Scholarlink

<p>Although the fifth-generation ECMWF reanalysis (ERA5) reanalysis dataset at 0.25º resolution and its land-specific product (ERA5-Land) at 0.1º resolution are widely used, they still remain too coarse for fine-scale applications. In this study, we dynamically downscaled ERA5 using the Weather Research and Forecasting (WRF) model to produce 2-km resolution rainfall data for Ho Chi Minh City, Vietnam (hereafter WRF-HCM). The model outputs were subsequently bias-corrected using quantile mapping, with daily data from 29 meteorological stations used for calibration and validation, resulting in products referred to as WRFC-HCM. Results show that, on a monthly scale, ERA5 generally outperforms WRF-HCM, WRFC-HCM, and ERA5-Land. However, on a daily scale, WRFC-HCM significantly improves accuracy, particularly in representing dry events. Seasonal probability density function analysis indicates that ERA5 overestimates light rain events, a bias substantially reduced by WRFC-HCM. For moderate and heavy rainfall (&gt; 20&#xa0;mm/day), ERA5 underestimates occurrences, while WRFC-HCM provides better representation. Lastly, in assessing extreme climate indices, ERA5 and ERA5-Land overestimate the number of wet days, leading to higher Consecutive Wet Days (CWD) values, while underestimating heavy rain events, resulting in lower values for maximum 1-day precipitation (Rx1day) and maximum consecutive 5-day precipitation (Rx5day). These biases lead to an overall underestimation of rainfall intensity. Compared to ERA5, WRFC-HCM shows improved performance by effectively reducing the number of wet days and increasing Rx1day, offering a more accurate characterization of rainfall extremes.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Performance and added value of a high-resolution (2 km) rainfall product based on WRF-downscaled ERA5 for Ho Chi Minh City, Vietnam

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月17日

查看全文
  • Tung Nguyen-Duy,
  • Thanh Ngo-Duc,
  • Marc Choisy,
  • Iago Perez Fernandez,
  • Sarah Sparrow

摘要

Although the fifth-generation ECMWF reanalysis (ERA5) reanalysis dataset at 0.25º resolution and its land-specific product (ERA5-Land) at 0.1º resolution are widely used, they still remain too coarse for fine-scale applications. In this study, we dynamically downscaled ERA5 using the Weather Research and Forecasting (WRF) model to produce 2-km resolution rainfall data for Ho Chi Minh City, Vietnam (hereafter WRF-HCM). The model outputs were subsequently bias-corrected using quantile mapping, with daily data from 29 meteorological stations used for calibration and validation, resulting in products referred to as WRFC-HCM. Results show that, on a monthly scale, ERA5 generally outperforms WRF-HCM, WRFC-HCM, and ERA5-Land. However, on a daily scale, WRFC-HCM significantly improves accuracy, particularly in representing dry events. Seasonal probability density function analysis indicates that ERA5 overestimates light rain events, a bias substantially reduced by WRFC-HCM. For moderate and heavy rainfall (> 20 mm/day), ERA5 underestimates occurrences, while WRFC-HCM provides better representation. Lastly, in assessing extreme climate indices, ERA5 and ERA5-Land overestimate the number of wet days, leading to higher Consecutive Wet Days (CWD) values, while underestimating heavy rain events, resulting in lower values for maximum 1-day precipitation (Rx1day) and maximum consecutive 5-day precipitation (Rx5day). These biases lead to an overall underestimation of rainfall intensity. Compared to ERA5, WRFC-HCM shows improved performance by effectively reducing the number of wet days and increasing Rx1day, offering a more accurate characterization of rainfall extremes.