主成分分析
均方误差
后发
谱线
能量(信号处理)
线性回归
产品(数学)
还原(数学)
数学
计算机科学
模式识别(心理学)
统计
算法
人工智能
机器学习
物理
天文
几何学
作者
Bruno Tochetto Primo,Fernanda M. Achete,Aline Kaji,Nicholas Barbosa,Ángel Ramiro,Filipe Salvio
标识
DOI:10.1115/omae2023-108536
摘要
Abstract This paper describes the Wave Spectra Correction (WASCO). The WASCO methodology is based on the combination of Principal Component Analysis (PCA) and multiple linear regression between in situ data (PNBOIA) and numerical model results from hindcast (ERA5) and forecast. Analysis of the WASCO show poor performance for low spectra energy, but consistent improvement for high energy spectra. An important benefit of WASCO product is the applicability to other regions worldwide, since it is not dependent on local parameters. The corrected reanalysis and forecast show a consistent error reduction when comparing with the non corrected pairs. For the forecast, MAE is reduced by 49.4%; the total RMSE and the frequency and direction integrated RMSE are reduced by 55%, 48.6% and 22%, respectively.
科研通智能强力驱动
Strongly Powered by AbleSci AI