基于RF_Adaboost神经网络的中国区域ZWD偏差改正模型

A Bias Correction Model of ZWD Based on RF_Adaboost Neural Network over China

ES评分 0

DOI 10.14075/j.jgg.2024.08.401
刊名
Journal of Geodesy and Geodynamics
年,卷(期) 2025, 45(8)
作者
作者单位

1 桂林理工大学测绘地理信息学院,桂林市雁山街319号,541006;
2 桂林理工大学旅游与风景园林学院,桂林市雁山街319号,541006

摘要
以往的中国区域ZWD模型只考虑高程、纬度、气象参数等因子的影响,未顾及季风气候和海洋水汽输送对ZWD估计的影响。针对该问题,以偏差改正为切入点,首先以探空站积分ZWD为参考值,分析2016—2019年由ERA5数据积分法和反算法得到的ERA5-ZWD估计值在中国区域的均方根误差和偏差的空间分布,然后利用2016—2018年ERA5和探空站数据建立基于RF_Adaboost神经网络的中国区域偏差改正模型。结果表明,2019年ERA5-ZWD估计值经过偏差改正后偏差大幅降低,均方根误差在中国东部沿海地区和南方地区显著降低。
Abstract
Previous ZWD models over China only consider the effects of elevation, latitude and meteorological parameters, without taking into account the impact of monsoon climate and ocean water vapor transport on ZWD estimation. To solve this problem, we take bias correction as the breakthrough point, analyze the spatial distribution of RMSE and bias of ERA5-ZWD estimation value obtained by the ERA5 data integration method and inversion method over China from 2016 to 2019. Then, the regional bias correction model over China based on RF_Adaboost neural network is established using the ERA5 and radiosonde data from 2016 to 2018. The results show that the bias of ERA5-ZWD estimation value in 2019 is significantly reduced after correction, and the RMSE is significantly reduced in the eastern coastal areas and southern regions of China.
关键词
ERA5; 天顶湿延迟; RF_Adaboost; 偏差改正
KeyWord
ERA5; zenith wet delay; RF_Adaboost; bias correction
基金项目
页码 791-795
  • 参考文献
  • 相关文献
  • 引用本文

彭祥天, 张露露, 林买金, 徐庆兵, 刘俊文, 谢劭峰. 基于RF_Adaboost神经网络的中国区域ZWD偏差改正模型 [J]. 大地测量与地球动力学. 2025; 45; (8). 791 - 795.

  • 文献评论

相关学者

相关机构