CN116151474A - Precipitation product downscaling method integrating multisource data - Google Patents
Precipitation product downscaling method integrating multisource data Download PDFInfo
- Publication number
- CN116151474A CN116151474A CN202310244992.1A CN202310244992A CN116151474A CN 116151474 A CN116151474 A CN 116151474A CN 202310244992 A CN202310244992 A CN 202310244992A CN 116151474 A CN116151474 A CN 116151474A
- Authority
- CN
- China
- Prior art keywords
- precipitation
- grid point
- intensity
- probability
- designated area
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000001556 precipitation Methods 0.000 title claims abstract description 181
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000010801 machine learning Methods 0.000 claims abstract description 7
- 238000012876 topography Methods 0.000 claims description 12
- 230000004927 fusion Effects 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000012886 linear function Methods 0.000 claims description 3
- 238000007477 logistic regression Methods 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 abstract description 5
- 238000004458 analytical method Methods 0.000 description 6
- 230000002354 daily effect Effects 0.000 description 3
- 238000007476 Maximum Likelihood Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000003203 everyday effect Effects 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Tourism & Hospitality (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Software Systems (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Game Theory and Decision Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Educational Administration (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method for reducing the scale of a precipitation product by fusing multisource data, which comprises the following steps: s10, constructing a rainfall event prediction model by using global features, local features and adjacent features of a designated area, and determining the rainfall probability of a target area in the designated area according to the rainfall event prediction model; s20, constructing a rainfall intensity prediction model by using an XGBoost machine learning model, and determining the rainfall intensity of a designated area according to the rainfall intensity prediction model; s30, fusing the precipitation probability and the precipitation intensity of the designated area to obtain a predicted value of the final precipitation intensity of the designated area. According to the method for reducing the scale of the precipitation product by fusing the multisource data, the precipitation intensity of the target grid point is obtained by fusing the predicted precipitation intensity of the target grid point and the precipitation probability, so that the prediction accuracy is improved, the occurrence of abnormal situations of low probability and high intensity is avoided, and the prediction accuracy of the precipitation intensity is improved.
Description
Technical Field
The invention belongs to the technical field of hydrology and weather, and particularly relates to a method for reducing the scale of a precipitation product by fusing multisource data.
Background
The precipitation product downscaling refers to integrating an original precipitation product with low spatial resolution into a precipitation product with high spatial resolution, wherein the original precipitation product generally adopts a method of weather re-analysis precipitation data or satellite precipitation observation data, although the spatial coverage is continuous, the spatial resolution is generally lower (> 10 km), the downscaling aims at fusing the observation data of a ground weather site and other auxiliary data (such as elevation, vegetation index and the like), the original precipitation data with low spatial resolution is processed based on a given fusion algorithm, and then the precipitation product with continuous spatial coverage and higher spatial resolution is generated, the existing fusion algorithm comprises a method of combining a probability density function with an optimal interpolation method, a geographic difference analysis method, a Bayesian fusion method, an artificial neural network, a K nearest neighbor method and the like, but the existing method is generally aimed at fusion of precipitation intensities (namely, the precipitation elements concerned are the precipitation amount of precipitation in unit time), fusion products (namely whether precipitation occurs in the given time or not) without directly giving precipitation events, and the precipitation events are fused with the precipitation intensities in the process, and in consideration of important weather events, the precipitation events are also important in the field, so that the precipitation quality is improved, and the precipitation quality is improved. In the process of specifically developing the method, the following technical difficulties are provided:
(1) The existing scale-down scheme does not combine precipitation events and precipitation intensity, and when a high-spatial-resolution precipitation intensity product is generated, the probability of precipitation is low, but the precipitation intensity is high, and the precision of the precipitation product after scale down can be seriously affected.
(2) What key features to choose to build a predictive model when fusing weather site observations, analyzing precipitation products, and other ancillary data to generate high resolution precipitation event products?
Disclosure of Invention
In order to solve the problems, the invention provides a method for reducing the scale of a precipitation product by fusing multi-source data, the precipitation intensity of a target grid point is obtained by fusing the predicted precipitation intensity of the target grid point and the precipitation probability, the prediction accuracy is improved, the occurrence of abnormal situations with low probability and high intensity is avoided, and the prediction accuracy of the precipitation intensity is improved.
In order to achieve the above purpose, the invention adopts a technical scheme that:
a method for downscaling a precipitation product incorporating multi-source data, comprising the steps of: s10, constructing a rainfall event prediction model by using global features, local features and adjacent features of a designated area, and determining the rainfall probability of a target area in the designated area according to the rainfall event prediction model; s20, constructing a rainfall intensity prediction model by using an XGBoost machine learning model, and determining the rainfall intensity of a designated area according to the rainfall intensity prediction model; s30, fusing the precipitation probability and the precipitation intensity of the designated area to obtain a predicted value of the final precipitation intensity of the designated area.
Further, step S10 includes: s11, selecting precipitation probability characteristics, namely selecting global characteristics, local characteristics and adjacent characteristics of a designated area; s12, selecting a model, and constructing a rainfall event prediction model by adopting logarithmic probability regression (logistic regression); s13, predicting the precipitation event, and obtaining the probability of the precipitation event of the target area in the designated area according to the precipitation event prediction model.
Further, the global characteristic is the precipitation intensity of a coarse grid point of a designated area, the coarse grid point space comprises a target grid point, and the time of the coarse grid point is consistent with that of the target grid point; the local features are that the target grid points are ground elevation (DEM), gradient, topography relief and vegetation index measured under corresponding topography observation data and vegetation index data; the adjacent features are real-time data of four nearest weather stations adjacent to the target grid point, wherein the real-time data comprise whether each nearest weather station adjacent to the target grid point is subjected to precipitation or not, the space distance between each nearest weather station adjacent to the target grid point, and the similarity between the local features of the grid points where each nearest weather station adjacent to the target grid point is located and the local features of given grid points;
the rainfall event prediction model is shown as a formula (1)
Wherein F (X) i,j ) As shown in formula (2)
F(X i,j )＝(w 1 +w 1,i )×x i,j,1 +(w 2 +w 2,i )×x i,j,2 +…+(w 17 +w 17,i )×x i,j,17 +(b+b i )
(2)
Wherein i is the ith day, j is the jth grid point, Y i,j Indicating whether the jth grid point occurs on the ith dayPrecipitation event, Y i,j =1 is occurrence of precipitation event, Y i,j =0 indicates that no precipitation event has occurred, F (X i,j ) Is a linear function of the corresponding grid point characteristics, x i,j,1 To x i,j,17 Features 1 to 17, w, respectively representing the jth grid point on the ith day 1 To w 17 Fixed coefficients, w, representing each feature separately 1,i To x i,j,17 Coefficients on the i-th day of features 1-17, respectively, b representing intercept, b i Represents the intercept on day i.
The step S13 includes the following steps: s131, determining global features, local features and adjacent features of target grid points in the coarse grid points as independent variables in a prediction model, and determining whether precipitation events occur to the target grid points; s132, when the target grid point generates a precipitation event, a precipitation probability calculation formula is shown as a formula (3)
P i,j Probability of precipitation event for ith and jth grid point, when precipitation probability P i,j When the temperature is more than or equal to 0.5, determining that a precipitation event occurs, and when P i,j <And 0.5, judging that no precipitation event occurs.
Further, step S20 includes: s21, selecting a precipitation intensity characteristic, wherein the precipitation intensity characteristic comprises longitude (Lon), latitude (Lat), ground elevation (DEM), gradient (Slope), topography relief (Aspect), vegetation index (NDVI) and precipitation intensity (R_GPM) of a coarse grid point precipitation product observed by a satellite after interpolation is carried out on the target grid point; s22, constructing a precipitation intensity model, and constructing a precipitation intensity model by adopting an XGBoost machine learning model, wherein the predicted precipitation intensity R of a target grid point i,j As shown in formula (4)
R i,j ＝F XGBoost (R_GPM i,j ,Lon j ,Lat j ,DEM j ,Slope j ,Aspect j ,NDVI i,j ) (4)。
Further, the precipitation intensity of the target grid point, which is the predicted value of the final precipitation intensity of the designated area in the step S30, is shown in the formula (5)
R_adjust i,j ＝P i,j (Y i,j ＝1|X i,j )×R i,j (5)。
Compared with the prior art, the invention has the beneficial effects that:
(1) According to the method for reducing the scale of the precipitation product by fusing the multi-source data, the precipitation intensity of the target grid point is obtained by fusing the predicted precipitation intensity of the target grid point and the precipitation probability, so that the prediction accuracy is improved, when the precipitation probability is approximately equal to 1, the precipitation intensity of the target grid point is equal to the predicted precipitation intensity of the target grid point, and when the precipitation probability is approximately equal to 0, the precipitation probability of the target grid point is very low, the occurrence of abnormal situations of low probability and high intensity is avoided, and the prediction accuracy of the precipitation intensity is improved.
(2) According to the method for reducing the scale of the precipitation product by fusing the multisource data, 1 global feature, 4 local features and 12 adjacent features are selected to establish the prediction model, so that the prediction accuracy is improved.
Drawings
The technical solution of the present invention and its advantageous effects will be made apparent by the following detailed description of the specific embodiments of the present invention with reference to the accompanying drawings.
FIGS. 1-2 are flow charts of methods of downscaling precipitation products incorporating multi-source data according to the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to fall within the scope of the invention.
As shown in fig. 1 to 2, for a certain grid point (spatial resolution is 1 km×1 km) on a certain day, whether a precipitation event occurs at the grid point is determined by using data such as re-analysis precipitation intensity data, weather site observation data, topography observation data, vegetation observation data, satellite precipitation intensity, etc., and the prediction model is a typical binary classification task, and mainly involves the following steps:
s10, constructing a rainfall event prediction model by using global features, local features and adjacent features of a designated area, and determining the rainfall probability of a target area in the designated area according to the rainfall event prediction model; s20, constructing a rainfall intensity prediction model by using an XGBoost machine learning model, and determining the rainfall intensity of a designated area according to the rainfall intensity prediction model; s30, fusing the precipitation probability and the precipitation intensity of the designated area to obtain a predicted value of the final precipitation intensity of the designated area.
The step S10 includes: s11, selecting precipitation probability characteristics, and selecting global characteristics, local characteristics and adjacent characteristics of a designated area. The global characteristic is the precipitation intensity of a coarse grid point of a designated area, the coarse grid point spatially contains a target grid point, the time of the coarse grid point is consistent with that of the target grid point, and the characteristic data is derived from analysis precipitation intensity data. Preferably, the spatial resolution of the coarse grid point is 10 km×10 km, and the spatial resolution of the target grid point is 1 km×1 km. The local features are that the target grid point is 4 data of ground elevation (DEM), gradient, topography relief and vegetation index measured under corresponding topography observation data and vegetation index data. The adjacent features are real-time data of four nearest weather stations adjacent to the target grid point, and the real-time data comprises whether each nearest weather station generates excessive rainfall (the excessive rainfall is represented by 1 and the excessive rainfall is not represented by 0), the space distance between each nearest weather station and the target grid point, and the similarity of the local features of the grid points where each nearest weather station is located and the local features of the given grid points.
S12, selecting a model, and constructing a rainfall event prediction model by adopting logarithmic probability regression (logistic regression). The rainfall event prediction model is shown as a formula (1)
Wherein F (X) i,j ) As shown in formula (2)
F(X i,j )＝(w 1 +w 1,i )×x i,j,1 +(w 2 +w 2,i )×x i,j,2 +…+(w 17 +w 17,i )×x i,j,17 +(b+b i ) (2)
Wherein i is the ith day, the value range of i corresponds to the time range of the data, j is the jth grid point, Y i,j Indicating whether precipitation event occurs at the jth grid point on the ith day, Y i,j =1 is occurrence of precipitation event, Y i,j =0 indicates that no precipitation event has occurred, F (X i,j ) Is a linear function of the corresponding grid point characteristics, x i,j,1 To x i,j,17 Features 1 to 17, w, respectively representing the jth grid point on the ith day 1 To w 17 Fixed coefficients, w, representing each feature separately 1,i To x i,j,17 Coefficients on the i-th day of features 1-17, respectively, b representing intercept, b i Representing the intercept on day i, these coefficients and intercepts are typically estimated using sample data in combination with maximum likelihood estimation.
S13, predicting the precipitation event, and obtaining the probability of the precipitation event of the target area in the designated area according to the precipitation event prediction model. The step S13 includes the following steps: s131, determining global features, local features and adjacent features of target grid points in the coarse grid points as independent variables in a prediction model, and determining whether precipitation events occur in the target grid points. S132, when the target grid point generates a precipitation event, a precipitation probability calculation formula is shown as a formula (3)
P i,j Probability of precipitation event for ith and jth grid point, when precipitation probability P i,j When the temperature is more than or equal to 0.5, determining that a precipitation event occurs, and when P i,j <0.5, judging that no precipitation event occurs, P i,j E (0, 1). The precipitation probability is a continuous number from 0 to 1, which can be conveniently applied to restrict the precipitation intensity.
Step S20 includes: s21, a precipitation intensity characteristic is selected, wherein the precipitation intensity characteristic comprises longitude (Lon), latitude (Lat), ground elevation (DEM), gradient (Slope), topography relief (Aspect), vegetation index (NDVI) and precipitation intensity (R_GPM) of a satellite observed coarse grid point precipitation product after interpolation is carried out on the target grid point. S22, constructing a precipitation intensity model, and constructing a precipitation intensity model by adopting an XGBoost machine learning model, wherein the predicted precipitation intensity R of a target grid point i,j As shown in formula (4)
R i,j ＝F XGBoost (R_GPM i,j ,Lon j ,Lat j ,DEM j ,Slope j ,Aspect j ,NDVI i,j ) (4)。
R i,j Represents the precipitation intensity of the grid point of the ith and jth grid points.
The predicted value of the final precipitation intensity of the designated area in the step S30, i.e., the precipitation intensity of the target grid point, is shown in the formula (5)
R_adjust i,j ＝P i,j (Y i,j ＝1|X i,j )×R i,j (5)。
For each 1 km×1 km grid point, the precipitation intensity is predicted by using the formula (4), the precipitation probability is predicted by using the formula (3), and the prediction result can be divided into four situations: scenario 1 is low probability and high intensity, i.e. precipitation events are unlikely to occur at grid points, but the precipitation intensity prediction value is very large; scenario 2 is low probability and low intensity, i.e. the grid points are unlikely to generate precipitation events, and the predicted value of the precipitation intensity is smaller; the situation 3 is high probability and high strength, namely the grid points are likely to generate precipitation events, and the precipitation strength predicted value is larger; scenario 4 is high probability and low intensity, i.e. precipitation events are likely to occur at grid points, but the precipitation intensity predictions are small. For scenario 3 and scenario 4, the predicted value of the precipitation intensity does not need to be constrained, while for scenario 1 and scenario 2, the precipitation intensity needs to be constrained. Under scenario 3 and scenario 4 conditions, P i,j Approximately equal to 1, R_adjust i,j Approximate R i,j While under scenario 1 and scenario 2 conditions, P i,j Approximately equal to 0, the precipitation prediction value R i,j The method reduces to a lower level, particularly aims at the abnormal situation of the scene 1, and avoids the abnormal situation with low probability and high strength.
Example 1
Specifically, a region of 2 degrees×2 degrees in Sichuan province is selected as an example, the specific latitude range is 32-34 degrees, the longitude range is 103-105 degrees, and the date is 8 months in 2020, and the steps of the specific embodiment are as follows: (1) The ground elevation Data (DEM) in the area range is from a weather detection data center in Sichuan province, the data space resolution is 1 km×1 km, the data are static data, and the gradient and topography relief data are obtained by calculating the DEM data without considering time change. Satellite precipitation observation data and vegetation index data are both from the data set disclosed by NASA in the united states, wherein the satellite precipitation observation data are IMERG Final Precipitation L V06, the vegetation index is from the MOD13A3 data set, and the vegetation index data have a spatial resolution of 1 km×1 km, which is a monthly average. The satellite precipitation observation data has a spatial resolution of 10 km×10 km and a temporal resolution of day. Precipitation re-analysis data are from the fifth generation ERA5-Land analysis data of the european medium weather forecast center (ECMWF), the spatial resolution is 10 km×10 km, the temporal resolution is hours, and the data are summed and integrated into daily precipitation intensity. The weather site position and weather site precipitation observation data come from weather detection data centers in Sichuan province. For each meteorological site, other data are subjected to space-time matching, in addition, 4 nearest meteorological sites are searched, whether precipitation occurs, site distance and similarity information of the 4 sites are extracted, a sample data set for modeling is finally constructed, and partial example data are given in table 1.
TABLE 1 partial sample data for two weather stations at day 8 month 10
(2) In the process of rainfall event prediction, at least 20 observation samples of the stations are contained every day, at least 15 days of effective observation data are available in 8 months, when coefficients in a logarithmic probability regression model are specifically calculated, the observation samples of the whole month (the sample size is 960) are adopted, but when the daily scale coefficients in the model are further calculated, the calculation of the daily scale coefficients cannot be effectively supported due to the small number of the stations by the sample size of the stations every day, so that when the rainfall event prediction model is constructed by adopting the logarithmic probability regression, only fixed coefficients in the model are considered. Bringing 960 samples into equations (1) and (2), and estimating the fixed coefficients and the intercept in equation (2) by combining the maximum likelihood estimation method, wherein the intercept is 25 in the embodiment, and the fixed coefficients corresponding to the global feature (1), the local feature (4) and the adjacent feature (12) are respectively: -0.08, -0.66, -0.001, -0.34, 0.005, -2.05, 0.41, -11.08, -1.39, -3.18, 1.33, -0.65, 3.57, -11.19, -0.42, -0.80, -1.09.
(3) And calculating the precipitation probability by using the coefficients calculated in the steps and combining the formula (3), wherein the predicted value of the precipitation probability of each station on 8 months and 10 days in 2020 is shown in Table 2.
(4) The longitude (Lon), latitude (Lat), ground elevation (DEM), slope (Slope), topography relief (Aspect), vegetation index (NDVI) and satellite observed coarse grid point precipitation product are selected to interpolate the target grid point precipitation intensity (r_gpm), wherein the individual sample data are shown in table 1.
(5) The predicted precipitation intensity is obtained according to formula (4), as shown in table 2.
(6) The predicted value of the final precipitation intensity is obtained according to formula (5), as shown in table 2.
TABLE 2 weather station precipitation observation data and precipitation probability, intensity forecast (8 months, 10 days 2020, latitude ranges from 32 to 34, longitude ranges from 103 to 105.)
The foregoing description is only exemplary embodiments of the present invention and is not intended to limit the scope of the present invention, and all equivalent structures or equivalent processes using the descriptions and the drawings of the present invention or directly or indirectly applied to other related technical fields are included in the scope of the present invention.
Claims (7)
1. A method for downscaling a precipitation product incorporating multi-source data, comprising the steps of:
s10, constructing a rainfall event prediction model by using global features, local features and adjacent features of a designated area, and determining the rainfall probability of a target area in the designated area according to the rainfall event prediction model;
s20, constructing a rainfall intensity prediction model by using an XGBoost machine learning model, and determining the rainfall intensity of a designated area according to the rainfall intensity prediction model;
s30, fusing the precipitation probability and the precipitation intensity of the designated area to obtain a predicted value of the final precipitation intensity of the designated area.
2. The method of downscaling a precipitation product incorporating multisource data according to claim 1, wherein step S10 comprises:
s11, selecting precipitation probability characteristics, namely selecting global characteristics, local characteristics and adjacent characteristics of a designated area;
s12, selecting a model, and constructing a rainfall event prediction model by adopting logarithmic probability regression (logistic regression);
s13, predicting the precipitation event, and obtaining the probability of the precipitation event of the target area in the designated area according to the precipitation event prediction model.
3. The method of downscaling a precipitation product incorporating multisource data according to claim 2, wherein,
the global characteristic is the precipitation intensity of a coarse grid point of a designated area, the coarse grid point space comprises a target grid point, and the time of the coarse grid point is consistent with that of the target grid point;
the local features are that the target grid points are ground elevation (DEM), gradient, topography relief and vegetation index measured under corresponding topography observation data and vegetation index data;
the adjacent features are real-time data of four nearest weather stations adjacent to the target grid point, and the real-time data comprises whether each nearest weather station adjacent to the target grid point is subjected to precipitation or not, the spatial distance between each nearest weather station adjacent to the target grid point, and the similarity of the local features of the grid point where each nearest weather station adjacent to the target grid point is located and the local features of the given grid point.
4. A method of downscaling a precipitation product incorporating multisource data according to claim 3, wherein the precipitation event prediction model is as shown in equation (1)
Wherein F (X) i，j ) As shown in formula (2)
F(X i，j )＝(w 1 +w 1，i )×x i，j，1 +(w 2 +w 2，i )×x i，j，2 +...+(w 17 +w 17，i )×x i，j，17 +(b+b i ) (2)
Wherein i is the ith day, j is the jth grid point, Y i，j Indicating whether precipitation event occurs at the jth grid point on the ith day, Y i，j =1 is occurrence of precipitation event, Y i，j =0 indicates that no precipitation event has occurred, F (X i，j ) Is a linear function of the corresponding grid point characteristics, x i，j，1 To x i，j，17 Features 1 to 17, w, respectively representing the jth grid point on the ith day 1 To w 17 Fixed coefficients, w, representing each feature separately 1，i To x i，j，17 The coefficients of the ith day of the 1 st to 17 th features respectively,b represents the intercept, b i Represents the intercept on day i.
5. The method for downscaling a precipitation product incorporating multi-source data according to claim 4, wherein said S13 comprises the steps of:
s131, determining global features, local features and adjacent features of target grid points in the coarse grid points as independent variables in a prediction model, and determining whether precipitation events occur to the target grid points;
s132, when the target grid point generates a precipitation event, a precipitation probability calculation formula is shown as a formula (3)
P i，j Probability of precipitation event for ith and jth grid point, when precipitation probability P i，j When the temperature is more than or equal to 0.5, determining that a precipitation event occurs, and when P i，j And when the temperature is less than 0.5, judging that no precipitation event occurs.
6. The method of downscaling a precipitation product incorporating multi-source data according to claim 5, wherein step S20 comprises:
s21, selecting a precipitation intensity characteristic, including longitude (Lon), latitude (Lat), ground elevation (DEM), gradient (Slope), topography relief (Aspect), vegetation index (NDVI) and precipitation intensity (R_GPM) after interpolation of satellite observed coarse grid point precipitation products onto a target grid point
S22, constructing a precipitation intensity model, and constructing a precipitation intensity model by adopting an XGBoost machine learning model, wherein the predicted precipitation intensity R of a target grid point i，j As shown in formula (4)
R i，j ＝F XGBoost (R_GPM i，j ，Lon j ，Lat j ，DEM j ，Slope j ，Aspect j ，NDVI i，j ) (4)。
7. The method for downscaling a precipitation product with multi-source data fusion according to claim 6, wherein the final precipitation intensity of the designated area in step S30 is predicted as the target grid point as shown in formula (5)
R_adjust i，j ＝P i，j (Y i，j ＝1|X i，j )×R i，j (5)。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211572343 | 2022-12-08 | ||
| CN2022115723436 | 2022-12-08 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116151474A true CN116151474A (en) | 2023-05-23 |
Family
ID=86354440
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310244992.1A Pending CN116151474A (en) | 2022-12-08 | 2023-03-15 | Precipitation product downscaling method integrating multisource data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116151474A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117349795A (en) * | 2023-12-04 | 2024-01-05 | 水利部交通运输部国家能源局南京水利科学研究院 | Precipitation fusion method and system based on ANN and GWR coupling |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111159640A (en) * | 2019-11-20 | 2020-05-15 | 北京玖天气象科技有限公司 | Small rain emptying method, system, electronic equipment and storage medium suitable for grid forecast |
| CN112800634A (en) * | 2021-04-07 | 2021-05-14 | 水利部交通运输部国家能源局南京水利科学研究院 | Rainfall estimation method and system coupling dry-wet state identification and multi-source information fusion |
| CN115203934A (en) * | 2022-07-12 | 2022-10-18 | 南京师范大学 | Mountain area water-reducing downscaling method based on Logistic regression |
-
2023
- 2023-03-15 CN CN202310244992.1A patent/CN116151474A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111159640A (en) * | 2019-11-20 | 2020-05-15 | 北京玖天气象科技有限公司 | Small rain emptying method, system, electronic equipment and storage medium suitable for grid forecast |
| CN112800634A (en) * | 2021-04-07 | 2021-05-14 | 水利部交通运输部国家能源局南京水利科学研究院 | Rainfall estimation method and system coupling dry-wet state identification and multi-source information fusion |
| CN115203934A (en) * | 2022-07-12 | 2022-10-18 | 南京师范大学 | Mountain area water-reducing downscaling method based on Logistic regression |
Non-Patent Citations (1)
| Title |
|---|
| 张钧民 等: "基于XGBoost的多源降水数据融合方法研究", 《热带地理》, vol. 41, no. 4, pages 1 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117349795A (en) * | 2023-12-04 | 2024-01-05 | 水利部交通运输部国家能源局南京水利科学研究院 | Precipitation fusion method and system based on ANN and GWR coupling |
| CN117349795B (en) * | 2023-12-04 | 2024-02-02 | 水利部交通运输部国家能源局南京水利科学研究院 | Precipitation fusion method and system based on ANN and GWR coupling |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111859800B (en) | Space-time estimation and prediction method for PM2.5 concentration distribution | |
| Shahabi et al. | Landslide susceptibility mapping using GIS-based statistical models and Remote sensing data in tropical environment | |
| Shahabi et al. | Remote sensing and GIS-based landslide susceptibility mapping using frequency ratio, logistic regression, and fuzzy logic methods at the central Zab basin, Iran | |
| Kuhnert et al. | Incorporating uncertainty in gully erosion calculations using the random forests modelling approach | |
| Tabari et al. | Comparison of artificial neural network and combined models in estimating spatial distribution of snow depth and snow water equivalent in Samsami basin of Iran | |
| Talchabhadel et al. | Evaluating the rainfall erosivity (R-factor) from daily rainfall data: an application for assessing climate change impact on soil loss in Westrapti River basin, Nepal | |
| Antonanzas et al. | Solar irradiation mapping with exogenous data from support vector regression machines estimations | |
| CN111260111A (en) | Runoff forecast improvement method based on meteorological big data | |
| Arowolo et al. | Comparison of spatial interpolation techniques to generate high‐resolution climate surfaces for Nigeria | |
| Li et al. | Possible impact of climate change on apple yield in Northwest China | |
| Beesley et al. | A comparison of the BAWAP and SILO spatially interpolated daily rainfall datasets | |
| CN112861072B (en) | Satellite-ground multi-source rainfall self-adaptive dynamic fusion method | |
| Almasi et al. | Assessment of the climate change impacts on flood frequency (case study: Bazoft Basin, Iran) | |
| CN113901384A (en) | Ground PM2.5 concentration modeling method considering global spatial autocorrelation and local heterogeneity | |
| Kumar et al. | Climate change projections of rainfall and its impact on the cropland suitability for rice and wheat crops in the Sone river command, Bihar | |
| CN116151474A (en) | Precipitation product downscaling method integrating multisource data | |
| CN116484189A (en) | ERA5 precipitation product downscaling method based on deep learning | |
| CN117349795B (en) | Precipitation fusion method and system based on ANN and GWR coupling | |
| Rahimi et al. | An uncertainty-based regional comparative analysis on the performance of different bias correction methods in statistical downscaling of precipitation | |
| Lange et al. | Small area estimation of poverty under structural change | |
| CN114781501A (en) | Multi-source precipitation fusion method based on principal component regression | |
| Wubneh et al. | Meteorological and hydrological drought monitoring and trend analysis for selected gauged watersheds in the Lake Tana basin, Ethiopia: under future climate change impact scenario | |
| Alessandrini et al. | Estimating net migration at high spatial resolution | |
| Tasan et al. | Leveraging GNSS tropospheric products for machine learning-based land subsidence prediction | |
| Senamaw et al. | Impacts of land-use/land-cover changes on water-borne soil erosion using geospatial technologies and RUSLE Model over Chimbel Watershed of Upper Blue Nile Basin in Ethiopia |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |