As shown, when the threshold is between 0 and 10 mm·h -1 In the meantime, the RMSE differences among the methods are relatively small. However, for heavier precipitation (threshold ≥15 mm·h), the differences are less pronounced. -1 The Hurdle-IMDL model showed a significantly lower RMSE than all baseline models. Regarding ME, most methods exhibited negative values, demonstrating a systematic underestimation, particularly of heavy precipitation. However, the Hurdle-IMDL model had the smallest absolute ME value among all methods, significantly reducing the underestimation. Diffusion was observed at thresholds between 0.1 and 5 mm·h. -1 The highest POD was observed, but it also showed the highest FAR, indicating an overestimation of samples with no precipitation and light rain. In contrast, the Hurdle-IMDL showed the highest POD at a threshold greater than or equal to 7 mm·h. -1 It reached its maximum POD during extremely heavy precipitation (threshold ≥ 20 mm·h), while its FAR only occurred during extremely heavy precipitation (threshold ≥ 20 mm·h). -1 The Hurdle-IMDL outperformed other methods in mitigating the underestimation of heavy to extreme precipitation, demonstrating its effectiveness. Finally, the Hurdle-IMDL consistently achieved higher ETS values ​​than all baseline models across all levels. The relative improvement became more significant as the threshold increased. Notably, for thresholds greater than or equal to 30 mm·h… -1In the case of precipitation imbalance, the baseline model's ETS is zero (indicating negligible detection capability), while Hurdle-IMDL's ETS consistently remains above 0.1. Overall, these results demonstrate that Hurdle-IMDL effectively addresses uneven precipitation distribution and enhances model inversion performance, particularly in the case of heavy to extreme precipitation.

[0043] Example 2: This example provides a non-transitory computer-readable storage medium storing computer instructions that cause a computer to execute a quantitative precipitation estimation method based on the Hurdle-IMDL framework. The method includes the following steps: Historical precipitation measurement data and meteorological satellite observation data are acquired to form a dataset. The precipitation labels in the dataset exhibit zero expansion and the non-zero portion shows a long-tail distribution characteristic. A biased quantitative precipitation estimation probability model is constructed based on the Hurdle model, which consists of two parts: a rain area detection probability model and a biased rainfall estimation probability model. The biased model is converted into an ideal model based on IMDL to obtain an ideal quantitative precipitation estimation probability model. Based on practical scenarios, empirical distributions are introduced into the ideal quantitative precipitation estimation probability model to construct an empirical quantitative precipitation estimation probability model and derive its negative log-likelihood function. An AI model is constructed to fit the above-mentioned empirical quantitative precipitation estimation probability model; the negative log-likelihood function is used as the loss function to train the AI ​​model; Meteorological satellite observation data of the area to be inverted are input into the trained AI model to obtain the estimated values ​​of the parameters of the above empirical quantitative precipitation estimation probability model, and then the precipitation is estimated based on the conditional expectation.

[0044] Example 3: This example provides an electronic device that may include a processor, a communication interface, a memory, and a communication bus. The processor, communication interface, and memory communicate with each other via the communication bus. The processor can call logical instructions from the memory to execute a quantitative precipitation estimation method based on the Hurdle-IMDL framework. This method includes the following steps: Historical precipitation measurement data and meteorological satellite observation data are acquired to form a dataset. The precipitation labels in the dataset exhibit zero expansion and the non-zero portion shows a long-tail distribution characteristic. A biased quantitative precipitation estimation probability model is constructed based on the Hurdle model, which consists of two parts: a rain area detection probability model and a biased rainfall estimation probability model. The biased model is converted into an ideal model based on IMDL to obtain an ideal quantitative precipitation estimation probability model. Based on practical scenarios, empirical distributions are introduced into the ideal quantitative precipitation estimation probability model to construct an empirical quantitative precipitation estimation probability model and derive its negative log-likelihood function. An AI model is constructed to fit the above-mentioned empirical quantitative precipitation estimation probability model; the negative log-likelihood function is used as the loss function to train the AI ​​model; Meteorological satellite observation data of the area to be inverted are input into the trained AI model to obtain the parameter estimates of the empirical quantitative precipitation estimation probability model, and then the precipitation is estimated based on the conditional expectation.

[0045] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0046] Example 4: This example provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can perform a quantitative precipitation estimation method based on the Hurdle-IMDL framework. This method includes the following steps: Historical precipitation measurement data and meteorological satellite observation data are acquired to form a dataset. The precipitation labels in the dataset exhibit zero expansion and the non-zero portion shows a long-tail distribution characteristic. A biased quantitative precipitation estimation probability model is constructed based on the Hurdle model, which consists of two parts: a rain area detection probability model and a biased rainfall estimation probability model. The biased model is converted into an ideal model based on IMDL to obtain an ideal quantitative precipitation estimation probability model. Based on practical scenarios, empirical distributions are introduced into the ideal quantitative precipitation estimation probability model to construct an empirical quantitative precipitation estimation probability model and derive its negative log-likelihood function. An AI model is constructed to fit the above-mentioned empirical quantitative precipitation estimation probability model; the negative log-likelihood function is used as the loss function to train the AI ​​model; Meteorological satellite observation data of the area to be inverted are input into the trained AI model to obtain the parameter estimates of the empirical quantitative precipitation estimation probability model, and then the precipitation is estimated based on the conditional expectation.

[0047] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0048] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0049] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A quantitative precipitation estimation method based on the Hurdle-IMDL framework, characterized in that, Includes the following steps: Historical precipitation measurement data and meteorological satellite observation data are acquired to form a dataset. The precipitation labels in the dataset exhibit zero expansion and the non-zero portion shows a long-tail distribution characteristic. A biased quantitative precipitation estimation probability model is constructed based on the Hurdle model, which consists of two parts: a rain area detection probability model and a biased rainfall estimation probability model. The biased rainfall estimation probability model is transformed into an ideal rainfall estimation probability model based on IMDL to obtain an ideal quantitative precipitation estimation probability model. Then, empirical distribution is introduced into the ideal quantitative precipitation estimation probability model to construct an empirical quantitative precipitation estimation probability model and derive its negative log-likelihood function. An AI model is constructed to fit the empirical quantitative precipitation estimation probability model; the negative log-likelihood function is used as the loss function to train the AI ​​model. Meteorological satellite observation data of the area to be inverted are input into the trained AI model to obtain the parameter estimates of the ideal quantitative precipitation estimation probability model, and then the precipitation is estimated based on the conditional expectation.

2. The quantitative precipitation estimation method based on the Hurdle-IMDL framework according to claim 1, characterized in that, The biased quantitative precipitation estimation probability model constructed based on the Hurdle model consists of two parts: a rain area detection probability model and a biased precipitation estimation probability model. The biased precipitation estimation probability model is transformed into an ideal precipitation estimation probability model based on IMDL, thus obtaining the ideal quantitative precipitation estimation probability model. Then, an empirical distribution is introduced into the ideal quantitative precipitation estimation probability model to construct an empirical quantitative precipitation estimation probability model and derive its negative log-likelihood function. Specifically: A biased quantitative precipitation estimation probability model is constructed based on the Hurdle model. It consists of two parts: a rain area detection probability model and a biased precipitation estimation probability model. These models model precipitation occurrence and precipitation amount respectively, in order to deal with the zero-inflation data characteristics of the dataset. For the biased rainfall estimation probability model, the IMDL method is used to establish a transformation relationship between the ideal rainfall estimation probability model and the biased rainfall estimation probability model. Based on the transformation relationship, the biased rainfall estimation probability model is converted into the ideal rainfall estimation probability model, and the biased quantitative precipitation estimation probability model is corrected into the ideal quantitative precipitation estimation probability model. The ideal quantitative precipitation estimation probability model is learned from the unprocessed long-tailed distribution data.

3. The quantitative precipitation estimation method based on the Hurdle-IMDL framework according to claim 2, characterized in that, The partial quantitative precipitation estimation probability model constructed based on the Hurdle model is as follows: Regarding precipitation and remote sensing signals Its conditional probability density function is expressed as: ； in, Indicates that in a given remote sensing signal The probability that precipitation will not occur under certain conditions; Indicates that in a given remote sensing signal Under these conditions, precipitation Take a specific positive value The probability of; The modeling process consists of two parts. The first part is modeling the occurrence of precipitation, corresponding to the estimation... The second part involves modeling precipitation and fitting the data. .

4. The quantitative precipitation estimation method based on the Hurdle-IMDL framework according to claim 3, characterized in that, For the precipitation modeling part of the biased quantitative precipitation estimation probability model, the transformation relationship between the ideal rainfall estimation probability model and the biased rainfall estimation probability model is established as follows: ； in, It is a biased rainfall estimation probability model. For the probability model of ideal rainfall estimation, The probability distribution of historical precipitation measurement data in the dataset; Based on the transformation relationship, the ideal quantitative precipitation estimation probability model is obtained by replacing the biased precipitation estimation probability model with the ideal precipitation estimation probability model, which is expressed as: 。 5. The quantitative precipitation estimation method based on the Hurdle-IMDL framework according to claim 4, characterized in that, The process involves introducing an empirical distribution into the ideal quantitative precipitation estimation probability model, constructing the empirical quantitative precipitation estimation probability model, and deriving its negative log-likelihood function; specifically: Using the log-normal distribution as the random variable and The empirical distribution of the ideal quantitative precipitation estimation probability model is parameterized as follows: ； in, These represent the parameters of the probabilistic model for estimating ideal rainfall. express The parameters, It is estimated from the dataset using statistical methods; Negative log-likelihood of an ideal quantitative precipitation estimation probability model The objective function used to estimate its parameters is expressed as: ； The parsing terms are as follows: ； ； ； ； in, The number of the label in the dataset. Indicates an indicator function.

6. The quantitative precipitation estimation method based on the Hurdle-IMDL framework according to claim 5, characterized in that, Based on the constructed ideal quantitative precipitation estimation probability model, the three parameters to be estimated are determined as follows: ,in Treat them as hyperparameters of the AI ​​model and assign them fixed values; parameter Estimation is performed using a constructed AI model, which is an improved U-Net network with an additional output module added at the network's end to simultaneously estimate... and ; An improved U-Net network extracts depth features from meteorological satellite observation data. Then it is processed by two branches. The first branch consists of stacked 1×1 convolutional layers, where the ReLU activation function is used after the first convolutional layer to estimate the parameters. The second branch estimates the parameters using a parallel sequence ending with a sigmoid activation function. .

7. The quantitative precipitation estimation method based on the Hurdle-IMDL framework according to claim 6, characterized in that, The estimation of precipitation based on conditional expectations is as follows: 。 8. A non-transitory computer-readable storage medium, characterized in that, It stores computer instructions that cause the computer to execute the quantitative precipitation estimation method based on the Hurdle-IMDL framework as described in any one of claims 1-7.

9. An electronic device, characterized in that, include: The system includes a processor, a communication interface, a memory, and a communication bus. The processor, communication interface, and memory communicate with each other via the communication bus. The processor calls logical instructions from the memory to execute the quantitative precipitation estimation method based on the Hurdle-IMDL framework as described in any one of claims 1-7.

10. A computer program product, characterized in that, The computer program product includes a computer program stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer performs the quantitative precipitation estimation method based on the Hurdle-IMDL framework as described in any one of claims 1-7.

Images