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

Evaluation of satellite-based rainfall estimates in the upper Blue Nile basin

  • Tamer A Gado,
  • Dina H Zamzam,
  • Yiping Guo,
  • Bakenaz A Zeidan

摘要

Abstract

Observed rainfall measurements are scarce in many areas of the Nile Basin, where gauge stations are usually sparse. Thus, it is necessary to find reliable sources, such as satellite precipitation products, which can provide uninterrupted precipitation time series. In this study, four high-resolution satellite rainfall products (TRMM-3B42RT, PERSIANN-CCS, GSMAP-NRT, and CHIRPS-V2) were evaluated against ground-based observations from 44 stations over the Upper Blue Nile Basin (UBNB). The satellite-based rainfall estimates were adjusted by two bias-correction methods: linear scaling (LS) and power transformation (PT). Precipitation estimates from the CHIRPS-V2, PERSIANN-CCS, and TRMM-3B42RT products have reliable agreement with ground measurements in estimating and detecting rainfall events. CHIRPS-V2 reveals its outstanding skill over other products, and GSMAP-NRT is the worst. Both bias-correction methods significantly improve all-product estimates, with LS better than PT. This study demonstrates the importance of evaluating and modifying satellite rainfall products for use in hydrometeorological applications in the UBNB.

Research highlights

The research evaluates four high-resolution satellite rainfall products (TRMM-3B42RT, PERSIANN-CCS, GSMAP-NRT, and CHIRPS-V2) against monthly ground-based observations from 44 stations over the Upper Blue Nile Basin (UBNB).

Two bias correction methods (linear scaling and power transformation) were applied to adjust the satellite-based rainfall estimates.

Rainfall estimates from CHIRPS-V2, PERSIANN-CCS, and TRMM-3B42RT products have reliable agreement with ground measurements in estimating and detecting rainfall events.

CHIRPS-V2 reveals its outstanding skill over other products and GSMAP-NRT is the worst.

Both bias-correction methods significantly improve all-product estimates, with linear scaling slightly better than power transformation.