A self-verification method for precipitation forecast stage accuracy based on scale effect

Abstract

The invention discloses a precipitation forecast stage accuracy self-verification method based on scale effect. It includes the following steps: S1, precipitation calculation at each scale; S2, precipitation forecast value calculation at each scale; S3, fitting accuracy calculation; S4, fitting accuracy evaluation. This method can be used to solve the problem that the actual forecast accuracy cannot be verified when the precipitation is forecasted after the model is built because the measured data has not yet appeared.

Description

Technical Field

[0001] The patent of this invention belongs to the field of hydrology, and in particular relates to a self-verification method for the stage accuracy of precipitation forecasting based on scale effect. Background Art

[0002] Precipitation refers to the phenomenon that water vapor in the air condenses and falls to the ground when it is cold. It is an important part of the hydrological cycle process. It can directly or indirectly replenish various water bodies on land. The water resources that humans can use mainly come from precipitation. Precipitation has an important impact on agricultural production, social and economic activities, human daily life, and the management and utilization of water resources. In recent years, affected by climate change and human activities, precipitation changes have shown many disordered and irregular characteristics in time and space, and natural disasters have also followed. Floods have broken out in some areas, and droughts and water shortages have occurred in some areas. This has not only brought great difficulties to human life and production, but also caused huge social and economic losses and casualties. Precipitation is used to measure the amount of precipitation in a certain area during a certain period of time. Its size is affected by many factors, that is, precipitation has obvious differences in different times, seasons and regions. Agriculture is highly dependent on precipitation, especially in arid and semi-arid areas. Precipitation can increase the total crop yield by 3%-14%. Studying the amount of precipitation during the crop growth period is of vital importance to the development of agriculture, society and economy. Due to the chaos and complexity of the precipitation formation mechanism, the temporal and spatial distribution of precipitation on a regional scale is becoming increasingly uneven, and the differences on different time scales are becoming increasingly greater. As the prediction scale is extended, the uncertainty of influencing factors continues to increase, and the difficulty of prediction becomes greater. Therefore, studying the laws of precipitation changes and accurately predicting the future development trend of precipitation are of great significance for responding to flood disasters, rationally formulating irrigation systems, and improving the efficiency of agricultural water resources utilization.

[0003] At present, the research on precipitation forecasting is deepening step by step, and the practicality of the research results is also gradually improving. However, due to the differences in regional geological conditions, climatic conditions and other aspects, the existing rainfall forecast models and theories are not very targeted, especially in the verification of precipitation forecast accuracy, there is still a lack of perfect theory. At present, the Nash coefficient (NSE), root mean square error (RMSE) and mean absolute relative error (MARE) are used to evaluate the fitting accuracy and reserved test accuracy of the model. The above model accuracy evaluation method only evaluates the accuracy of the fitting stage and the reserved test stage. After the model is built, when the precipitation is predicted externally, the actual forecast accuracy cannot be verified because the measured data has not yet appeared. Summary of the invention

[0004] The purpose of the present invention is to address the problems of the above-mentioned prior art and propose a self-verification method for the stage accuracy of precipitation forecast based on scale effect. This method can be used to solve the problem that the actual forecast accuracy cannot be verified due to the lack of measured data when the precipitation is forecasted after the model is built. The technical solution of the present invention is implemented as follows: A self-verification method for the stage accuracy of precipitation forecast based on scale effect, comprising the following steps:

[0005] Calculation of precipitation at various scales: Collect precipitation sequences for each month during the growing season Where: t is a positive integer from 0 to n; i = 5, 6, 7, 8, 9; the annual and growing season precipitation are Y p (t) and SYQ p (t), then:

[0006]

[0007] Y p =SYQ p / ρ (2)

[0008] Where: ρ is the ratio of multi-year growth period precipitation to multi-year precipitation, n is the number of years of existing hydrological precipitation series data;

[0009] S2. Calculation of precipitation forecast values ​​at each scale: The forecast values ​​of precipitation series for each month of the growing period calculated by the prediction model are The predicted precipitation in the year and growing period are According to the relationship between precipitation at different scales, we can get:

[0010]

[0011]

[0012] Where: is the sum of the precipitation forecast values ​​for each month during the growing period, is the annual precipitation calculated based on the relationship between the predicted precipitation value during the growing period and ρ;

[0013] S3. Fitting accuracy calculation: and As the measured and fitted values ​​of the annual precipitation and growth period precipitation forecast values, the Nash coefficient NSE, root mean square error RMSE and mean relative error MARE were calculated respectively;

[0014]

[0015]

[0016]

[0017] Where: N is the sequence length, is the fitted value of precipitation, x t is the observed precipitation value at time t, The average precipitation amount.

[0018] S4. Fitting accuracy evaluation: When NSE=1, RMSE=0, and MARE=0, the fitting value is exactly the same as the measured value; the closer NSE is to 1, and the closer RMSE and MARE are to 0, the better the fitting and prediction accuracy of the model.

[0019] The advantage of the present invention is that it can be used to solve the problem that the actual forecast accuracy cannot be verified because the measured data has not yet appeared when the extended forecast of precipitation is carried out after the hydrological data model is built. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flow chart of the uncertainty cause analysis method of runoff change based on random simulation. DETAILED DESCRIPTION

[0021] The following is a detailed description of the implementation scheme of the present invention in conjunction with the accompanying drawings. A precipitation forecast stage accuracy self-verification method based on scale effect, comprising the following steps:

[0022] Calculation of precipitation at various scales: Collect precipitation sequences for each month during the growing season Where: t is a positive integer from 0 to n; i = 5, 6, 7, 8, 9; the annual and growing season precipitation are Y p (t) and SYQ p (t), then:

[0023]

[0024] Y p =SYQ p / ρ (2)

[0025] Where: ρ is the ratio of multi-year growth period precipitation to multi-year precipitation, n is the number of years of existing hydrological precipitation series data;

[0026] S2. Calculation of precipitation forecast values ​​at each scale: The forecast values ​​of precipitation series for each month of the growing period calculated by the prediction model are The predicted precipitation in the year and growing period are According to the relationship between precipitation at different scales, we can get:

[0027]

[0028]

[0029] Where: is the sum of the precipitation forecast values ​​for each month during the growing period, is the annual precipitation calculated based on the relationship between the predicted precipitation value during the growing period and ρ;

[0030] S3. Fitting accuracy calculation: and As the measured and fitted values ​​of the annual precipitation and growth period precipitation forecast values, the Nash coefficient NSE, root mean square error RMSE and mean relative error MARE were calculated respectively;

[0031]

[0032]

[0033]

[0034] Where: N is the sequence length, is the fitted value of precipitation, x t is the observed precipitation value at time t, The average precipitation amount.

[0035] S4. Fitting accuracy evaluation: When NSE=1, RMSE=0, and MARE=0, the fitting value is exactly the same as the measured value; the closer NSE is to 1, and the closer RMSE and MARE are to 0, the better the fitting and prediction accuracy of the model.

Claims

1. A precipitation forecast stage accuracy self-verification method based on scale effect, characterized by: The following steps are involved: S1. Calculation of precipitation at each scale: Collect precipitation sequences for each month during the growing period Where: t is a positive integer from 0 to n; i = 5, 6, 7, 8, 9; the annual and growing season precipitation are Y p (t) and SYQ p (t), then: Y p =SYQ p / ρ (2) Where: ρ is the ratio of multi-year growth period precipitation to multi-year precipitation, n is the number of years of existing hydrological precipitation series data; S2. Calculation of precipitation forecast values ​​at each scale: The forecast values ​​of precipitation series for each month of the growing period calculated by the prediction model are The predicted precipitation in the year and growing period are According to the relationship between precipitation at different scales, we can get: Where: is the sum of the precipitation forecast values ​​for each month during the growing period, is the annual precipitation calculated based on the relationship between the predicted precipitation value during the growing period and ρ; S3. Fitting accuracy calculation: and As the measured and fitted values ​​of the annual precipitation and growth period precipitation forecast values, the Nash coefficient NSE, root mean square error RMSE and mean relative error MARE were calculated respectively; S4. Fitting accuracy evaluation: When NSE=1, RMSE=0, and MARE=0, the fitting value is exactly the same as the measured value; the closer NSE is to 1, and the closer RMSE and MARE are to 0, the better the fitting and prediction accuracy of the model.

2. The precipitation forecast stage accuracy self-verification method based on scale effect according to claim 1 is characterized by: The specific calculation formula of the Nash coefficient NSE described in S3 is as follows: Where: N is the sequence length, is the fitted value of precipitation, x t is the observed precipitation value at time t, The average precipitation amount.

3. The precipitation forecast stage accuracy self-verification method based on scale effect according to claim 1 is characterized by: The specific calculation formula of the root mean square error RMSE described in S3 is as follows: Where: N is the sequence length, is the fitted value of precipitation, x t is the observed precipitation value at time t, The average precipitation amount.

4. The precipitation forecast stage accuracy self-verification method based on scale effect according to claim 1 is characterized by: The specific calculation formula of the mean relative error MARE described in S3 is as follows: Where: N is the sequence length, is the fitted value of precipitation, x t is the observed precipitation value at time t, The average precipitation amount.

Images