Method and device for the revision of the multiple classification of the precipitation phase
By using a stepwise conditional binary classification and a spatial downscaling algorithm for the probability variability of precipitation phases, the problems of probability constraints and sample size imbalance in multi-class precipitation phase forecasting are solved, achieving high-resolution and accurate precipitation phase forecasting.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NATIONAL METEOROLOGICAL CENTRE
- Filing Date
- 2026-02-10
- Publication Date
- 2026-05-12
AI Technical Summary
Existing technologies cannot guarantee the constraint that the sum of probabilities of different precipitation phases is 1 in multi-class precipitation phase probability forecasting, and the imbalance of sample size affects the correction effect, resulting in inaccurate forecasts. Spatial downscaling methods lack reliability.
A stepwise conditional binary classification algorithm for precipitation phase probability correction and a spatial downscaling algorithm for precipitation phase probability variability are adopted. The stepwise conditional binary classification method is used to determine the preliminary corrected multiphase precipitation probability, and spatial downscaling is performed in combination with temperature variability to ensure that the sum of probabilities is 1 and to improve spatial resolution.
It improves the accuracy and spatial resolution of multi-class precipitation phase probability forecasts, overcomes the impact of sample size imbalance, and achieves efficient correction and spatial continuity.
Smart Images
Figure CN121682452B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of information processing technology, and specifically to a method and apparatus for correcting multiple precipitation phases. Background Technology
[0002] Precipitation exists in various phases, including rain, snow, sleet, freezing rain, ice pellets, and graupel. Therefore, precipitation phase is a typical discrete multi-class variable. Ensemble numerical weather prediction models provide basic probabilistic forecast products for different precipitation phases. However, the reliability of these products is generally low, and their spatial resolution is coarse (above 9 km). Therefore, statistical post-processing corrections and spatial downscaling are needed to correct the probabilistic precipitation phase forecasts output by these weather forecast models. This aims to eliminate forecast bias and improve forecast skill, while also increasing the spatial resolution of the forecasts to meet the needs of refined weather forecasting.
[0003] Currently, most post-processing methods for discrete variables in probabilistic forecasts of meteorological elements such as temperature, precipitation, and precipitation phases are designed for binary events, such as the presence or absence of precipitation, or the presence or absence of heavy rain. For multi-class events, multiple binary classification methods are employed, such as classifying precipitation phases sequentially into rain and non-rain, snow and non-snow, freezing rain and non-freezing rain, and then performing probability corrections for each phase. This method has two drawbacks. First, the sum of the probabilities of different precipitation phases (including no precipitation) must be 1, a constraint that must be satisfied before and after correction. However, existing correction methods cannot guarantee this constraint and can only achieve this by normalizing the probabilities of different precipitation phases after correction, which inevitably affects the objective correction effect of the probabilistic forecast. Second, the sample sizes for different precipitation phases are extremely unbalanced. The ratio of non-rain to rain samples is on the order of 10:1, and the ratio of non-rain / snow mix to rain / snow mix is on the order of 100:1. This significant imbalance in sample sizes greatly affects the effectiveness of objective correction, causing the final correction probability to favor the event with the larger sample size. These two shortcomings limit the effectiveness of the multiple binary classification correction method for the discretized multi-class event of precipitation phase.
[0004] Spatial downscaling methods for discrete variables like precipitation phases are currently mainly based on the linear changes in temperature and the rain / snow boundary, assuming a temperature lapse rate of 0.6℃ / 100 meters. Within a certain area (typically tens of kilometers square), a temperature change of about 1℃ can lead to a transition between rain and snow. Based on this assumption, spatial downscaling is performed to linearize the precipitation phase. However, there is currently no reliable statistical downscaling method for the probability of precipitation phases. Summary of the Invention
[0005] The main objective of this invention is to provide a correction method and apparatus for multi-class precipitation phases to address the shortcomings in related technologies.
[0006] To achieve the above objectives, according to a first aspect of the present invention, a correction method for multi-class precipitation phases is provided, comprising: determining the probability of multi-phase precipitation after preliminary correction using a stepwise conditional binary classification precipitation phase probability correction algorithm for a full spatiotemporal sample; and performing spatial downscaling processing on the probability of the multi-phase precipitation after preliminary correction using a precipitation phase probability variability spatial downscaling algorithm to obtain a high-resolution precipitation phase probability forecast correction product.
[0007] Optionally, for the full spatiotemporal sample, the stepwise conditional binary classification precipitation phase probability correction algorithm is used to determine the probability of multiphase precipitation after preliminary correction. This includes: classifying all events into two categories: no precipitation and precipitation, and correcting them using a binary classification method to obtain the corrected probability of precipitation P1; classifying precipitation events into rainfall and non-rainfall, and correcting them using a binary classification method to obtain the rainfall probability P_rain under precipitation conditions; determining the preliminary corrected probability of rainfall based on the precipitation probability and the rainfall probability; classifying the non-rainfall events into snowfall and non-solid precipitation, and correcting them using a binary classification method to obtain the snowfall probability P_snow under precipitation and non-rainfall conditions, and determining the preliminary corrected probability of rainfall based on the precipitation probability and the non-rainfall probability. The corrected snowfall probability is determined by considering the snow probability P_snow, the precipitation probability P1, and the rainfall probability P_rain. The non-solid precipitation events are then divided into sleet and freezing rain, and a binary classification correction method is used to obtain the sleet probability P_sleet and freezing rain probability P_freezing under the condition of precipitation but no rainfall or snowfall. The corrected sleet probability is determined based on the sleet probability P_sleet, snowfall probability P_snow, rainfall probability P_rain, and precipitation probability P1. Similarly, the corrected freezing rain probability is determined based on the freezing rain probability P_freezing, snowfall probability P_snow, rainfall probability P_rain, and precipitation probability P1.
[0008] Optionally, a spatial downscaling algorithm for the probability variability of precipitation phases is used to spatially downscale the probabilities of the preliminarily corrected multiphase precipitation, including: calculating the difference between the terrain height of the coarse-resolution product and the terrain height of the high-resolution product; calculating the temperature variability dT of the high-resolution product relative to the coarse-resolution product based on the wet adiabatic lapse rate and the difference; fitting the probability of rain and snow at different temperatures using a logistic regression function to obtain the probability functions of rain and snow; differentiating the probability functions of rain and snow respectively to obtain the variability dPr of the probability of rain and the variability dPs of the probability of snow at different temperatures; and combining the temperature variability dT and the probability variability dPr of the precipitation. The snowfall probability variability dPs and the high-resolution temperature forecast product are used to calculate the precipitation probability correction value and the snowfall probability correction value, respectively, and then the sleet probability correction value is obtained. The precipitation probability correction value, snowfall probability correction value and sleet probability correction value are added to the probability of the preliminarily corrected multiphase precipitation to obtain the downscaled probability of rain, snow, sleet and freezing rain, forming the high-resolution precipitation phase probability forecast correction product.
[0009] Optionally, for a precipitation event, determining the preliminary probability of rainfall includes: determining the corrected preliminary probability of rainfall based on the precipitation probability and the rainfall probability includes: calculating the final corrected preliminary probability of rainfall as P_rain×P1 based on the conditional probability formula.
[0010] Optionally, determining the corrected snowfall probability based on the snowfall probability P_snow, the precipitation probability P1, and the rainfall probability P_rain includes: determining the snowfall probability according to the conditional probability formula as: P_snow×(1-P_rain)×P1.
[0011] Optionally, the corrected probability of sleet is determined based on the probability of sleet P_sleet, the probability of snowfall P_snow, the probability of rainfall P_rain, and the probability of precipitation P1; and the corrected probability of freezing rain is determined based on the probability of freezing rain P_freezing, the probability of snowfall P_snow, the probability of rainfall P_rain, and the probability of precipitation P1, including: calculating the final corrected probability of sleet according to the conditional probability formula as: P_sleet×(1-P_snow)×(1-P_rain)×P1, and calculating the final corrected probability of freezing rain according to the conditional probability formula as: P_freezing×(1-P_snow)×(1-P_rain)×P1.
[0012] According to a second aspect of the present invention, a correction device for multi-phase precipitation is provided, comprising: a probability determination unit for multi-phase precipitation, configured to determine the probability of multi-phase precipitation after preliminary correction using a stepwise conditional binary classification precipitation phase probability correction algorithm for a full range of spatiotemporal samples; and a high-resolution precipitation phase probability forecast correction unit, configured to perform spatial downscaling processing on the probability of the multi-phase precipitation after preliminary correction using a precipitation phase probability variability spatial downscaling algorithm to obtain a high-resolution precipitation phase probability forecast correction product.
[0013] According to a third aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing the computer to perform the method described in any one of the first aspects.
[0014] According to a fourth aspect of the present invention, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to cause the at least one processor to perform the method described in any implementation of the first aspect.
[0015] This embodiment addresses a correction method for multi-class precipitation phase probabilities. It includes using a stepwise conditional binary classification algorithm to determine the probabilities of multi-phase precipitation after initial correction, based on a full spatiotemporal sample. A spatial downscaling algorithm based on precipitation phase probability variability is then used to spatially downscale the probabilities of the initially corrected multi-phase precipitation, resulting in a high-resolution precipitation phase probability forecast correction product. By employing a stepwise binary classification algorithm based on conditional probability, the differences between sample sizes are minimized in each binary classification process because conditional probability only considers a few types of events, thus minimizing the correction effect caused by sample imbalance. This significantly improves the forecasting accuracy of the corrected precipitation phase probabilities. The probabilistic spatial downscaling algorithm based on precipitation phase probability variability correction, when dealing with kilometer-level resolution downscaling, uses matrix operations, improving computational efficiency while ensuring the spatial continuity of precipitation phase probabilities. This solves the problems existing in related technologies. Attached Figure Description
[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0017] Figure 1 This is a flowchart of the correction method for multi-class precipitation phases according to an embodiment of the present invention;
[0018] Figure 2 This is a schematic diagram illustrating the application of the correction method for multi-class precipitation phases in an embodiment of the present invention;
[0019] Figure 3 This is a schematic diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation
[0020] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0021] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of the invention described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0022] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.
[0023] According to embodiments of the present invention, a correction method for multi-class precipitation phases is provided, such as... Figure 1 As shown, steps 101 to 102 are included below:
[0024] Step 101: For the full spatiotemporal sample, the probability of multiphase precipitation after preliminary correction is determined by the stepwise conditional binary classification precipitation phase probability correction algorithm.
[0025] In this step, the initial probability of precipitation phases output by the ensemble numerical weather prediction model is a raw, unprocessed probabilistic forecast result based on atmospheric dynamics / thermodynamics, representing the likelihood of various precipitation phases (rain, snow, sleet, freezing rain, etc.) occurring in the target area over a certain future period. It is a probabilistic forecast output for precipitation phases, a discrete multi-category variable, rather than a deterministic conclusion. Correction yields the corrected probabilities. The probabilities of multi-phase precipitation include the probabilities of rain, snow, sleet, and freezing rain.
[0026] Step 102: Using the spatial downscaling algorithm of precipitation phase probability variability, the probabilities of the preliminarily corrected multiphase precipitation are spatially downscaled to obtain a high-resolution precipitation phase probability forecast correction product.
[0027] In this embodiment, the core output of the correction product is the final correction probability of four precipitation phases. Based on this product, users can determine the final probability of rain, snow, sleet, and freezing rain in a target area, compare it with a preset risk threshold, and generate decision support information for scenarios such as flood control, traffic management, and agricultural production.
[0028] As an optional implementation of this embodiment, for the full spatiotemporal sample, the stepwise conditional binary classification precipitation phase probability correction algorithm is used to determine the probability of multiphase precipitation after preliminary correction. This includes: classifying all events into two categories: no precipitation and precipitation, and correcting them using a binary classification method to obtain the corrected probability of precipitation P1; classifying precipitation events into rainfall and non-rainfall events, and correcting them using a binary classification method to obtain the probability of rainfall under precipitation conditions P_rain; determining the preliminary probability of rainfall after correction based on the precipitation probability and the rainfall probability; and classifying the non-rainfall events into snowfall and non-solid precipitation, and correcting them using a binary classification method to obtain the probability of snowfall under both precipitation and non-rainfall conditions P_snow. Based on the snowfall probability P_snow, the precipitation probability P1, and the rainfall probability P_rain, the corrected snowfall probability is determined. The non-solid precipitation events are divided into sleet and freezing rain, and a binary classification correction method is used to correct them, resulting in the sleet probability P_sleet and freezing rain probability P_freezing under the condition of precipitation but no rainfall and no snowfall. Based on the sleet probability P_sleet, snowfall probability P_snow, rainfall probability P_rain, and precipitation probability P1, the corrected sleet probability is determined. And based on the freezing rain probability P_freezing, snowfall probability P_snow, rainfall probability P_rain, and precipitation probability P1, the corrected freezing rain probability is determined.
[0029] As an optional implementation of this embodiment, for a precipitation event, determining the preliminary probability of rainfall includes: calculating the final corrected preliminary probability of rainfall as P_rain×P1 based on the conditional probability formula.
[0030] As an optional implementation of this embodiment, the corrected snowfall probability is determined based on the snowfall probability P_snow, the precipitation probability P1, and the rainfall probability P_rain, including: determining the snowfall probability as P_snow×(1-P_rain)×P1 according to the conditional probability formula.
[0031] As an optional implementation of this embodiment, the corrected probability of sleet is determined based on the probability of sleet P_sleet, the probability of snowfall P_snow, the probability of rainfall P_rain, and the probability of precipitation P1; and the corrected probability of freezing rain is determined based on the probability of freezing rain P_freezing, the probability of snowfall P_snow, the probability of rainfall P_rain, and the probability of precipitation P1, including: calculating the final corrected probability of sleet according to the conditional probability formula as: P_sleet×(1-P_snow)×(1-P_rain)×P1, and calculating the final corrected probability of freezing rain according to the conditional probability formula as: P_freezing×(1-P_snow)×(1-P_rain)×P1.
[0032] Among the above-mentioned optional implementation methods, the stepwise conditional binary classification precipitation phase probability correction algorithm adopts a stepwise binary classification correction method based on conditional probability.
[0033] The first step is to divide all events (full spatiotemporal sample) into two categories: no precipitation and precipitation.
[0034] For example, each independent event can be classified by combining a quantization threshold (e.g., cumulative precipitation over 6 hours ≥ 0.1 mm at a spatiotemporal location) with spatiotemporal matching (e.g., spatiotemporally matching forecast grids with observation stations to ensure each event corresponds to a unique observation result). For instance, all forecast grids and time-series events in the target area can be traversed, and the measured precipitation after event matching can be extracted for each event one by one; the event can then be determined as either having no precipitation or having precipitation by comparing it with the quantization threshold.
[0035] refer to Figure 2The first step involves using a binary classification correction method, resulting in a corrected precipitation probability of P1. The second step considers only the precipitation samples, dividing them into two categories under the conditional probability of precipitation: rain (liquid) and non-rain (non-liquid). The binary classification correction method is then applied to correct these two categories, resulting in the precipitation probability P_rain and the non-rain probability P_noRain under the given precipitation condition. The final corrected precipitation probability is then P_rain × P1. The third step considers only the non-rain samples within the precipitation samples, dividing them into snow (solid precipitation) and non-solid precipitation under the condition of precipitation and non-rain. The correction method is then applied to form the snowfall probability P_snow and the non-snowfall probability P_noSnow under the given precipitation and non-rain conditions. The final corrected snow probability is then P_snow × (1 - P_rain) × P1. Fourth, considering only the non-solid precipitation samples, the remaining samples are sleet and freezing rain. The binary classification correction method is then used, generating sleet and freezing rain probabilities P_sleet and P_freezing, respectively. Based on the conditional probability formula, the final corrected probabilities of sleet and freezing rain are:
[0036] P_sleet×(1-P_snow)×(1-P_rain)×P1
[0037] P_freezing×(1-P_snow)×(1-P_rain)×P1.
[0038] In the above process, since each binary classification correction is based on the conditional probability of the previous step, the constraint that the sum of the probabilities of different precipitation phases is always satisfied is always met. This overcomes the shortcomings of related technologies that use multiple independent binary classification corrections, which cannot guarantee that the sum of the probabilities of various precipitation phases is 1, and require subsequent normalization processing to correct, resulting in a decrease in the objective correction effect of probability forecasts.
[0039] By using a step-by-step sample selection process, each step corrects only the subset of target samples selected in the previous step, minimizing the impact of sample imbalance on the correction effect and improving the accuracy of probability forecasts for low-frequency phases such as sleet and freezing rain.
[0040] As an optional implementation of this embodiment, a spatial downscaling algorithm for the probability variability of precipitation phases is used to spatially downscale the probabilities of the preliminarily corrected multiphase precipitation. This includes: calculating the difference between the terrain height of the coarse-resolution product and the terrain height of the high-resolution product; calculating the temperature variability dT of the high-resolution product relative to the coarse-resolution product based on the wet adiabatic lapse rate and the difference; fitting the probability of rain and snow at different temperatures using a logistic regression function to obtain the probability functions of rain and snow; differentiating the probability functions of rain and snow to obtain the variability dPr of the probability of rain and the variability dPs of the probability of snow at different temperatures; and combining the temperature variability dT and the probability variability dPr of the precipitation. The snowfall probability variability dPs and the high-resolution temperature forecast product are used to calculate the precipitation probability correction value and the snowfall probability correction value, respectively, and then the sleet probability correction value is obtained. The precipitation probability correction value, snowfall probability correction value and sleet probability correction value are added to the probability of the preliminarily corrected multiphase precipitation to obtain the downscaled probability of rain, snow, sleet and freezing rain, forming the high-resolution precipitation phase probability forecast correction product.
[0041] In this optional implementation, the spatial downscaling algorithm for precipitation phase probability variability first calculates the difference in terrain height between products at different resolutions, where the coarse-resolution bilinear interpolation is performed to the high-resolution version. When precipitation occurs, the vertical temperature variability is generally the wet adiabatic lapse rate, taken as 0.6℃ / 100m. Based on this vertical variability, the temperature variability at high resolution relative to the coarse resolution can be calculated. The frequency of rain and snow occurrence with temperature exhibits a logistic regression function; therefore, the probability of rain and snow occurrence at different temperatures is fitted using a logistic regression function. Based on the fitted probability functions of rain and snow occurrence, the variability of the probability of rainfall and snowfall at different temperatures can be calculated, which is the derivative of the probability function. Combining the aforementioned temperature variability and the high-resolution temperature forecast product, the correction value for the precipitation phase probability from coarse resolution to high resolution can be calculated. Adding this correction value to the corrected probabilities of rain, snow, and sleet from the previous step yields the downscaled precipitation phase probability forecast correction product. This achieves objective correction and spatial downscaling of the precipitation phase probability.
[0042] For example, the coarse resolution threshold range and the fine resolution threshold range can be preset. For instance, the coarse resolution can be at the 50km level and the fine resolution can be at the 1km level. No limitation is made here.
[0043] This implementation of the precipitation phase probability variability spatial downscaling algorithm can accurately characterize the precipitation phase differences in small-scale regions, filling the gap in existing technologies for precipitation phase probability spatial downscaling and meeting the core requirements of refined weather forecasting.
[0044] In summary, the objective correction method for multi-class discrete variables of precipitation phases in this embodiment can always maintain the constraint that the sum of probabilities of multi-class precipitation phases is always 1, without forcibly normalizing the probabilities of multi-class precipitation phases, thus avoiding the technique of reducing probability forecasts during the normalization process.
[0045] The stepwise binary classification algorithm based on conditional probability, by considering only a few classes of events in each binary classification process, can minimize the differences between sample sizes and eliminate the poor correction effect caused by sample imbalance to the greatest extent. Correction experiments also show that this method can significantly improve the forecasting skill of precipitation phase probability after correction, and is significantly better than the multi-class binary classification algorithm.
[0046] The probabilistic spatial downscaling algorithm based on precipitation phase probability variability correction is adopted. When dealing with kilometer-level resolution downscaling, the matrix operation method is used, which improves the computational efficiency and ensures the spatial continuity of precipitation phase probability.
[0047] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.
[0048] According to an embodiment of the present invention, a correction device for multi-class precipitation phases is also provided, comprising a probability determination unit for multi-phase precipitation, used to determine the probability of multi-phase precipitation after preliminary correction using a stepwise conditional binary classification precipitation phase probability correction algorithm for the full spatiotemporal sample; and a high-resolution precipitation phase probability forecast correction unit, used to perform spatial downscaling processing on the probability of the multi-phase precipitation after preliminary correction using a precipitation phase probability variability spatial downscaling algorithm to obtain a high-resolution precipitation phase probability forecast correction product.
[0049] According to embodiments of the present invention, the present invention also provides an electronic device, the electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to implement the methods described in any of the above embodiments.
[0050] According to embodiments of the present invention, the present invention also provides a readable storage medium storing computer instructions that enable a computer to perform the methods described in any of the above embodiments when executed.
[0051] According to embodiments of the present invention, the present invention also provides a computer program product that, when executed by a processor, can implement the methods described in any of the above embodiments.
[0052] Figure 3 A schematic block diagram of an example electronic device 300 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices.
[0053] like Figure 3 As shown, the electronic device 300 includes a computing unit 301, which can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) 302 or a computer program loaded from a storage unit 308 into a random access memory (RAM) 303. The RAM 303 may also store various programs and data required for the operation of the electronic device 300. The computing unit 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.
[0054] Multiple components in electronic device 300 are connected to I / O interface 305, including: input unit 306, such as keyboard, mouse, etc.; output unit 307, such as various types of displays, speakers, etc.; storage unit 308, such as disk, optical disk, etc.; and communication unit 309, such as network card, modem, wireless transceiver, etc. Communication unit 309 allows electronic device 300 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.
[0055] The computing unit 301 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 301 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 301 performs the various methods and processes described above, such as the object matching method. For example, in some embodiments, the object matching method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 308. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 300 via ROM 302 and / or communication unit 309. When the computer program is loaded into RAM 303 and executed by the computing unit 301, one or more steps of the methods described above may be performed.
[0056] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.
[0057] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.
[0058] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
Claims
1. A correction method for multi-class precipitation phases, characterized in that, include: For the full spatiotemporal sample, a stepwise conditional binary classification precipitation phase probability correction algorithm is used to determine the probability of multiphase precipitation after preliminary correction. This includes: classifying all events into two categories: no precipitation and precipitation, and correcting them using a binary classification method to obtain the corrected probability of precipitation P1; classifying precipitation events into rainfall and non-rainfall events, and correcting them using a binary classification method to obtain the rainfall probability P_rain under precipitation conditions; determining the preliminary corrected probability of rainfall based on the precipitation probability and the rainfall probability; classifying the non-rainfall events into snowfall and non-solid precipitation, and correcting them using a binary classification method to obtain the snowfall probability P_snow under precipitation and non-rainfall conditions, and determining the preliminary corrected probability of rainfall based on the snowfall probability... The corrected snowfall probability is determined using the probability P_snow, the probability of precipitation P1, and the probability of rainfall P_rain. The non-solid precipitation events are then divided into sleet and freezing rain, and a binary classification correction method is used to obtain the sleet probability P_sleet and freezing rain probability P_freezing under the condition of precipitation but no rainfall or snowfall. The corrected sleet probability is determined based on the sleet probability P_sleet, the snowfall probability P_snow, the rainfall probability P_rain, and the probability of precipitation P1. Similarly, the corrected freezing rain probability is determined based on the freezing rain probability P_freezing, the snowfall probability P_snow, the rainfall probability P_rain, and the probability of precipitation P1. The spatial downscaling algorithm for precipitation phase probability variability is used to spatially downscale the probabilities of the preliminarily corrected multiphase precipitation to obtain a high-resolution precipitation phase probability forecast correction product.
2. The correction method for multi-class precipitation phases according to claim 1, characterized in that, The spatial downscaling algorithm for precipitation phase probability variability is used to perform spatial downscaling on the initially corrected probabilities of multiphase precipitation, including: Calculate the difference between the terrain height of the coarse-resolution product and the terrain height of the high-resolution product; based on the wet adiabatic lapse rate and in conjunction with the difference, calculate the temperature variability dT of the high-resolution product relative to the coarse-resolution product. Logistic regression is used to fit the probability of rain and snow at different temperatures to obtain the probability functions of rain and snow. The derivatives of the probability functions of rain and snow are obtained to obtain the variability dPr of the probability of rain and the variability dPs of the probability of snow at different temperatures. By combining the temperature variability dT, the precipitation probability variability dPr, and the snowfall probability variability dPs, as well as the high-resolution temperature forecast product, the precipitation probability correction value and the snowfall probability correction value are calculated respectively, and then the sleet probability correction value is obtained. The precipitation probability correction value, the snowfall probability correction value, and the sleet probability correction value are added to the probability of the multiphase precipitation after the initial correction, respectively, to obtain the downscaled probability of rain, snow, sleet, and freezing rain, forming the high-resolution precipitation phase probability forecast correction product.
3. The correction method for multi-class precipitation phases according to claim 1, characterized in that, Based on the probability of precipitation and the probability of rainfall, the corrected preliminary probability of rainfall is determined as follows: Based on the conditional probability formula, the final corrected preliminary probability of rainfall is calculated as P_rain×P1.
4. The correction method for multi-class precipitation phases according to claim 1, characterized in that, Based on the snowfall probability P_snow, the precipitation probability P1, and the rainfall probability P_rain, the corrected snowfall probability is determined as follows: The probability of snowfall is determined by the conditional probability formula as: P_snow×(1-P_rain)×P1.
5. The correction method for multi-class precipitation phases according to claim 1, characterized in that, The corrected probability of sleet is determined based on the probability of sleet (P_sleet), the probability of snowfall (P_snow), the probability of precipitation (P_rain), and the probability of precipitation (P1); and the corrected probability of freezing rain is determined based on the probability of freezing rain (P_freezing), the probability of snowfall (P_snow), the probability of precipitation (P_rain), and the probability of precipitation (P1), including: The final corrected probability of sleet, calculated using the conditional probability formula, is: P_sleet×(1-P_snow)×(1-P_rain)×P1, The final corrected probability of freezing rain, calculated using the conditional probability formula, is: P_freezing×(1-P_snow)×(1-P_rain)×P1.
6. A correction device for multi-class precipitation phases, characterized in that, include: The probability determination unit for multiphase precipitation is used to determine the initial corrected probability of multiphase precipitation for the full spatiotemporal sample using a stepwise conditional binary classification precipitation phase probability correction algorithm. This includes: classifying all events into two categories: no precipitation and precipitation, and correcting them using a binary classification method to obtain the corrected probability of precipitation P1; classifying precipitation events into rainfall and non-rainfall events, and correcting them using a binary classification method to obtain the rainfall probability P_rain under precipitation conditions; determining the initial corrected rainfall probability based on the precipitation probability and the rainfall probability; and classifying the non-rainfall events into snowfall and non-solid precipitation, and correcting them using a binary classification method to obtain the snowfall probability P_snow under both precipitation and non-rainfall conditions. Based on the snowfall probability P_snow, the precipitation probability P1, and the rainfall probability P_rain, the corrected snowfall probability is determined. The non-solid precipitation events are divided into sleet and freezing rain, and a binary classification correction method is used to correct them, obtaining the sleet probability P_sleet and freezing rain probability P_freezing under the condition of precipitation but no rainfall and no snowfall. The corrected sleet probability is determined based on the sleet probability P_sleet, snowfall probability P_snow, rainfall probability P_rain, and precipitation probability P1; and the corrected freezing rain probability is determined based on the freezing rain probability P_freezing, snowfall probability P_snow, rainfall probability P_rain, and precipitation probability P1. The high-resolution precipitation phase probability forecast correction unit is used to perform spatial downscaling on the probabilities of the preliminarily corrected multiphase precipitation using a precipitation phase probability variability spatial downscaling algorithm to obtain a high-resolution precipitation phase probability forecast correction product.
7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to perform the method according to any one of claims 1-5.
8. An electronic device, characterized in that, include: At least one processor; And a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to cause the at least one processor to perform the method according to any one of claims 1-5.
9. A computer program product, characterized in that, When executed by a processor, the computer program implements the method described in any one of claims 1-5.