Method for observing local heavy rainfall in alpine region

Abstract

The invention discloses a method for observing local heavy rainfall in an alpine region, and particularly relates to the technical field of rainfall observation. In order to solve the problem of systematic prediction deviation caused by the fact that a traditional model mistakenly takes the altitude as a positive correlation covariable in a mountain leeside slope or a sinking airflow control area, a local slope direction difference index and a water vapor potential disturbance index are innovatively introduced as alternative covariables, and the actual relation between the altitude and rainfall is dynamically judged through multivariate correlation analysis. Therefore, the covariable setting of the common Kriging model is flexibly adjusted; through residual analysis and systematic deviation value calculation, the method further achieves the recognition and spatial correction of rainfall hot spots and microclimate characteristics, and remarkably improves the accuracy of interpolation prediction and the reliability of regional climate risk recognition.

Description

Technical Field

[0001] The present invention relates to the technical field of precipitation observation, and in particular to a method for observing local heavy precipitation in alpine areas. Background Art

[0002] Localized heavy precipitation observations in alpine regions involve monitoring and recording sudden, intense precipitation events (such as rainstorms and snowstorms) over a small area in high-altitude, low-temperature areas (such as the Qinghai-Tibet Plateau or high-latitude mountainous regions). These observations aim to provide information on the spatiotemporal distribution, intensity variations, and formation mechanisms of precipitation. This helps improve early warning capabilities for extreme weather, enhance the accuracy of regional climate models, and provide a scientific basis for disaster prevention and mitigation and water resources management.

[0003] The existing technology has the following shortcomings:

[0004] In specific alpine terrains, such as the leeward slopes of mountains or areas controlled by downdrafts over the plateau, precipitation may decrease with increasing altitude, demonstrating an inverse correlation that contradicts the conventional wisdom that "orographic uplift enhances precipitation." When interpolation algorithms (such as cokriging) incorporate altitude as a covariate and assume a positive correlation with precipitation, model predictions can exhibit systematic biases, leading to the misidentification of hotspots at higher altitudes and the misidentification of actually dry and warm areas as high-risk, distorting microclimate analysis and disaster warning results. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for observing local heavy rainfall in high-altitude and cold regions to solve the shortcomings of the background technology.

[0006] In order to achieve the above-mentioned object, the present invention provides the following technical solution: a method for observing local heavy rainfall in alpine areas, comprising:

[0007] Collect observation data in alpine areas, including surface precipitation observation data, digital elevation model data, meteorological element data, and radar remote sensing data;

[0008] Extracting a local aspect difference index and a water vapor potential disturbance index for optimizing a precipitation spatial interpolation model based on the observation data;

[0009] Multivariate correlation analysis was performed on the elevation, local aspect difference index, and water vapor potential disturbance index with precipitation data, and the correlation coefficient α between elevation and precipitation was calculated. If α was lower than the preset significance threshold, it was determined that there was no positive correlation between elevation and precipitation.

[0010] A co-kriging interpolation model is constructed, and the covariate settings are dynamically adjusted based on the analysis results. Specifically, if α>0, altitude is used as the covariate; if α≤0, the local aspect difference index or water vapor potential disturbance index is used instead of altitude as the covariate.

[0011] Perform spatial interpolation operations, output precipitation distribution grid maps, and calculate the predicted systematic deviation value to determine the degree of error accumulation caused by the covariate setting of the model;

[0012] Based on the interpolation results and the predicted systematic deviation values, precipitation hotspots and microclimate effects are identified and corrected, and an optimized precipitation spatial distribution map of the high-altitude and cold regions is output.

[0013] Preferably, the method for extracting the local slope difference index LADI is as follows: constructing a wind direction weight correspondence relationship, dividing the 360° wind direction circle into intervals of 10° to 30°, and for each direction segment, statistically calculating the frequency of occurrence of the wind direction corresponding to the direction segment f i , normalize the wind direction frequency to the weight coefficient w i For a certain terrain unit, its slope is S, then its wind direction θ i The angle difference Δ i =Δ i =min(|S-θ i |,360-|S-θ i |); calculate the local aspect difference index LADI, the expression is: N is the total number of wind direction segments.

[0014] Preferably, the method for extracting the water vapor potential disturbance index WVPI is: obtaining the altitude layer parameters including: air temperature T, specific humidity q, air pressure v, potential height Z, and calculating Frequency squared N 2 Used to measure the stability of the air bag when it rises / sinks. The expression is: Where: g is the acceleration due to gravity, D is the potential temperature; Represents the gradient of potential temperature with height, potential temperature calculation: Calculate the water vapor humidity ratio R q : Calculate the water vapor potential perturbation index WVPI, the expression is: in is the average stability in the selected layer.

[0015] Preferably, the precipitation P, altitude H, LADI and WVPI of the observation station or grid point are collected, and all variables are processed for missing values ​​and unit consistency; the correlation between each variable and precipitation is calculated using the Pearson correlation coefficient formula r X,P , the expression is: Where, X∈{H, LADI, WVPI}; X i The altitude, local slope difference index and water vapor potential disturbance index of the i-th sample point; P iis the corresponding precipitation; are the mean values ​​of H, LADI, WVPI and precipitation respectively; n is the total number of sample points; the correlation coefficient between the variable altitude H and precipitation P is named as: α = r H,P ; Perform a two-sided t-test on the correlation coefficient α and calculate its significance level: Find the critical value of the t distribution with n-2 degrees of freedom and calculate the corresponding significance level.

[0016] Preferably, if α<0 or p>αthreshold, the correlation is not significant, and it is determined that there is no positive correlation between altitude and precipitation, and altitude should not be used as a covariate in the interpolation model; αthreshold is the significance threshold.

[0017] Preferably, calculating the predicted systematic deviation value SPI includes:

[0018] Two interpolation models were constructed for comparison, including: Model A: co-kriging interpolation model; Model B: ordinary kriging interpolation model; the two models were used to cross-validate the same data set; for each observation point i, it was deleted and its value was predicted by interpolation using the remaining points, and the prediction error e was calculated. i , the expression is: Where, is the interpolated predicted precipitation value, For the actual observed precipitation value, the root mean square error RMSE of model A and model B is calculated for all points, and the expression is: r is the total number of observation points, RMSE A Represents the prediction error of the interpolation model with the introduction of covariates, RMSE B represents the prediction error of the baseline model without introducing covariates, is the prediction error of model A for the i-th observation point, is the prediction error of model B for the i-th observation point; the prediction systematic deviation value SPI is calculated, and the expression is:

[0019] Preferably, based on the interpolation results and the predicted systematic deviation values, precipitation hotspots and microclimate effects are identified and corrected, specifically including:

[0020] Perform co-kriging interpolation and output a preliminary precipitation distribution raster map P CK (x,y); perform cross validation and calculate the prediction error residual e for each observation point i ;

[0021] Perform co-kriging and ordinary kriging interpolation respectively and calculate their RMSE values;

[0022] For each covariate j, its spatial sensitivity is calculated on the cross-validation residual set, S j The (x,y) expression is: e(x,y) represents the difference between the model prediction value and the measured precipitation at the spatial point (x,y), V j (x,y) is the value of the jth covariate, which is the input variable for interpolation models such as cokriging, and constructs a weighted residual correction map: Where W j (x,y) is the normalized sensitivity; is the residual interpolation map; the revised precipitation forecast map: P adj (x,y)=P CK (x,y)-R(x,y); output the precipitation map P of the high-cold area after covariate sensitive weighted residual correction adj (x,y).

[0023] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0024] 1. This invention addresses the problem of systematic deviation caused by improper covariate settings in precipitation prediction under complex terrain conditions in high-altitude and cold regions. It innovatively proposes a high-precision spatial interpolation method for precipitation based on multi-source observation data, adaptive covariate discrimination, and sensitivity residual correction. By introducing the local aspect difference index and the water vapor potential disturbance index to replace the altitude variable that has an inverse correlation with precipitation in a specific area, the rationality and adaptability of the model covariate selection are effectively improved. The dynamic judgment of variable relationships is combined with multivariate correlation analysis and significance testing to avoid elevation misjudgment and hotspot shift caused by default assumption errors during the interpolation process.

[0025] 2. This method introduces a systematic prediction bias to quantify the cumulative error effect of the covariate. Furthermore, a covariate sensitivity-weighted residual correction algorithm is employed to quantitatively correct the initial precipitation interpolation results and re-identify hotspots. The resulting optimized precipitation spatial distribution map not only more accurately reflects local microclimate characteristics but also significantly improves the spatial location accuracy of high-risk areas for dry-warm or heavy precipitation events. This provides more reliable data support and technical solutions for disaster warning, water resource scheduling, and ecological assessments in high-altitude and cold regions. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments described in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0027] Figure 1 This is a mind map of the method of the present invention. DETAILED DESCRIPTION

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of 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. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0029] For examples, see Figure 1 As shown, the method for observing local heavy rainfall in a high-altitude cold region described in this embodiment includes:

[0030] Collect observation data in alpine areas, including surface precipitation observation data, digital elevation model data, meteorological element data, and radar remote sensing data;

[0031] Extracting a local aspect difference index and a water vapor potential disturbance index for optimizing a precipitation spatial interpolation model based on the observation data;

[0032] Multivariate correlation analysis was performed on the elevation, local aspect difference index, and water vapor potential disturbance index with precipitation data, and the correlation coefficient α between elevation and precipitation was calculated. If α was lower than the preset significance threshold, it was determined that there was no positive correlation between elevation and precipitation.

[0033] A co-kriging interpolation model is constructed, and the covariate settings are dynamically adjusted based on the analysis results. Specifically, if α>0, altitude is used as the covariate; if α≤0, the local aspect difference index or water vapor potential disturbance index is used instead of altitude as the covariate.

[0034] Perform spatial interpolation operations, output precipitation distribution grid maps, and calculate the predicted systematic deviation value to determine the degree of error accumulation caused by the covariate setting of the model;

[0035] Based on the interpolation results and the predicted systematic deviation values, precipitation hotspots and microclimate effects are identified and corrected, and an optimized precipitation spatial distribution map of the high-altitude and cold regions is output.

[0036] Collecting observational data in alpine regions is fundamental for conducting spatial analysis and interpolation modeling of localized heavy rainfall. Given the harsh climate, complex terrain, and inconvenient transportation in these regions, observational data collection requires a high degree of systematicity, fault tolerance, and precision control.

[0037] Multiple meteorological stations should be deployed on high-altitude ridges, valleys, and windward and leeward slopes, taking into account topographic relief, wind direction frequency, and climate representativeness. Tipping bucket rain gauges or weighing rain sensors should be equipped with windshields and heating systems to prevent freezing from rain and snow. A sampling frequency of 1 to 5 minutes is recommended to support the capture of minute-level sudden precipitation events. A mounting system with tilt correction should be used to achieve ±0.2mm measurement accuracy. Data should be uploaded to a central server via BeiDou, GPRS, or LoRa, with resumable uploads and local cache backups. High-resolution DEM datasets are preferred, such as SRTM 1Arc-Second (~30m resolution), ASTER GDEM (~30m resolution), and Copernicus GLO-30 (~30m resolution).

[0038] Seamlessly stitch multiple DEM data together; unify the projection to WGS84 / UTM projection; remove abnormal pixels through edge detection and elevation slope anomaly judgment; calculate slope (Slope), aspect (Aspect), terrain potential index (TPI), etc. based on DEM for subsequent terrain-related analysis.

[0039] Conventional meteorological elements include: temperature, humidity, wind speed, wind direction, and air pressure; standard sensors configured in automatic weather stations should all be resistant to low temperatures, and a temperature sensor with a range of -50°C is recommended; three or more tower-type observation devices (such as 2m, 10m, and 30m) are set up at typical locations to collect wind, temperature, and humidity profiles; synchronized with precipitation data, a sampling rate of 5 minutes or higher is recommended.

[0040] Use X-band or C-band weather radar, deployed at a high-viewing-angle position; provide radar reflectivity (Z), radial velocity (V), and dual-polarization parameters (such as ZDR); typically scan once every 5 minutes with a horizontal spatial resolution of 0.5 to 1 km; estimate surface precipitation intensity using the ZR relationship.

[0041] Satellite remote sensing data sources: FY-4A / VISSR, FY-3D / MERSI: provide infrared, water vapor, and cloud image data; GPM (Global Precipitation Measurement) / IMERG products: obtain hourly precipitation; MODIS / Aqua&Terra: used to obtain cloud top temperature, cloud cover, and water vapor path; Resolution and time: GPM products: 10 km, 30 minutes; MODIS: 250m~1km, 1-2 times / day; a "multi-source fusion" strategy is adopted to combine ground radar and remote sensing inversion results to improve the accuracy of precipitation estimation, especially in areas with sparse stations.

[0042] Compare ground station data with radar estimates and satellite precipitation; use statistical thresholds and machine learning methods to automatically remove or mark anomalous observations; unify data spatial resolution (e.g., interpolate to a 1km×1km grid), and align time to standard moments (e.g., on the hour or at 5-minute intervals); and convert data into standard formats such as NetCDF, GeoTIFF, and Shapefile to facilitate reading and processing by GIS and model analysis platforms.

[0043] Based on the observation data, a local aspect difference index and a water vapor potential disturbance index are extracted for optimizing the precipitation spatial interpolation model.

[0044] Among them, the local aspect difference index LADI is used to quantitatively represent the angle difference between the slope of the terrain where the observation point is located and the direction of the dominant moist air flow, so as to reveal the influence of windward or leeward terrain on precipitation formation.

[0045] The extraction method is to construct a wind direction weight correspondence, usually dividing the 360° wind direction circle into intervals of 10° to 30°, and dividing the dominant wind direction into multiple direction segments, such as 0°, 30°, 60°, 90°, 120°, 150°, etc. up to 330°, for a total of 12 equally spaced segments. For each direction segment, the frequency of occurrence of the wind direction corresponding to the direction segment is calculated based on historical wind rose diagrams or years of meteorological observation data. i For example, the 0° direction (due north wind) accounts for 5%, the 30° direction accounts for 10%, the 60° direction accounts for 15%, etc.

[0046] Normalize the wind direction frequency to the weight coefficient w i , the expression is: Among them, N is the total number of wind direction segments, f i is the frequency; for a certain terrain unit, its slope is S (unit: degree), then its relationship with wind direction θ i The angle difference Δ i =Δ i =min(|S-θ i |,360-|S-θ i |); (make sure the angle is between [0°, 180°]); calculate the local aspect difference index LADI, the expression is: The local slope aspect difference index is the weighted average angle difference, which reflects the "relative windward" or "leeward" nature of the slope aspect under the action of the comprehensive wind field.

[0047] The water vapor potential perturbation index (WVPI) is used to describe the influence of water vapor distribution and vertical perturbation between atmospheric layers in high-altitude cold regions on precipitation suppression or enhancement, and to reveal unfavorable precipitation structures such as inversion layers and water vapor cutoff layers.

[0048] The extraction method is: obtain altitude layer parameters including: air temperature T, specific humidity q, air pressure p, geopotential height Z. Recommended layers: 1000hPa, 925hPa, 850hPa, 700hPa, 600hPa, 500hPa.

[0049] calculate Frequency squared N 2 (Stratification stability index), this frequency is used to measure the stability of the air bag when it rises / sinks, the expression is: Where: g is the acceleration due to gravity, about 9.8m / s 2 ; D is potential temperature, unit K; Indicates the gradient of potential temperature with height. Potential temperature calculation: v is the atmospheric pressure, q is the specific humidity (the mass of water vapor per unit mass of air);

[0050] Calculate the water vapor humidity ratio R q , select two key layers (such as ground layer 1000hPa and 850hPa) specific humidity: If R q <1 indicates the presence of upper-level water vapor attenuation, which may be a dry layer or a suppression zone caused by temperature inversion. The water vapor potential perturbation index WVPI is calculated as: in is the average stability in the selected stratification (the average or maximum value of 850–500 hPa can be taken).

[0051] Collect the following data at the observing stations or grid points:

[0052] For precipitation P, altitude H, LADI, and WVPI, all variables are processed for missing values ​​and have their units consistent. LADI and WVPI can be normalized to 0-1 to eliminate the impact of scale differences.

[0053] The correlation between each variable and precipitation was calculated using the Pearson correlation coefficient formula. X,P , the expression is: Where X∈{H, LADI, WVPI}; X i The altitude, local slope difference index and water vapor potential disturbance index of the i-th sample point; P i is the corresponding precipitation; are the mean values ​​of H, LADI, WVPI and precipitation respectively; n is the total number of sample points; the correlation coefficient between the variable altitude (H) and precipitation (P) is named as: α = r H,P ; Perform a two-sided t-test on the correlation coefficient α and calculate its significance level p-value: Find the critical value of the t distribution with n-2 degrees of freedom and calculate the corresponding p-value; usually set the significance threshold αthreshold = 0.1 or 0.05; if p>αthreshold, α is considered not significant.

[0054] If α < 0 or p > αthreshold, the correlation is not significant, indicating that there is no positive correlation between altitude and precipitation and altitude should not be used as a covariate in the interpolation model. If altitude is excluded, LADI and WVPI can be considered as covariates.

[0055] Perform spatial interpolation operations, output a precipitation distribution grid map, and calculate the predicted systematic deviation value to determine the degree of error accumulation caused by the covariate setting of the model, including:

[0056] The purpose of interpolation is to expand limited point precipitation observation data into a continuous spatial precipitation surface, which is used to identify local heavy precipitation hotspots and spatial distribution patterns.

[0057] Common methods for selecting interpolation algorithms: Ordinary Kriging: no covariates, based only on spatial autocorrelation; Co-Kriging: introducing one or more covariates (such as altitude, LADI, WVPI); Regression Kriging (RK): regression first, then residual interpolation.

[0058] Based on the spatial distance between points and precipitation differences, theoretical variograms (commonly used models such as spherical, exponential, and linear) are fitted. When performing co-kriging, a covariance model between the main variable and the covariate needs to be constructed.

[0059] Use GIS software (such as ArcGIS Geostatistical Analyst, QGIS+SAGA, Rgstat package); set the interpolation grid resolution (such as 1km×1km); and generate a complete precipitation distribution raster map (GeoTIFF, ASCII, NetCDF, etc. formats).

[0060] The predicted systematic bias (SPI) was calculated to assess whether the covariate (such as altitude) introduced systematic bias in the interpolation model and to determine whether it should be retained or replaced.

[0061] Two interpolation models were constructed for comparison, including: Model A: co-kriging interpolation model (introducing covariates such as altitude); Model B: ordinary kriging interpolation model (not introducing covariates); and the two models were cross-validated on the same dataset.

[0062] For each observation point i, delete the point, use the remaining points to predict its value by interpolation, and calculate the prediction error e i , the expression is: Where, is the interpolated predicted precipitation value, For the actual observed precipitation value, the root mean square error RMSE of model A and model B is calculated for all points, and the expression is: r is the total number of observation points, RMSE A Represents the prediction error of the interpolation model with the introduction of covariates, RMSE B represents the prediction error of the baseline model without introducing covariates, is the prediction error of model A for the i-th observation point, is the prediction error of model B for the i-th observation point; the prediction systematic deviation value SPI is calculated, and the expression is:

[0063] SPI<1.0: The introduction of covariates improves prediction accuracy, indicating that it has a positive effect on the spatial distribution of precipitation; SPI≈1.0: The introduction of covariates has no substantial effect; SPI>1.0: The introduction of covariates causes a decrease in prediction performance, and there is a risk of cumulative systematic errors. It is recommended to remove or replace the covariates.

[0064] Based on the interpolation results and the predicted systematic deviation values, precipitation hotspots and microclimate effects are identified and corrected, and an optimized spatial distribution map of precipitation in the alpine region is output, including:

[0065] Perform co-kriging interpolation and output a preliminary precipitation distribution raster map P CK (x,y); perform cross validation and calculate the prediction error residual e for each observation point i .

[0066] Execute co-kriging (including covariates) and ordinary kriging (without covariates) interpolation respectively, and calculate their RMSE values.

[0067] For each covariate j (such as altitude, LADI, WVPI), its spatial sensitivity is calculated on the cross-validation residual set, S j The (x,y) expression is: e(x,y) represents the difference between the model prediction value and the measured precipitation at the spatial point (x,y), V j (x,y) is the value of the jth covariate, which is the input variable used in interpolation models such as cokriging, such as altitude, elevation, LADI, WVPI, etc. It can be extracted by linear fitting residuals and covariate gradients, or regression model slopes; it represents the strength of the covariate's impact on the prediction error at that point. Construct a weighted residual correction plot: Where W j (x,y) is the normalized sensitivity; It is a residual interpolation map (IDW or OK interpolation can be used); in essence, it redistributes the error according to the influence.

[0068] Corrected precipitation forecast map: P adj (x,y)=P CK (x,y)-R(x,y); output the precipitation map of the high-cold area after covariate sensitive weighted residual correction; repair the systematic bias area caused by the misuse of covariates; while retaining the real local microclimate gradient characteristics.

[0069] P adj (x,y) performs hotspot identification; compares the changes in hotspot distribution center and range before and after correction; marks hotspot areas that are incorrectly located or missed due to incorrect covariates, and identifies newly formed or exposed microclimate high-sensitivity areas (such as potential dry areas under the control of inversion dry layers).

[0070] Output the final precipitation spatial distribution map P adj (x,y), in formats such as GeoTIFF, NetCDF, and ASCII Grid; layers include: corrected precipitation intensity; significant hotspot areas; systematic bias correction annotation areas (which can be used as a quality control layer); and distribution maps of microclimate susceptible areas.

[0071] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0072] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0073] It should be understood that the term "and / or" herein is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent three situations: A exists alone, A and B exist at the same time, and B exists alone, where A and B may be singular or plural. In addition, the character " / " herein generally indicates that the objects associated with each other are in an "or" relationship, but it may also indicate an "and / or" relationship, which can be understood by referring to the context. A person of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0074] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A method for observing local heavy rainfall in alpine regions, characterized by: include: Collect observation data in alpine areas, including surface precipitation observation data, digital elevation model data, meteorological element data, and radar remote sensing data; Extracting a local aspect difference index and a water vapor potential disturbance index for optimizing a precipitation spatial interpolation model based on the observation data; Multivariate correlation analysis was performed on the elevation, local aspect difference index, and water vapor potential disturbance index with precipitation data, and the correlation coefficient α between elevation and precipitation was calculated. If α was lower than the preset significance threshold, it was determined that there was no positive correlation between elevation and precipitation. A co-kriging interpolation model is constructed, and the covariate settings are dynamically adjusted based on the analysis results. Specifically, if α>0, altitude is used as the covariate; if α≤0, the local aspect difference index or water vapor potential disturbance index is used instead of altitude as the covariate. Perform spatial interpolation operations, output precipitation distribution grid maps, and calculate the predicted systematic deviation value to determine the degree of error accumulation caused by the covariate setting of the model; Based on the interpolation results and the predicted systematic deviation values, precipitation hotspots and microclimate effects are identified and corrected, and an optimized precipitation spatial distribution map of the high-altitude and cold regions is output.

2. The method for observing local heavy rainfall in alpine regions according to claim 1, characterized in that: The extraction method of the local slope difference index LADI is as follows: construct the wind direction weight correspondence relationship, divide the 360° wind direction circle into intervals of 10° to 30°, and for each direction segment, count the occurrence frequency f of the wind direction corresponding to the direction segment. i , normalize the wind direction frequency to the weight coefficient w i For a certain terrain unit, its slope is S, then its wind direction θ i The angle difference Δ i =Δ i =min(|S-θ i |,360-|S-θ i |); calculate the local aspect difference index LADI, the expression is: N is the total number of wind direction segments.

3. The method for observing local heavy rainfall in alpine regions according to claim 2, characterized in that: The extraction method of water vapor potential disturbance index WVPI is as follows: obtain the altitude parameters including air temperature T, specific humidity q, air pressure v, potential height Z, calculate Brunt– Frequency squared N 2 Used to measure the stability of an air bag as it rises / sinks. The expression is: Where: g is the acceleration due to gravity, D is the potential temperature; Represents the gradient of potential temperature with height, potential temperature calculation: Calculate the water vapor humidity ratio R q : Calculate the water vapor potential perturbation index WVPI, the expression is: in is the average stability in the selected layer.

4. The method for observing local heavy rainfall in alpine regions according to claim 3, characterized in that: The precipitation P, altitude H, LADI and WVPI of the observation station or grid point were collected, and missing values ​​and unit consistency were processed for all variables; the correlation r between each variable and precipitation was calculated using the Pearson correlation coefficient formula. X,P , the expression is: Where, X∈{H, LADI, WVPI}; X i The altitude, local slope difference index and water vapor potential disturbance index of the i-th sample point; P i is the corresponding precipitation; are the mean values ​​of H, LADI, WVPI and precipitation respectively; n is the total number of sample points; the correlation coefficient between the variable altitude H and precipitation P is named as: α = r H,P ; Perform a two-sided t-test on the correlation coefficient α and calculate its significance level: Find the critical value of the t distribution with n-2 degrees of freedom and calculate the corresponding significance level.

5. The method for observing local heavy rainfall in alpine regions according to claim 4, characterized in that: If α<0 or p>αthreshold, the correlation is not significant, and it is determined that there is no positive correlation between altitude and precipitation, and altitude should not be used as a covariate in the interpolation model; αthreshold is the significance threshold.

6. The method for observing local heavy rainfall in alpine regions according to claim 5, characterized in that: Calculate the predicted systematic deviation value SPI, including: Two interpolation models were constructed for comparison, including: Model A: co-kriging interpolation model; Model B: ordinary kriging interpolation model; cross-validation was performed on the same dataset using both models; for each observation point i, its value was predicted by interpolation using the remaining points, and the prediction error e was calculated. i , the expression is: Where, is the interpolated predicted precipitation value, For the actual observed precipitation value, the root mean square error RMSE of model A and model B is calculated for all points, and the expression is: r is the total number of observation points, RMSE A Represents the prediction error of the interpolation model with the introduction of covariates, RMSE B represents the prediction error of the baseline model without introducing covariates, is the prediction error of model A for the i-th observation point, is the prediction error of model B for the i-th observation point; the prediction systematic deviation value SPI is calculated as follows:

7. The method for observing local heavy rainfall in alpine regions according to claim 6, characterized in that: Based on the interpolation results and the predicted systematic deviation values, precipitation hotspots and microclimate effects are identified and corrected, including: Perform co-kriging interpolation and output a preliminary precipitation distribution raster map P CK (x,y); perform cross validation and calculate the prediction error residual e for each observation point i ; Perform co-kriging and ordinary kriging interpolation respectively and calculate their RMSE values; For each covariate j, its spatial sensitivity is calculated on the cross-validation residual set, S j The (x,y) expression is: e(x,y) represents the difference between the model prediction value and the measured precipitation at the spatial point (x,y), V j (x,y) is the value of the jth covariate, which is the input variable for interpolation models such as cokriging, and constructs a weighted residual correction map: Where W j (x,y) is the normalized sensitivity; is the residual interpolation map; the revised precipitation forecast map: P adj (x,y)=P CK (x,y)-R(x,y); output the precipitation map P of the high-cold area after covariate sensitive weighted residual correction adj (x,y).

Images