Quantitative precipitation estimation method based on hurdle-imdl framework

By using the Hurdle-IMDL framework and the improved U-Net network, the underfitting problem of quantitative precipitation estimation models when precipitation distribution is unbalanced is solved, and effective inversion and accurate estimation of extreme precipitation are achieved.

CN120972290BActive Publication Date: 2026-01-16NANJING METEOROLOGICAL SCI & TECH INNOVATION RES INST
2 Cites 0 Cited by

Patent Information

Application Number
CN202511501837.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-10-21
Publication Date
2026-01-16
Estimated Expiration
2045-10-21

AI Technical Summary

Technical Problem

Existing quantitative precipitation estimation models struggle to effectively invert strong to extreme precipitation when label distributions are unbalanced, especially when precipitation distributions are unbalanced. Conventional methods suffer from underfitting and misestimation problems.

Method used

The Hurdle-IMDL framework is adopted, the Hurdle model is used to handle the zero-inflation problem, and the IMDL method is used to transform the biased rainfall estimation model into an ideal model. The improved U-Net network is used for training to construct an ideal quantitative precipitation estimation probability model, and the parameters are optimized using the negative log-likelihood function.

Benefits of technology

It achieves automatic correction of the tendency to underestimate heavy rainfall during the training phase, outputs debiased pixel-level precipitation intensity, improves the reliability of extreme precipitation detection and intensity estimation, and has quantitative precipitation estimation results with complete spatial details, strong real-time performance and good generalization ability.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The application discloses a kind of quantitative precipitation estimation methods based on Hurdle-IMDL framework, comprising: obtaining historical precipitation measurement data and meteorological satellite observation data;Based on Hurdle model, construct biased quantitative precipitation estimation probability model, revise the biased rainfall estimation probability submodel according to the IMDL method, into experience distribution to construct experience quantitative precipitation estimation probability model and derive its negative log-likelihood function;Build AI model, use negative log-likelihood function as loss function to optimize AI model;Meteorological satellite observation data of the region to be inverted is input into the AI model trained, to obtain the estimated value of the parameter of experience quantitative precipitation estimation probability model, then estimate precipitation according to conditional expectation.The application uses Hurdle model to solve zero inflation problem, and uses IMDL learning method to deal with long tail problem, to improve the inversion accuracy of strong to extreme precipitation.
Need to check novelty before this filing date? Find Prior Art