• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to project page

2023 Fiscal Year Final Research Report

Development of a heavy rainfall prediction method combining numerical weather prediction models and deep learning methods

Research Project

  • PDF
Project/Area Number 19H02246
Research Category

Grant-in-Aid for Scientific Research (B)

Allocation TypeSingle-year Grants
Section一般
Review Section Basic Section 22040:Hydroengineering-related
Research InstitutionUniversity of Yamanashi

Principal Investigator

Souma Kazuyoshi  山梨大学, 大学院総合研究部, 准教授 (40452320)

Co-Investigator(Kenkyū-buntansha) 古屋 貴彦  山梨大学, 大学院総合研究部, 准教授 (00770835)
宮本 崇  山梨大学, 大学院総合研究部, 准教授 (30637989)
馬籠 純  山梨大学, 大学院総合研究部, 准教授 (70377597)
石平 博  山梨大学, 大学院総合研究部, 教授 (80293439)
Project Period (FY) 2019-04-01 – 2023-03-31
Keywords深層学習 / 機械学習 / 数値気象モデル / 降雨予測
Outline of Final Research Achievements

In this study, we developed a method to correct the prediction results from a numerical weather model (a computer program to predict the three-dimensional temperature and wind speed, etc., based on physical equations) using deep learning (artificial intelligence that imitates human neurons). The U-Net (an advanced artificial intelligence that imitates the optic nerve used in image analysis), a type of deep learning method, was introduced. In addition, we improved the data expansion method (a method to augment data to compensate for data shortage) when inputting numerical weather model output to the deep learning method. The results suggest that the combination of numerical weather model prediction and correction methods using U-Net can improve the prediction of heavy rainfall areas, which is important for disaster adaptation.

Free Research Field

水工学

Academic Significance and Societal Importance of the Research Achievements

数時間から1日程度先までの降水量を予測するためには,数値気象モデル(物理式に基づき3次元の気温や風速等をコンピュータで予測するプログラム)による予測が重要となる.本研究で開発した手法を用いることで,数値気象モデルのみでは降水の予測には至らない豪雨でも,降水の原因となる上昇気流や水平風速の予測結果を深層学習(人間の神経細胞を模した人工知能)へ入力して自動的に降水量予測結果を補正できる.それによって定量的な豪雨予測の信頼性が向上し,土砂・浸水危険度予測の信頼性向上とその早期避難への活用が期待でき,災害に強い都市づくりに貢献できると期待される.

URL: 

Published: 2025-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi