CN-122018045-A - Single-point rainfall forecast optimization method and system with multiple time periods independently calibrated

CN122018045ACN 122018045 ACN122018045 ACN 122018045ACN-122018045-A

Abstract

The invention provides a single-point rainfall forecast optimization method and a system for multi-period independent calibration, which comprise the steps of obtaining forecast data and observation data and preprocessing to obtain forecast observation matching samples; dividing the forecast observation matching sample into a plurality of independent time periods, setting a cold start period, setting an attenuation coefficient, dynamically updating the forecast CDF and the observation CDF, presetting a multi-gradient precipitation threshold value and calculating to obtain a corresponding optimization threshold value, and carrying out interpolation calculation on the forecast rainfall intensity to obtain the optimized rainfall intensity. The method has the advantages of independent optimization according to time periods, accurate adaptation of different time-efficient error characteristics, comprehensive improvement of prediction accuracy of each time period, CDF initialization through sample accumulation in a cold start stage without dependence on prior parameters, multi-step preprocessing to eliminate data space-time difference, guarantee of prediction and observation data space matching, avoidance of sample dislocation, attenuation coefficient association of attenuation windows, dynamic allocation of new and old data weights, and consideration of stability of historical data and timeliness of recent data.

Inventors

ZHANG PEIMING
Guo Yinxiu
LIU YONGCHENG
CHEN JUNFENG
XU XIN
LI FENGHUI
WANG PENGCHENG
LI XINGGUO

Assignees

天津云遥宇航科技有限公司
无锡云遥宇航气象科技有限公司
北京云遥宇航科技有限公司
上海云遥宇航气象科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260331

Claims (8)

1. The single-point rainfall forecast optimization method for multi-period independent calibration is characterized by comprising the following steps of: Acquiring forecast data and observation data and preprocessing to obtain a forecast observation matching sample; dividing the forecast observation matching sample into a plurality of independent time periods; setting a cold start period, accumulating the forecast observation matching samples of each independent period, and respectively calculating a daily forecast CDF and a daily observation CDF, wherein when the cold start period reaches the standard, the forecast CDF and the observation CDF of each independent period are obtained by calculating the average value; Setting attenuation coefficients, and dynamically updating the forecast CDF and the observation CDF; presetting a multi-gradient precipitation threshold value, and determining a corresponding optimization threshold value through probability matching of the forecast CDF and the observation CDF; And carrying out interpolation calculation on the forecast rainfall intensity based on the multi-gradient rainfall threshold and the optimization threshold to obtain the optimized rainfall intensity.
2. The multi-period independently calibrated single point precipitation forecast optimization method of claim 1, wherein the preprocessing comprises the steps of: A target site is taken as a center, a rectangular area taking an observation distance as a radius is defined, and the forecast data and the observation data are cut; carrying out space-time alignment on the forecast data and the observed data to eliminate grid dislocation; Carrying out space resampling on the forecast data and the observed data through a linear interpolation algorithm to ensure that the grid quantity is consistent; and positioning grid points around the target site as calculation input of bilinear interpolation, and calculating a forecast value and an observation value of the target site position to form the forecast observation matching sample.
3. The multi-period independently calibrated single point precipitation forecast optimization method of claim 1, wherein setting attenuation coefficients, dynamically updating the forecast CDF and the observed CDF comprises the steps of: Presetting an attenuation window, and calculating according to the attenuation window to obtain the attenuation coefficient; Based on a day forecast observation matching sample, the calculating the day forecast CDF and the day observation CDF for the day; Setting historical weight and recent weight, and respectively updating the forecast CDF and the observation CDF in a dynamic weighting mode.
4. The method for optimizing single-point rainfall forecast of multi-period independent calibration of claim 1, wherein a correction factor is introduced into the calculation of the optimized rainfall intensity, and a range interval of the correction factor is set.
5. The utility model provides a single-point precipitation forecast optimizing system of independent calibration of multispeed, which characterized in that includes: the data processing module is used for acquiring forecast data and observation data and preprocessing the forecast data and the observation data to obtain a forecast observation matching sample; The multi-period dividing module comprises a plurality of exclusive optimizing units and is used for dividing the forecast observation matching samples according to forecast aging; the cold start processing module is used for setting a cold start period and accumulating the forecast observation matching samples in the period to calculate daily forecast CDF and daily observation CDF; The dynamic updating module is used for setting attenuation coefficients and updating the forecast CDF and the observation CDF of each exclusive optimizing unit; And the precipitation optimization module is used for setting gradient precipitation thresholds, determining corresponding optimization thresholds through probability matching of the forecast CDF and the observation CDF, calculating correction factors through logarithmic linear interpolation, and outputting optimized rainfall intensity.
6. The multi-period independent calibrated single point precipitation forecast optimization system of claim 5, further comprising an output monitoring module for performing optimization accuracy verification, anomaly monitoring, and periodic maintenance.
7. The multi-period independently calibrated single point precipitation forecast optimization system of claim 5, wherein the proprietary parameters of the proprietary optimization unit include a period name, a current forecast CDF, a current observed CDF, a historical sample set, a cold start status, and cold start CDF data.
8. The system of claim 5, wherein the dynamic update module sets a sample upper limit for each of the dedicated optimization units to limit the total amount of historical samples, and automatically deletes the earliest matching sample of the forecast observation beyond the sample upper limit.

Description

Single-point rainfall forecast optimization method and system with multiple time periods independently calibrated Technical Field The invention belongs to the technical field of rainfall prediction, and particularly relates to a single-point rainfall prediction optimization method and system with multiple time periods for independent calibration. Background The numerical weather forecast is the core of modern weather service, global numerical forecast products of European middle weather forecast center (ECMWF) are widely used for forecasting weather elements such as precipitation due to wide space coverage and long forecast time, and Chinese regional precipitation analysis products (CMPA) of China weather bureau become key observation data sources for correcting precipitation forecast deviation and verifying effect with high spatial resolution and hour-by-hour update. The single-point rainfall prediction is a core requirement of weather service, and directly affects decisions in the fields of agriculture, flood control, traffic and the like, but the ECMWF rainfall prediction has systematic deviations such as small rain leakage report, heavy rain misreport and the like, and is based on the mode parameterization scheme, initial field error and characteristic influence of underlying surfaces, and the characteristic difference of different prediction aging errors is obvious. The existing rainfall prediction optimization technology has the defects of insufficient generalization of multi-period optimization, no independent mechanism designed for different time-efficiency errors, low optimization reliability due to lack of historical samples in a cold start stage, stiff historical data updating mechanism, incapability of responding to weather seasonal changes, single rainfall intensity threshold design and the like. Therefore, there is a need for a precipitation prediction optimization method that solves the above-mentioned problems. Disclosure of Invention In order to solve the technical problems, the invention provides a single-point precipitation prediction optimization method and system for multi-period independent calibration, which are particularly suitable for single-point precipitation prediction optimization for multi-period independent calibration. The technical scheme adopted by the invention is that in the first aspect, the single-point rainfall forecast optimization method for multi-period independent calibration is provided, and comprises the following steps: Acquiring forecast data and observation data and preprocessing to obtain a forecast observation matching sample; dividing the forecast observation matching sample into a plurality of independent time periods; setting a cold start period, accumulating the forecast observation matching samples of each independent period, and respectively calculating a daily forecast CDF and a daily observation CDF, wherein when the cold start period reaches the standard, the forecast CDF and the observation CDF of each independent period are obtained by calculating the average value; Setting attenuation coefficients, and dynamically updating the forecast CDF and the observation CDF; presetting a multi-gradient precipitation threshold value, and determining a corresponding optimization threshold value through probability matching of the forecast CDF and the observation CDF; And carrying out interpolation calculation on the forecast rainfall intensity based on the multi-gradient rainfall threshold and the optimization threshold to obtain the optimized rainfall intensity. Further, the pretreatment comprises the following steps: A target site is taken as a center, a rectangular area taking an observation distance as a radius is defined, and the forecast data and the observation data are cut; carrying out space-time alignment on the forecast data and the observed data to eliminate grid dislocation; Carrying out space resampling on the forecast data and the observed data through a linear interpolation algorithm to ensure that the grid quantity is consistent; and positioning grid points around the target site as calculation input of bilinear interpolation, and calculating a forecast value and an observation value of the target site position to form the forecast observation matching sample. Further, setting an attenuation coefficient, and dynamically updating the forecast CDF and the observation CDF includes the following steps: Presetting an attenuation window, and calculating according to the attenuation window to obtain the attenuation coefficient; Based on a day forecast observation matching sample, the calculating the day forecast CDF and the day observation CDF for the day; Setting historical weight and recent weight, and respectively updating the forecast CDF and the observation CDF in a dynamic weighting mode. Further, a correction factor is introduced into the calculation of the optimized rainfall intensity, and a range interval of the correction factor is set. I