Mountainous area low-order small-watershed rainfall spatial distribution mode analysis method and system

Abstract

The invention relates to the technical field of small watershed landslide early warning, and discloses a mountainous area low-order small watershed rainfall spatial distribution pattern analysis method and system, and the method comprises the steps: employing a Pearson correlation coefficient to evaluate the correlation of rainfall data of at least one rainfall monitoring point at different time scales; forming a data set based on the evaluation result, establishing a rainfall monitoring space model, and training the rainfall monitoring space model by using the training set; inputting the rainfall data into the rainfall monitoring space model for simulation, dividing a rainfall data target area according to a simulation result, and uploading a division result to a cloud platform; and comparing the real-time data with the prediction data, and adjusting the rainfall monitoring space model according to a comparison result. The system comprises a historical rainfall data acquisition module, a spatial model construction module and a real-time adjustment verification module. According to the method, a spatial model capable of accurately predicting rainfall can be constructed, and the generalization ability and prediction precision of the model are improved.

Description

Technical Field

[0001] The invention relates to the technical field of small watershed landslide early warning, and in particular to a method and system for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas. Background Art

[0002] At present, the critical rainfall threshold for early warning of landslide disasters still has a high false alarm rate. One of the reasons is that there is a huge uncertainty in rainfall as an input item, which is particularly serious in mountainous areas with complex terrain. Explaining this problem is still a huge challenge for today's rain gauge monitoring network. Through two rainfall stations (1632m and 2200m) at different altitudes in a ditch basin (upstream sub-basin of a ditch) for two years of rainfall monitoring, an attempt was made to reveal the spatial distribution pattern of rainfall from the perspective of low-order small watersheds in mountainous areas. The results show that rainfall at different altitudes in low-order watersheds in mountainous areas is highly correlated and significantly different. First, at different time scales (month, week, day, hour), the PCC of rainfall data at both locations reached above 0.9. However, the rainfall in the upper reaches of the basin is significantly greater than that in the lower reaches, and the average vertical rainfall gradient between the two during the monitoring period is 35.6mm / 100m. The reason for this phenomenon is the higher rainfall intensity and longer rainfall duration in the upper reaches of the basin, and the contribution of high-intensity rainfall is greater. Secondly, the comparative analysis of daily rainfall levels shows that when high-intensity rainfall occurs, the rainfall in the upper reaches of the basin is one level higher than that in the lower reaches, which is more obvious in years with abundant rainfall. In this way, in the future, we may be able to use the rainfall level at the bottom of the valley at low altitudes to predict the rainfall level at high altitudes where slope instability is prone to occur and there is no monitoring equipment based on this feature. Finally, the identification and analysis of IPEs show that when warning of rainfall-type landslide disasters, more attention should be paid to the rainfall conditions in the upper reaches of the basin, because these locations are more prone to extreme and abnormally high-intensity rainfall events. In the early warning of landslides based on rainfall critical thresholds, special attention should be paid to the quality of rainfall data. The way to improve is to increase the number of rain gauges in potential landslide source areas, rather than the number of rain gauges in low-altitude residential areas.

[0003] Prior art 1, Chinese patent, patent number: 202410011156.3 discloses a flash flood forecasting method that couples physical mechanism with deep learning model, which involves the field of flash flood forecasting in small watersheds. It includes the following steps: obtaining meteorological and hydrological data; calibrating the parameters of the physical mechanism hydrological model according to the meteorological and hydrological data, and establishing a physical mechanism hydrological model; forecasting flow data based on the physical mechanism hydrological model, and extracting the first time series features of the forecast flow data and the historical flow data respectively; obtaining rainfall map data, and using the graph convolutional neural network and the gated recurrent unit to extract the second time series features of the rainfall map data; the first time series features and the second time series features are fully connected and calculated to realize the prediction of flash flood flow. Although the graph convolutional neural network based on the spatial domain is connected with the gated recurrent unit to mine the time series features of rainfall data, it fully considers the uneven spatial distribution of short-duration rainfall in mountainous areas and improves the prediction accuracy of the model; however, the temporal and spatial resolution is low, resulting in large errors.

[0004] Prior art 2, Chinese patent, patent number: 202410558495.3 discloses a real-time rainfall field assimilation method that integrates rain gauges, monitoring images and numerical weather forecasts. By decomposing the rainfall layer in the monitoring image, a deep learning model is established to solve the rainfall intensity, and the point-distributed ground rain gauges and monitoring image rainfall data are spatially interpolated to obtain the spatial distribution of two independent source rainfalls; and combined with the rainfall spatial distribution data in the numerical weather forecast, the Bayesian fusion technology is used to obtain a more accurate real-time rainfall field result after assimilation. Although the provided method for obtaining a more accurate real-time rainfall field assimilation overcomes the drawbacks of the current independent source rainfall data in the field of meteorology and hydrology that it is difficult to simultaneously guarantee "accurate point estimation and reliable spatial changes" or the problem that it is difficult to consider monitoring image rainfall measurement when assimilating rainfall information from multiple sources, it is suitable for real-time forecasting and warning of urban rainstorms and floods, and provides a technical basis for urban flood control, disaster reduction and emergency management; however, the uncertainty analysis is insufficient, resulting in the inability to accurately forecast.

[0005] Prior art three, Chinese patent, patent number: 202410234086.8 belongs to the field of fluid simulation technology, and involves a dynamic simulation method considering the interaction between runoff and debris flow, including constructing a rainfall spatial distribution model; constructing a vegetation interception model; constructing a soil infiltration coupling model; determining rainfall data, vegetation interception rainwater data, soil infiltration rainwater data; determining residual rainwater data; obtaining characteristic data of runoff and debris flow; constructing a double-layer depth average model of runoff and debris flow propagation based on the water absorption rate parameter of debris flow to runoff; spatially discretizing the double-layer depth average model to obtain a dynamic model; simulating the behavior and interaction process of runoff and debris flow. Although the process of rainfall, vegetation interception, soil infiltration, runoff generation and debris flow propagation is considered, a deep average double-layer model describing the dynamics of runoff and debris flow is proposed, and the water absorption rate parameter is introduced, and the behavior and interaction process of runoff and debris flow are accurately and effectively simulated using the dynamic model; however, the temporal and spatial resolution is low, resulting in large errors.

[0006] At present, the existing technologies 1, 2 and 3 have the problems of low temporal and spatial resolution and insufficient uncertainty analysis. To solve the above problems, the present invention provides a method and system for analyzing the spatial distribution pattern of rainfall in low-order small watersheds in mountainous areas. Summary of the invention

[0007] The main purpose of the present invention is to provide a method and system for analyzing the spatial distribution pattern of rainfall in low-order small watersheds in mountainous areas, so as to solve the problems of low temporal and spatial resolution and insufficient uncertainty analysis in the prior art.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas, comprising the following steps:

[0010] Obtain historical rainfall data from at least one rainfall monitoring station, merge them according to different time scales such as month, week, day and hour, and use Pearson correlation coefficient to evaluate the correlation of rainfall data from at least one rainfall monitoring station on different time scales;

[0011] Based on the evaluation results, a data set is formed, the data set is preprocessed, and the preprocessed data set is divided into a training set, a validation set, and a test set; a rainfall monitoring spatial model is established, and the rainfall monitoring spatial model is trained using the training set;

[0012] The rainfall data is input into the rainfall monitoring spatial model for simulation, the target area of ​​the rainfall data is divided according to the simulation results, and the division results are uploaded to the cloud platform; the real-time data is compared with the predicted data, and the rainfall monitoring spatial model is adjusted according to the comparison results.

[0013] As a further improvement of the present invention, the process of using the Pearson correlation coefficient to evaluate the correlation of rainfall data of two monitoring points at different time scales includes the following steps:

[0014] Obtain historical rainfall data from at least one rainfall monitoring station; perform preprocessing operations on the historical rainfall data to remove outliers and fill missing values; merge the historical rainfall data according to different time scales such as month, week, day and hour;

[0015] Gamma distribution is used to analyze the spatial variation characteristics of daily rainfall; and the spatial variation characteristics of hourly rainfall are analyzed by dividing independent rainfall events; independent rainfall events are identified by combining rainfall characteristics and lithological conditions;

[0016] Based on the identification results, Pearson was used to evaluate the identification results and the correlation of rainfall data of at least one rainfall monitoring station on different time scales was evaluated.

[0017] As a further improvement of the present invention, the process of merging historical rainfall data according to different time scales such as month, week, day and hour includes the following steps:

[0018] Delete the historical rainfall data with negative rainfall values ​​or data above the preset maximum rainfall threshold; fill in the data based on the average value of the previous and next rainfall data;

[0019] Divide the historical rainfall data into hours, and divide it into 24 hours to get the daily rainfall; add up the rainfall of 7 days to get the weekly rainfall; add up the rainfall of 30 days to get the total rainfall of each month;

[0020] The hourly, daily, weekly and monthly rainfall is transmitted to the cloud platform server, which stores it and performs visualization.

[0021] As a further improvement of the present invention, the process of identifying independent rainfall events includes the following steps:

[0022] The spatial variation characteristics of daily rainfall are calculated by fitting the daily rainfall input into the gamma distribution, and the estimated values ​​of shape parameters and scale parameters are obtained;

[0023] Among them, the spatial characteristic change of daily rainfall is calculated as:

[0024] Let x be the random rainfall in a certain period of time, and F(x) be the probability density function;

[0025]

[0026] Among them, α is the shape parameter and β is the scale parameter; Γ(α) is the gamma function, whose shape is controlled by α and β; when the shape parameter α<1, the gamma distribution is left-skewed, when α>1, the gamma distribution is right-skewed, and when α=1, the gamma distribution degenerates into an exponential distribution. The larger the scale parameter β, the narrower the gamma distribution range and the steeper the image.

[0027] Define the threshold of the interval between rainfall events; classify rainfall events, and obtain the temporal and spatial distribution of rainfall events by analyzing the rainfall distribution in different time periods;

[0028] Independent rainfall events are identified based on rainfall characteristics and lithological conditions; the joint probability density distribution function of rainfall intensity and duration is fitted, and the time interval distribution function of rainfall events is described in combination with the Poisson process to calibrate the interval time of rainfall events.

[0029] As a further improvement of the present invention, the process of evaluating the correlation of rainfall data of at least one rainfall monitoring station on different time scales comprises the following steps:

[0030] Extract the rainfall data characteristics of the hour, day, week and month, and measure the rainfall data between different time periods based on the rainfall data characteristics of the hour, day, week and month;

[0031] Among them, the rainfall data between different time periods are:

[0032]

[0033] Among them, x i and i are the values ​​of the two variables at the i-th time point, and is the average of these two variables;

[0034] The Pearson correlation coefficient is checked. If the significance level is less than the preset significance threshold, it means that the Pearson correlation coefficient is significant.

[0035] The rainfall data on different time scales and the relationship characteristics of the Pearson correlation on different time scales are transmitted to the cloud platform, and the cloud platform draws the rainfall data on different time scales and the relationship characteristics of the Pearson correlation on different time scales and displays them in a visual form.

[0036] As a further improvement of the present invention, the process of extracting the rainfall data features of the hour, day, week and month includes the following steps:

[0037] Perform rainfall trend analysis on historical rainfall data to obtain rainfall trend data, perform multi-scale wavelet transform on rainfall trend data, extract rainfall characteristics at different time scales, and obtain multi-scale rainfall characteristic data;

[0038] Based on rainfall trends and multi-humidity rainfall characteristics data, the probability distribution of extreme rainfall events is calculated to obtain the probability distribution of extreme rainfall events; based on the probability distribution of rainfall events, multi-scale rainfall characteristics and rainfall trend data, rainfall analysis data is obtained;

[0039] Among them, rainfall distribution data include rainfall intensity, rainfall duration and rainfall interval period;

[0040] Based on rainfall analysis data, rainfall data between different time periods are measured; and the data between different time periods are spatially divided.

[0041] As a further improvement of the present invention, the process of obtaining the probability distribution of extreme rainfall events comprises the following steps:

[0042] The rainfall intensity is randomly sampled to determine the transfer path probability of rainfall of different intensities; the rainfall duration is analyzed and mined to obtain the hidden duration state probability distribution;

[0043] Extract the associated features of the rainfall interval period, construct the conditional probability relationship between the rainfall interval period, rainfall intensity and duration, and obtain the interval period conditional relationship; combine the data of the transfer path probability, the hidden duration state probability distribution and the interval period conditional probability relationship to obtain the rainfall combination data;

[0044] Construct rainfall probability density data based on rainfall combination data; adjust the probability distribution of rainfall probability density data to obtain adjusted probability density data; obtain the probability distribution of future rainfall and future predicted rainfall based on the adjusted probability density data and the probability of rainfall events; and perform spatial division based on the future rainfall probability distribution and future predicted rainfall.

[0045] As a further improvement of the present invention, the process of training the rainfall monitoring spatial model using the training set includes the following steps:

[0046] The rainfall data at different time scales and the relationship between Pearson correlations at different scales are preprocessed by missing value and outlier monitoring and data standardization to obtain a data set; the data set is divided into a training set, a test set and a validation set;

[0047] A rainfall monitoring spatial model is constructed based on the rainfall data and lithologic deformation data of the target area, and the rainfall monitoring spatial model is iteratively trained based on the lithologic deformation data, rainfall time series characteristics, and rainfall analysis data until the number of training iterations reaches a preset number of training iterations, and the rainfall monitoring spatial model is completed;

[0048] The validation set is used to validate the rainfall monitoring spatial model. If the preset validation threshold is reached, the final rainfall monitoring spatial model is obtained.

[0049] As a further improvement of the present invention, the lithology-rainfall coupling loss function in the model training formula system is expressed as:

[0050]

[0051] In the formula, S m represents the displacement of the mth deformation monitoring point; H 1 represents the Sobolev space norm; TV represents the total variation regularization term; ∑J θ represents the rainfall covariance matrix;

[0052] The spatiotemporal feature fusion equation is expressed as:

[0053]

[0054] In the formula, represents the three-dimensional space-time convolution operation; γ represents the vertical attenuation coefficient; z c represents the base depth of feature channel c;

[0055] The adaptive momentum update rule is expressed as:

[0056]

[0057] Where η k represents the dynamic learning rate; τ represents the decay constant in the training phase; ⊙ represents the Hadamard product.

[0058] To achieve the above object, the present invention also provides the following technical solutions:

[0059] A system for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas, which is applied to the method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas, and the system for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas comprises:

[0060] A module for obtaining historical rainfall data is used to obtain historical rainfall data of at least one rainfall monitoring station, merge the historical rainfall data according to different time scales of month, week, day and hour, and use the Pearson correlation coefficient to evaluate the correlation of rainfall data of at least one rainfall monitoring point at different time scales;

[0061] Construct a spatial model module to form a data set based on the evaluation results, preprocess the data set, and divide the preprocessed data set into a training set, a validation set, and a test set; establish a rainfall monitoring spatial model, and use the training set to train the rainfall monitoring spatial model;

[0062] The real-time adjustment and verification module is used to input rainfall data into the rainfall monitoring spatial model for simulation, divide the target area of ​​rainfall data according to the simulation results, and upload the division results to the cloud platform; compare the real-time data with the predicted data, and adjust the rainfall monitoring spatial model according to the comparison results.

[0063] The present invention obtains historical rainfall data from at least one rainfall monitoring station and merges them according to different time scales such as month, week, day and hour, and uses the Pearson correlation coefficient to evaluate the correlation of rainfall data at different time scales; through data merging and correlation evaluation at multiple time scales, the spatiotemporal variation characteristics of rainfall data can be more comprehensively understood. Through the division of data sets and model training, a spatial model that can accurately predict rainfall can be constructed to improve the generalization ability and prediction accuracy of the model; through the comparison of real-time data and predicted data and model adjustment, the model performance can be continuously optimized to improve the accuracy and timeliness of rainfall prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A schematic diagram of the steps of an embodiment of a method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas of the present invention;

[0065] Figure 2 A schematic flow chart of an embodiment of a method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas of the present invention and steps of using the Pearson correlation coefficient to evaluate the correlation of rainfall data of two monitoring points at different time scales;

[0066] Figure 3 This is a schematic flow chart of the steps of merging historical rainfall data according to different time scales, such as month, week, day and hour, in one embodiment of the method for analyzing spatial distribution pattern of rainfall in low-order small watersheds in mountainous areas of the present invention;

[0067] Figure 4 A schematic flow chart of steps for identifying independent rainfall events in an embodiment of a method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas of the present invention;

[0068] Figure 5 A schematic flow chart of the steps of evaluating the correlation of rainfall data of at least one rainfall monitoring station at different time scales in one embodiment of the method for analyzing spatial distribution pattern of rainfall in low-order small watersheds in mountainous areas of the present invention;

[0069] Figure 6A schematic flow chart of the steps of extracting hourly, daily, weekly and monthly rainfall data characteristics in one embodiment of a method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas of the present invention;

[0070] Figure 7 A schematic flow chart of the steps of obtaining the probability distribution of extreme rainfall events in one embodiment of a method for analyzing the spatial distribution pattern of rainfall in a low-order small watershed in a mountainous area of ​​the present invention;

[0071] Figure 8 A schematic flow chart of the steps of using a training set to train a rainfall monitoring spatial model in an embodiment of a method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas of the present invention;

[0072] Fig. 9 It is a schematic flow chart of the steps of adjusting the rainfall monitoring spatial model according to the comparison results in one embodiment of the method for analyzing the spatial distribution pattern of rainfall in a low-order small watershed in a mountainous area of ​​the present invention;

[0073] Fig.10 It is a functional module schematic diagram of an embodiment of a system for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas of the present invention;

[0074] Fig.11 It is a structural schematic diagram of an embodiment of an electronic device of the present invention;

[0075] Fig.12 It is a schematic structural diagram of an embodiment of the storage medium of the present invention. DETAILED DESCRIPTION

[0076] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0077] The terms "first", "second" and "third" in the present invention are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined as "first", "second" and "third" can explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. All directional indications (such as up, down, left, right, front, back...) in the embodiments of the present invention are only used to explain the relative position relationship, movement, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication also changes accordingly. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally also includes steps or units that are not listed, or optionally also includes other steps or units inherent to these processes, methods, products or devices.

[0078] Reference to an "embodiment" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiment may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0079] like Figure 1 As shown, this embodiment provides an embodiment of a method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas. In this embodiment, the method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas specifically includes the following steps:

[0080] Step S1: Obtain historical rainfall data of at least one rainfall monitoring station, merge them according to different time scales of month, week, day and hour, and use Pearson correlation coefficient to evaluate the correlation of rainfall data of at least one rainfall monitoring point at different time scales;

[0081] Step S2: forming a data set based on the evaluation results, preprocessing the data set, and dividing the preprocessed data set into a training set, a validation set, and a test set; establishing a rainfall monitoring spatial model, and using the training set to train the rainfall monitoring spatial model;

[0082] Step S3: Input rainfall data into the rainfall monitoring spatial model for simulation, divide the rainfall data target area according to the simulation results, and upload the division results to the cloud platform; compare the real-time data with the predicted data, and adjust the rainfall monitoring spatial model according to the comparison results.

[0083] Preferably, in step S1 of this embodiment, historical rainfall data of at least one rainfall monitoring station is obtained, and the data are merged according to different time scales of month, week, day and hour, and the correlation of rainfall data at different time scales is evaluated using the Pearson correlation coefficient; step S1 can more comprehensively understand the spatiotemporal variation characteristics of rainfall data through data merging and correlation evaluation at multiple time scales. Step S2 forms a data set based on the evaluation results, preprocesses the data set, and divides the preprocessed data set into a training set, a validation set and a test set; establishes a rainfall monitoring spatial model, and trains the rainfall monitoring spatial model using the training set; step S2 can construct a spatial model that can accurately predict rainfall through the division of the data set and model training, thereby improving the generalization ability and prediction accuracy of the model. Step S3 inputs rainfall data into the rainfall monitoring spatial model for simulation, divides the target area of ​​rainfall data according to the simulation results, and uploads the division results to the cloud platform; compares real-time data with predicted data, and adjusts the rainfall monitoring spatial model according to the comparison results. Step S3 can continuously optimize the model performance and improve the accuracy and timeliness of rainfall prediction through the comparison of real-time data with predicted data and model adjustment.

[0084] Furthermore, if Figure 2 As shown in step S1, the process of using the Pearson correlation coefficient to evaluate the correlation of rainfall data of two monitoring points at different time scales specifically includes the following steps:

[0085] Step S11: acquiring historical rainfall data from at least one rainfall monitoring station; performing preprocessing operations such as removing outliers and filling missing values ​​on the historical rainfall data; merging the historical rainfall data according to different time scales such as month, week, day and hour;

[0086] Step S12: Analyze the spatial variation characteristics of daily rainfall using gamma distribution; and analyze the spatial variation characteristics of hourly rainfall by dividing independent rainfall events; and identify independent rainfall events in combination with rainfall characteristics and lithological conditions;

[0087] Step S13: Based on the recognition results, Pearson is used to evaluate the recognition results and to evaluate the correlation of rainfall data of at least one rainfall monitoring station at different time scales.

[0088] Preferably, step S11 of this embodiment obtains historical rainfall data from at least one rainfall monitoring station, pre-processes the data (such as removing outliers and filling missing values), and then merges the data by month, week, day and different time scales. This embodiment improves the quality and availability of data and provides accurate data support for the spatial and temporal variation analysis of rainfall. Step S12 uses gamma distribution to analyze the spatial variation characteristics of daily rainfall, and analyzes the spatial variation characteristics of hourly rainfall by dividing independent rainfall events. Independent rainfall events are identified in combination with rainfall characteristics and lithological conditions. This embodiment uses statistical methods and geographic information system (GIS) tools to reveal the spatial distribution law and variation characteristics of rainfall. The technical effect is to enhance the understanding of spatial variation of rainfall and provide a scientific basis for regional rainfall monitoring and disaster warning. Step S13 uses the Pearson correlation coefficient to evaluate the recognition results based on the recognition results, and evaluates the correlation of rainfall data of at least one rainfall monitoring station at different time scales. This embodiment evaluates the correlation of rainfall data at different time scales through statistical analysis methods, providing a quantitative basis for the study of spatial and temporal variation of rainfall. The technical effect is to improve the accuracy of correlation analysis of rainfall data and provide scientific support for rainfall prediction and management.

[0089] Furthermore, if Figure 3 As shown, the process of merging historical rainfall data according to different time scales of month, week, day and hour in step S11 specifically includes the following steps:

[0090] Step S111: Delete the historical rainfall data with negative rainfall values ​​or data exceeding the preset maximum rainfall threshold; fill in the data according to the average value of the previous and next rainfall data;

[0091] Step S112: divide the historical rainfall data into hours, and obtain the daily rainfall by 24 hours; add up the rainfall of 7 days to obtain the weekly rainfall; add up the rainfall of 30 days to obtain the total rainfall of each month;

[0092] Step S113: Transmitting the hourly, daily, weekly and monthly rainfall to the cloud platform server, which stores and visualizes it.

[0093] Preferably, in step S111 of this embodiment, abnormal values ​​(such as negative values ​​or data above the preset maximum rainfall threshold) in the historical rainfall data are deleted and filled according to the average value of the previous and next rainfall data. Step S112 divides the historical rainfall data by hours, days, weeks and months, and calculates daily rainfall, weekly rainfall and monthly rainfall respectively; converts continuous rainfall data into rainfall at different time scales to facilitate subsequent statistical analysis and prediction. Step S113 transmits hourly, daily, weekly and monthly rainfall to the cloud platform server for storage and visualization; realizes centralized management and visualization of data, which is convenient for users to view and analyze rainfall data.

[0094] Furthermore, if Figure 4 As shown, the process of identifying independent rainfall events in step S12 specifically includes the following steps:

[0095] Step S121: fitting the daily rainfall input into the gamma distribution to calculate the spatial variation characteristics of the daily rainfall, and obtain the estimated values ​​of the shape parameter and the scale parameter;

[0096] Among them, the spatial characteristic change of daily rainfall is calculated as:

[0097] Let x be the random rainfall in a certain period of time, and F(x) be the probability density function;

[0098]

[0099] Among them, α is the shape parameter and β is the scale parameter; Γ(α) is the gamma function, whose shape is controlled by α and β; when the shape parameter α<1, the gamma distribution is left-skewed, when α>1, the gamma distribution is right-skewed, and when α=1, the gamma distribution degenerates into an exponential distribution. The larger the scale parameter β, the narrower the gamma distribution range and the steeper the image.

[0100] Step S122: defining a threshold value of the interval time of rainfall events; classifying rainfall events, and obtaining the temporal and spatial distribution law of rainfall events by analyzing the rainfall distribution in different time periods;

[0101] Step S123: Identify independent rainfall events according to rainfall characteristics and lithological conditions; fit the joint probability density distribution function of rainfall intensity and duration, and combine the time interval distribution function of rainfall events described by the Poisson process to calibrate the interval time of rainfall events.

[0102] Preferably, step S121 of this embodiment uses the gamma distribution model to perform statistical modeling on daily rainfall, and estimates the shape parameters and scale parameters by the maximum likelihood estimation method (MLE) or the L-moments method; it can accurately describe the probability distribution characteristics of daily rainfall, and provide basic data for subsequent spatial change analysis. Step S122 divides the continuous rainfall process into different sessions by setting the rainfall event interval threshold, and analyzes the rainfall distribution of each session; it reveals the spatiotemporal distribution law of rainfall events, and provides a basis for the classification and feature extraction of rainfall events. Step S123 combines the Poisson process and the joint probability density distribution function to describe the time interval distribution of rainfall events, and extracts the joint distribution of rainfall intensity and duration by identifying independent rainfall events; it can more accurately describe the time and intensity characteristics of rainfall events, and provide support for rainfall simulation and risk assessment.

[0103] Furthermore, if Figure 5 As shown, the process of evaluating the correlation of rainfall data of at least one rainfall monitoring station at different time scales in step S13 specifically includes the following steps:

[0104] Step S131: extracting rainfall data features of the hour, day, week and month, and measuring rainfall data between different time periods based on the rainfall data features of the hour, day, week and month;

[0105] Among them, the rainfall data between different time periods are:

[0106]

[0107] Among them, x i and i are the values ​​of the two variables at the i-th time point, and is the average of these two variables;

[0108] Step S132: Check the Pearson correlation coefficient. If the significance level is less than the preset significance threshold, it means that the Pearson correlation coefficient has significant correlation.

[0109] Step S133: The rainfall data on different time scales and the relationship characteristics of the Pearson correlation on different time scales are transmitted to the cloud platform, and the cloud platform draws the rainfall data on different time scales and the relationship characteristics of the Pearson correlation on different time scales and displays them in a visual form.

[0110] Preferably, in step S131 of this embodiment, the rainfall data features of the hour, day, week and month are extracted, and the rainfall data between different time periods are measured based on these features; by extracting data features at multiple time scales, the changing trend and periodic characteristics of rainfall can be analyzed more comprehensively, providing basic data support for subsequent statistical analysis and visualization. Step S132 verifies the Pearson correlation coefficient, and the significant level value is less than the preset correlation significant threshold; ensures that the Pearson correlation coefficient is statistically significant, thereby ensuring the reliability and effectiveness of the correlation analysis, and avoiding misjudgment or omission of correlation. Step S133 transmits the relationship features of rainfall data on different time scales and Pearson correlations on different time scales to the cloud platform, and the cloud platform draws the relationship features of rainfall data on different time scales and Pearson correlations on different time scales, and displays them in a visual form; this embodiment uses the visualization tool of the cloud platform to display complex data relationships in an intuitive form, which is convenient for researchers to quickly understand the spatiotemporal variation laws and correlation features of rainfall data, and improve the efficiency and accuracy of data analysis.

[0111] Furthermore, if Figure 6 As shown, the process of extracting the rainfall data features of the hour, day, week and month in step S131 specifically includes the following steps:

[0112] Step S1321: performing rainfall trend analysis on historical rainfall data to obtain rainfall trend data, performing multi-scale wavelet transform on the rainfall trend data to extract rainfall characteristics at different time scales to obtain multi-scale rainfall characteristic data;

[0113] Step S1322: Calculate the probability distribution of extreme rainfall events based on rainfall trends and multi-humidity rainfall characteristic data to obtain the probability distribution of extreme rainfall events; obtain rainfall analysis data based on the calculated probability distribution of rainfall events, multi-scale rainfall characteristics and rainfall trend data;

[0114] Among them, rainfall distribution data include rainfall intensity, rainfall duration and rainfall interval period;

[0115] Step S1323: Measure the rainfall data between different time periods based on the rainfall analysis data; and spatially divide the data between different time periods.

[0116] Preferably, in step S1321 of this example, by performing trend analysis on historical rainfall data, rainfall trend data is extracted, and rainfall characteristics at different time scales are extracted using multi-scale wavelet transform; the long-term change trend and periodic characteristics of rainfall data can be revealed, and basic data support can be provided for subsequent rainfall analysis. Step S1322 calculates the probability distribution of extreme rainfall events based on rainfall trend and multi-scale rainfall characteristic data, and generates rainfall analysis data in combination with characteristics such as rainfall intensity, duration and interval period; the probability of occurrence of extreme rainfall events can be accurately assessed, and a scientific basis can be provided for flood risk management and disaster prevention and mitigation. Step S1323 measures the rainfall differences between different time periods based on rainfall analysis data, and spatially divides rainfall data; the rainfall differences between different time periods and regions can be revealed, and support can be provided for regional rainfall management and planning.

[0117] Furthermore, if Figure 7 As shown, the process of obtaining the probability distribution of extreme rainfall events in step S1322 specifically includes the following steps:

[0118] Step S13221: Randomly sample the rainfall intensity to determine the transfer path probability of rainfall of different intensities; perform state hidden analysis and mining on the rainfall duration to obtain the hidden duration state probability distribution;

[0119] Step S13222: extracting the associated features of the rainfall interval period, constructing the conditional probability relationship of the rainfall interval period, rainfall intensity and duration, and obtaining the interval period conditional relationship; combining the data of the transfer path probability, the hidden duration state probability distribution and the interval period conditional probability relationship, and obtaining the rainfall combination data;

[0120] Step S13223: construct rainfall probability density data based on rainfall combination data; adjust the probability distribution of rainfall probability density data to obtain adjusted probability density data, and obtain the probability distribution of future rainfall and future predicted rainfall based on the adjusted probability density data and the probability of occurrence of rainfall events; perform spatial division based on the future rainfall probability distribution and future predicted rainfall.

[0121] Preferably, in step S13221 of this embodiment, the rainfall intensity is randomly sampled to determine the transfer path probability of rainfall of different intensities, and the state hidden analysis and mining of the rainfall duration are performed to obtain the hidden duration state probability distribution. This embodiment utilizes the probability distribution of rainfall events and the calculation method of the transfer path probability, which can effectively capture the random characteristics of rainfall intensity and duration, thereby improving the prediction accuracy of rainfall patterns. Step S13222 extracts the associated features of the rainfall interval period, constructs the conditional probability relationship of the rainfall interval period, rainfall intensity and duration, and combines the data of the transfer path probability, the hidden duration state probability distribution and the interval period conditional probability relationship to obtain rainfall combination data. This embodiment can more comprehensively describe the complexity and diversity of rainfall events through the extraction of associated features and the construction of conditional probability relationships, thereby improving the comprehensive analysis capability of rainfall patterns. Step S13223 constructs rainfall probability density data based on the rainfall combination data, and adjusts the probability distribution of the rainfall probability density data to obtain the adjusted probability density data. Based on the adjusted probability density data and the probability of rainfall event occurrence, the probability distribution of future rainfall and the future predicted rainfall are obtained. Finally, spatial division is performed based on the probability distribution of future rainfall and the future predicted rainfall. This embodiment can more accurately predict the probability distribution and rainfall of future rainfall through the construction and adjustment of probability density data, thereby providing a scientific basis for flood prevention and disaster reduction.

[0122] Furthermore, if Figure 8 As shown, the process of using the training set to train the rainfall monitoring spatial model in step S2 specifically includes the following steps:

[0123] Step S21: Preprocess the rainfall data at different time scales and the Pearson correlation relationship at different scales by performing missing value and outlier monitoring and data standardization to obtain a data set; divide the data set into a training set, a test set, and a validation set;

[0124] Step S22: constructing a rainfall monitoring spatial model based on the rainfall data and lithologic deformation data of the target area, and iteratively training the rainfall monitoring spatial model based on the lithologic deformation data, rainfall time series characteristics, and rainfall analysis data until the number of training iterations reaches a preset number of training iterations, thereby completing the rainfall monitoring spatial model;

[0125] Step S23: Use the verification set to verify the rainfall monitoring spatial model. If the preset verification threshold is reached, the final rainfall monitoring spatial model is obtained.

[0126] Among them, in step S21, the data preprocessing formula system, the multi-scale Pearson coupling correction factor is expressed as:

[0127]

[0128] Rhodopsin, Ψ represents the spatiotemporal coupling correlation coefficient matrix; ω τ represents the time scale attenuation weight factor; R t,i' represents the rainfall observation value of the i'th station at time scale t; μ, σ represent the mean and standard deviation of the corresponding time scale;

[0129] The sliding window outlier discrimination criterion is expressed as:

[0130]

[0131] When D w >Q 3 +1.5IQR+k·exp(-α·w) is considered an outlier; k represents the terrain complexity correction factor (k=1+0.1·DEM / 100); represents the second-order time derivative of rainfall in window w;

[0132] The multimodal normalization equation is expressed as:

[0133]

[0134] Where p represents the spatial grid index (p∈{1,...,P}); represents the terrain gradient modulus of the grid p; β represents the lithology permeability coefficient (β = 0.05·Ksat / 100);

[0135] Step S22: Model training formula system, lithology-rainfall coupling loss function is expressed as:

[0136]

[0137] In the formula, S m represents the displacement of the mth deformation monitoring point; H 1 represents the Sobolev space norm; TV represents the total variation regularization term; ∑j θ represents the rainfall covariance matrix;

[0138] The spatiotemporal feature fusion equation is expressed as:

[0139]

[0140] In the formula, represents the three-dimensional space-time convolution operation; γ represents the vertical attenuation coefficient; z c represents the base depth of feature channel c;

[0141] The adaptive momentum update rule is expressed as:

[0142]

[0143] Where η k represents the dynamic learning rate; τ represents the attenuation constant in the training phase; ⊙ represents the Hadamard product;

[0144] Step S23 verifies the formula system, and the multi-index fusion verification function is expressed as:

[0145]

[0146] RMSE norm =RSME / (R max -R min )

[0147]

[0148] Where, CSI represents the critical success index; POD represents the hit rate, and FAR represents the false alarm rate;

[0149] The space-time continuity constraint is expressed as:

[0150]

[0151] Where Ω represents the watershed spatial domain; ‖·‖ F represents the Frobenius norm; ∈ topo represents the terrain curvature tolerance threshold; A represents the watershed area; the spatiotemporal index represents t (time), k (time scale), x, y, z (spatial coordinates); the hydrological parameters represent R (precipitation), S (shape variable), G (terrain gradient); the model parameters represent θ (network weight), W (convolution kernel), λ (regularization coefficient); the operator represents (tensor product), ⊙(element product), (3D convolution); statistical indicators represent Q3 (third quartile) and IQR (interquartile range).

[0152] Preferably, step S21 of this embodiment ensures the quality and consistency of rainfall data through preprocessing methods such as missing value and outlier monitoring and data standardization, thereby providing a reliable data basis for subsequent model training. This embodiment plays a role in improving data quality and accuracy, and ensures the reliability and effectiveness of model training. Step S22 constructs a rainfall monitoring spatial model based on the rainfall data and lithologic deformation data of the target area, and iteratively trains the model through lithologic deformation data, rainfall time series characteristics, and rainfall analysis data until a preset number of training iterations is reached. This embodiment optimizes model performance and improves prediction accuracy, ensuring the generalization ability and robustness of the model under different conditions. Step S23 uses a validation set to verify the rainfall monitoring spatial model to ensure that the model reaches a preset validation threshold. This embodiment evaluates model performance and ensures model reliability, ensuring the effectiveness and accuracy of the final model in practical applications.

[0153] Furthermore, if Fig. 9 As shown, the process of adjusting the rainfall monitoring spatial model according to the comparison result in step S3 specifically includes the following steps:

[0154] Step S31: inputting rainfall data into the rainfall monitoring space model for simulation, and dividing the target area into severe landslide, landslide and mild landslide grades in combination with lithology data; uploading the division results to the cloud platform;

[0155] Step S32: Real-time monitoring of the target area is performed, and the real-time monitoring data is compared with the model simulation results. If it is greater than a preset comparison threshold, the rainfall monitoring spatial model is adjusted according to the real-time monitoring data; if it is less than, the simulation is continued;

[0156] Step S33: Establish a real-time warning system and adjust the spatial division in real time according to the prediction results; if the rainfall monitoring spatial model detects rainfall data greater than the preset warning value, an early warning will be immediately issued to the relevant departments.

[0157] Among them, step S31 landslide grade classification formula system, lithology-rainfall coupling risk discriminant formula:

[0158]

[0159] In the formula, represents the landslide risk index of grid p; represents the piecewise function fusion operator ( , otherwise perform multiplication); ρ p Indicates rock density (kg / m 3 );S crit represents the critical deformation displacement (mm); J F represents the Jacobi matrix eigenvalue spectrum;

[0160] Dynamic boundary equation for grade division:

[0161]

[0162] In the formula, Ω m Indicates the mth level landslide risk area (m=1,2,3 corresponds to severe / moderate / mild); Ω m represents the quantile threshold (Q1=0.95, Q2=0.75, Q3=0.5); Represents the Laplacian operator of terrain curvature;

[0163] Cloud platform data compression encoding representation:

[0164]

[0165] In the formula, Represents the cloud platform storage accuracy control parameter; Φ compress represents a compressed hash function;

[0166] Step S32: Model dynamically adjusts the formula system, and the real-time data-model difference is expressed as:

[0167]

[0168] Where erf is the error function; ε is the numerical stability constant; σ S represents the standard deviation of deformation monitoring noise;

[0169] The parameter online correction equation is expressed as:

[0170]

[0171] Where K represents the sensitivity gain coefficient; t last Indicates the last modification timestamp;

[0172] Step S33: Early warning system formula system, rainstorm-landslide coupling early warning index is expressed as:

[0173]

[0174] Where erfc represents the residual error function; t sat represents the soil water saturation time; σ sat represents the time diffusion coefficient of the saturation process.

[0175] Preferably, in step S31 of this embodiment, rainfall data is input into the rainfall monitoring space model for simulation, and the target area is divided into landslide grades in combination with lithology data, and the results are uploaded to the cloud platform; by combining rainfall data and lithology data, landslide risk assessment and spatial division of the target area are realized, and the accuracy and timeliness of landslide warning are improved. Step S32 monitors the target area in real time, compares the real-time monitoring data with the model simulation results, and adjusts the model according to the comparison results. This embodiment dynamically adjusts the rainfall monitoring space model through real-time monitoring and model comparison to ensure the real-time and accuracy of the model and improve the response speed and reliability of the early warning system. Step S33 establishes a real-time early warning system, adjusts the spatial division in real time according to the prediction results, and issues an early warning immediately when rainfall data greater than the preset early warning value is monitored; Step S33 issues an early warning to relevant departments in a timely manner through the real-time early warning system to reduce the losses caused by landslide disasters and improve the efficiency and effectiveness of emergency response.

[0176] like Fig.10As shown, this embodiment also provides a system for analyzing spatial distribution patterns of rainfall in small low-order watersheds in mountainous areas. In this embodiment, the system for analyzing spatial distribution patterns of rainfall in small low-order watersheds in mountainous areas is applied to the method for analyzing spatial distribution patterns of rainfall in small low-order watersheds in mountainous areas in the above-mentioned embodiment. The system for analyzing spatial distribution patterns of rainfall in small low-order watersheds in mountainous areas comprises:

[0177] The module 1 for obtaining historical rainfall data is used to obtain historical rainfall data of at least one rainfall monitoring station, merge the historical rainfall data according to different time scales of month, week, day and hour, and use the Pearson correlation coefficient to evaluate the correlation of rainfall data of at least one rainfall monitoring point at different time scales;

[0178] Constructing a spatial model module 2, which is used to form a data set based on the evaluation results, preprocess the data set, and divide the preprocessed data set into a training set, a validation set, and a test set; establishing a rainfall monitoring spatial model, and using the training set to train the rainfall monitoring spatial model;

[0179] The real-time adjustment verification module 3 is used to input rainfall data into the rainfall monitoring space model for simulation, divide the rainfall data target area according to the simulation results, and upload the division results to the cloud platform; compare the real-time data with the predicted data, and adjust the rainfall monitoring space model according to the comparison results.

[0180] Preferably, the module 1 for obtaining historical rainfall data in this embodiment obtains historical rainfall data of at least one rainfall monitoring station through a web crawler or database query, and merges the data at different time scales of month, week, day and hour; ensures the comprehensiveness of the data and the diversity of time resolution, and provides basic data support for subsequent analysis. The module 2 for constructing a spatial model evaluates the correlation of rainfall data at different time scales based on the Pearson correlation coefficient, forms a data set, preprocesses the data set, and divides the preprocessed data set into a training set, a validation set and a test set; this embodiment improves the accuracy and generalization ability of the model through correlation analysis and data preprocessing, and provides a reliable data basis for establishing a rainfall monitoring spatial model. The real-time adjustment and verification module 3 inputs the real-time rainfall data into the rainfall monitoring spatial model for simulation, divides the target area of ​​the rainfall data according to the simulation results, and uploads the division results to the cloud platform; compares the real-time data with the predicted data, and adjusts the rainfall monitoring spatial model according to the comparison results. This embodiment realizes real-time monitoring and dynamic adjustment of rainfall data, improves the real-time and accuracy of the model, and provides timely and effective decision support for disaster prevention and mitigation.

[0181] like Fig.11 As shown, this embodiment provides an embodiment of an electronic device. In this embodiment, the electronic device 4 includes a processor 41 and a memory 42 coupled thereto.

[0182] The memory 42 stores program instructions for implementing the method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas according to any of the above embodiments.

[0183] The processor 41 is used to execute the program instructions stored in the memory 42 to lay out the analysis method of the spatial distribution pattern of rainfall in low-order small watersheds in mountainous areas.

[0184] The processor 41 may also be referred to as a CPU (Central Processing Unit). The processor 41 may be an integrated circuit chip having signal processing capabilities. The processor 41 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0185] Further, Fig.12 The schematic diagram of the structure of the storage medium of an embodiment of the present application is that the storage medium 5 of the embodiment of the present application stores program instructions 51 that can implement all the above methods, wherein the program instructions 51 can be stored in the above storage medium in the form of a software product, including several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) or a processor to execute all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or terminal devices such as a computer, a server, a mobile phone, and a tablet.

[0186] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0187] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of software functional units. The above is only an implementation mode of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly used in other related technical fields, is also included in the patent protection scope of the present invention.

[0188] The specific implementation methods of the invention are described in detail above, but they are only examples, and the invention is not limited to the specific implementation methods described above. For those skilled in the art, any equivalent modification or substitution of the invention is also within the scope of the invention, therefore, the equalization, modification, improvement, etc. made without departing from the spirit and principle of the invention should be included in the scope of the invention.

Claims

1. A method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas, characterized in that: The method for analyzing the spatial distribution pattern of rainfall in low-order small watersheds in mountainous areas comprises the following steps: Obtain historical rainfall data from at least one rainfall monitoring station, merge them according to different time scales such as month, week, day and hour, and use Pearson correlation coefficient to evaluate the correlation of rainfall data from at least one rainfall monitoring station on different time scales; Based on the evaluation results, a data set is formed, the data set is preprocessed, and the preprocessed data set is divided into a training set, a validation set, and a test set; a rainfall monitoring spatial model is established, and the rainfall monitoring spatial model is trained using the training set; The rainfall data is input into the rainfall monitoring spatial model for simulation, the target area of ​​the rainfall data is divided according to the simulation results, and the division results are uploaded to the cloud platform; the real-time data is compared with the predicted data, and the rainfall monitoring spatial model is adjusted according to the comparison results.

2. The method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas according to claim 1 is characterized in that: The process of using the Pearson correlation coefficient to evaluate the correlation of rainfall data at two monitoring points at different time scales includes the following steps: Obtain historical rainfall data from at least one rainfall monitoring station; perform preprocessing operations on the historical rainfall data to remove outliers and fill missing values; merge the historical rainfall data according to different time scales such as month, week, day and hour; Gamma distribution is used to analyze the spatial variation characteristics of daily rainfall; and the spatial variation characteristics of hourly rainfall are analyzed by dividing independent rainfall events; independent rainfall events are identified by combining rainfall characteristics and lithological conditions; Based on the identification results, Pearson was used to evaluate the identification results and the correlation of rainfall data of at least one rainfall monitoring station on different time scales was evaluated.

3. The method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas according to claim 1 is characterized in that: The process of combining historical rainfall data at different time scales, such as monthly, weekly, daily and instantaneous, includes the following steps: Delete the historical rainfall data with negative rainfall values ​​or data above the preset maximum rainfall threshold; fill in the data based on the average value of the previous and next rainfall data; Divide the historical rainfall data into hours, and divide it into 24 hours to get the daily rainfall; add up the rainfall of 7 days to get the weekly rainfall; add up the rainfall of 30 days to get the total rainfall of each month; The hourly, daily, weekly and monthly rainfall is transmitted to the cloud platform server, which stores it and performs visualization.

4. The method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas according to claim 2 is characterized in that: The process of identifying independent rainfall events includes the following steps: The spatial variation characteristics of daily rainfall are calculated by fitting the daily rainfall input into the gamma distribution, and the estimated values ​​of shape parameters and scale parameters are obtained; Among them, the spatial characteristic change of daily rainfall is calculated as: Let x be the random rainfall in a certain period of time, and F(x) be the probability density function; Among them, α is the shape parameter and β is the scale parameter; Γ(α) is the gamma function, whose shape is controlled by α and β; when the shape parameter α<1, the gamma distribution is left-skewed, when α>1, the gamma distribution is right-skewed, and when α=1, the gamma distribution degenerates into an exponential distribution. The larger the scale parameter β, the narrower the gamma distribution range and the steeper the image. Define the threshold of the interval between rainfall events; classify rainfall events, and obtain the temporal and spatial distribution of rainfall events by analyzing the rainfall distribution in different time periods; Independent rainfall events are identified based on rainfall characteristics and lithological conditions; the joint probability density distribution function of rainfall intensity and duration is fitted, and the time interval distribution function of rainfall events is described in combination with the Poisson process to calibrate the interval time of rainfall events.

5. The method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas according to claim 2 is characterized in that: The process of evaluating the relevance of rainfall data at different time scales from at least one rain gauge station includes the following steps: Extract the rainfall data characteristics of the hour, day, week and month, and measure the rainfall data between different time periods based on the rainfall data characteristics of the hour, day, week and month; Among them, the rainfall data between different time periods are: Among them, x i and i are the values ​​of the two variables at the i-th time point, and is the average of these two variables; The Pearson correlation coefficient is checked. If the significance level is less than the preset significance threshold, it means that the Pearson correlation coefficient is significant. The rainfall data on different time scales and the relationship characteristics of the Pearson correlation on different time scales are transmitted to the cloud platform, and the cloud platform draws the rainfall data on different time scales and the relationship characteristics of the Pearson correlation on different time scales and displays them in a visual form.

6. The method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas according to claim 5 is characterized in that: The process of extracting the characteristics of rainfall data for the hour, day, week, and month includes the following steps: Perform rainfall trend analysis on historical rainfall data to obtain rainfall trend data, perform multi-scale wavelet transform on rainfall trend data, extract rainfall characteristics at different time scales, and obtain multi-scale rainfall characteristic data; Based on rainfall trends and multi-humidity rainfall characteristics data, the probability distribution of extreme rainfall events is calculated to obtain the probability distribution of extreme rainfall events; based on the probability distribution of rainfall events, multi-scale rainfall characteristics and rainfall trend data, rainfall analysis data is obtained; Among them, rainfall distribution data include rainfall intensity, rainfall duration and rainfall interval period; Based on rainfall analysis data, rainfall data between different time periods are measured; and the data between different time periods are spatially divided.

7. The method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas according to claim 6 is characterized in that: The process of obtaining the probability distribution of extreme rainfall events includes the following steps: The rainfall intensity is randomly sampled to determine the transfer path probability of rainfall of different intensities; the rainfall duration is analyzed and mined to obtain the hidden duration state probability distribution; Extract the associated features of the rainfall interval period, construct the conditional probability relationship between the rainfall interval period, rainfall intensity and duration, and obtain the interval period conditional relationship; combine the data of the transfer path probability, the hidden duration state probability distribution and the interval period conditional probability relationship to obtain the rainfall combination data; Construct rainfall probability density data based on rainfall combination data; adjust the probability distribution of rainfall probability density data to obtain adjusted probability density data; obtain the probability distribution of future rainfall and future predicted rainfall based on the adjusted probability density data and the probability of rainfall events; and perform spatial division based on the future rainfall probability distribution and future predicted rainfall.

8. The method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas according to claim 1 is characterized in that: The process of training the rainfall monitoring spatial model using the training set includes the following steps: The rainfall data at different time scales and the relationship between Pearson correlations at different scales are preprocessed by missing value and outlier monitoring and data standardization to obtain a data set; the data set is divided into a training set, a test set, and a validation set; A rainfall monitoring spatial model is constructed based on the rainfall data and lithologic deformation data of the target area, and the rainfall monitoring spatial model is iteratively trained based on the lithologic deformation data, rainfall time series characteristics, and rainfall analysis data until the number of training iterations reaches a preset number of training iterations, and the rainfall monitoring spatial model is completed; The validation set is used to validate the rainfall monitoring spatial model. If the preset validation threshold is reached, the final rainfall monitoring spatial model is obtained.

9. The method for analyzing spatial distribution patterns of rainfall in low-order small watersheds in mountainous areas according to claim 8 is characterized in that: The lithology-rainfall coupling loss function in the model training formula system is expressed as: In the formula, S m represents the displacement of the mth deformation monitoring point; H 1 represents the Sobolev space norm; TV represents the total variation regularization term; ∑J θ represents the rainfall covariance matrix; The spatiotemporal feature fusion equation is expressed as: In the formula, represents the three-dimensional space-time convolution operation; γ represents the vertical attenuation coefficient; z c represents the base depth of feature channel c; The adaptive momentum update rule is expressed as: Where η k represents the dynamic learning rate; τ represents the attenuation constant in the training phase; ⊙ represents the Hadamard product.

10. A system for analyzing spatial distribution patterns of rainfall in small low-order watersheds in mountainous areas, which is applied to the method for analyzing spatial distribution patterns of rainfall in small low-order watersheds in mountainous areas as claimed in any one of claims 1 to 9, characterized in that: The spatial distribution pattern analysis system of rainfall in low-order small watersheds in mountainous areas comprises: A module for obtaining historical rainfall data is used to obtain historical rainfall data of at least one rainfall monitoring station, merge the historical rainfall data according to different time scales of month, week, day and hour, and use the Pearson correlation coefficient to evaluate the correlation of rainfall data of at least one rainfall monitoring point at different time scales; Construct a spatial model module to form a data set based on the evaluation results, preprocess the data set, and divide the preprocessed data set into a training set, a validation set, and a test set; establish a rainfall monitoring spatial model, and use the training set to train the rainfall monitoring spatial model; The real-time adjustment and verification module is used to input rainfall data into the rainfall monitoring spatial model for simulation, divide the target area of ​​rainfall data according to the simulation results, and upload the division results to the cloud platform; compare the real-time data with the predicted data, and adjust the rainfall monitoring spatial model according to the comparison results.

Images