Scholarlink

<p>This study evaluates the quality and importance of aircraft-based observations (ABOs) assimilation in a regional Numerical Weather Prediction (NWP) model over the Indian subcontinent from 1 July to 15 August 2023 through Observing System Experiments (OSEs). The research employed the Weather Research and Forecasting (WRF) model along with the regional Gridpoint Statistical Interpolation (GSI) 3D-VAR data assimilation system to conduct two experiments: CNT (including all conventional data along with aircraft data) and EXP (mirroring CNT but excluding aircraft data). Assimilation of ABOs revealed that the maximum mean root mean square error difference (RMSD) of analysis innovations of ABO temperature and zonal wind is decreased by 39% and 34% than background innovation. CNT experiment significantly reduced the bias and root mean square error (RMSE) compared to EXP, particularly at cruise altitude during all cycles, where the assimilation of ABOs was the highest. The maximum percentage improvement in mean RMSD from the CNT experiment for surface wind (<i>U</i>), temperature (<i>T</i>), and specific humidity (<i>q</i>) at 18 UTC were 3.7%, 1.8%, and 3.5%, respectively, followed by <i>U</i> (2%) and <i>T</i> (0.6%) at 06 UTC over EXP. For upper air observations (300–200 hPa), the zonal wind component exhibits a significant improvement of approximately 8% at 18 UTC, followed by 4.2% at 06 UTC, as verified against pilot balloons or radiosondes. A case study on a deep depression (DD) that occurred over the Bay of Bengal (BoB) from 1 to 3 August 2023 highlighted significant improvements in the maximum wind speed at the 10 m level and a reduction in the direct position error of the DD track, due to enhanced analysis accuracy in CNT. The spatial verification for Day 1 reveals that CNT has lower location error, reduced RMSE, and higher correlation than EXP. Pattern error is the dominant contributor to total error for both Day 1 and Day 2. The fractional skill score (FSS) confirms CNT’s superior performance on Day 1 across all thresholds, while neither model achieves the target FSS for &gt;80 mm. The 850 hPa wind shows that CNT consistently has smaller discrepancies from observations than EXP. It provides a more reliable representation than EXP, particularly during early forecast hours. Overall, the assimilation of ABOs in a continuous cyclic mode provides more frequent adjustments of the atmospheric fields towards the actual conditions and thus leads to an improved forecast.</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

Evaluating the impact of aircraft observations on meteorological forecasts over the Indian subcontinent during southwest monsoon

  • 期刊论文

  • 发表日期：

    2025年9月20日

查看全文
  • Kumarjit Saha,
  • S Indira Rani,
  • S Desamsetti,
  • Greeshma M Mohan

摘要

This study evaluates the quality and importance of aircraft-based observations (ABOs) assimilation in a regional Numerical Weather Prediction (NWP) model over the Indian subcontinent from 1 July to 15 August 2023 through Observing System Experiments (OSEs). The research employed the Weather Research and Forecasting (WRF) model along with the regional Gridpoint Statistical Interpolation (GSI) 3D-VAR data assimilation system to conduct two experiments: CNT (including all conventional data along with aircraft data) and EXP (mirroring CNT but excluding aircraft data). Assimilation of ABOs revealed that the maximum mean root mean square error difference (RMSD) of analysis innovations of ABO temperature and zonal wind is decreased by 39% and 34% than background innovation. CNT experiment significantly reduced the bias and root mean square error (RMSE) compared to EXP, particularly at cruise altitude during all cycles, where the assimilation of ABOs was the highest. The maximum percentage improvement in mean RMSD from the CNT experiment for surface wind (U), temperature (T), and specific humidity (q) at 18 UTC were 3.7%, 1.8%, and 3.5%, respectively, followed by U (2%) and T (0.6%) at 06 UTC over EXP. For upper air observations (300–200 hPa), the zonal wind component exhibits a significant improvement of approximately 8% at 18 UTC, followed by 4.2% at 06 UTC, as verified against pilot balloons or radiosondes. A case study on a deep depression (DD) that occurred over the Bay of Bengal (BoB) from 1 to 3 August 2023 highlighted significant improvements in the maximum wind speed at the 10 m level and a reduction in the direct position error of the DD track, due to enhanced analysis accuracy in CNT. The spatial verification for Day 1 reveals that CNT has lower location error, reduced RMSE, and higher correlation than EXP. Pattern error is the dominant contributor to total error for both Day 1 and Day 2. The fractional skill score (FSS) confirms CNT’s superior performance on Day 1 across all thresholds, while neither model achieves the target FSS for >80 mm. The 850 hPa wind shows that CNT consistently has smaller discrepancies from observations than EXP. It provides a more reliable representation than EXP, particularly during early forecast hours. Overall, the assimilation of ABOs in a continuous cyclic mode provides more frequent adjustments of the atmospheric fields towards the actual conditions and thus leads to an improved forecast.