Scholarlink

<p>The present study focuses on the nowcasting of convective storms over India using an in-house nowcasting model built on the Python-based pySTEPS (short-term ensemble prediction systems) framework. This model integrates advanced techniques, including the optical flow algorithm, autoregressive process, and cascade decomposition, to accurately identify advection, growth, and decay processes in storm evolution. To achieve this, reconstructed radar reflectivity has been utilized, derived from integrated observations of INSAT-3D and spaceborne radar from the Global Precipitation Mission (GPM). The model effectively captures convective storms with high reconstructed radar reflectivity, and its performance has been assessed through multiple case studies. The nowcasts were validated against observed reconstructed radar reflectivity and corresponding precipitation intensity from satellite data. For verification, key statistical metrics were computed, including the confusion matrix and skill scores such as Fractional Skill Score (FSS), Heidke Skill Score (HSS), False Alarm Ratio (FAR), Probability of Detection (POD), and Critical Success Index (CSI). The model exhibited high accuracy, achieving scores of 0.87 (FSS), 0.84 (HSS), 0.90 (POD), and 0.81 (CSI) for a 30-minute lead time at a 1 dBZ threshold, with a FAR of 0.19. Further analysis of FSS across different reflectivity bins showed consistently high scores (~ 0.8 or above), indicating the model’s ability to nowcast a wide range of reconstructed radar reflectivity levels with minimal variation in accuracy. Overall, the model exhibits strong predictive capability and holds significant potential for operational nowcasting of convective storms over India.</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

Next-gen storm nowcasting: harnessing satellite-based radar reflectivity over India

  • 期刊论文

  • 发表日期：

    2025年7月17日

查看全文
  • Anoop Kumar Mishra,
  • Abhisek Chhari,
  • Gajendra Kumar,
  • Bipasha Paul Shukla,
  • Randhir Singh,
  • Pradeep Kumar Thapliyal,
  • Sourav Adhikary,
  • Krishna Kumar Shukla

摘要

The present study focuses on the nowcasting of convective storms over India using an in-house nowcasting model built on the Python-based pySTEPS (short-term ensemble prediction systems) framework. This model integrates advanced techniques, including the optical flow algorithm, autoregressive process, and cascade decomposition, to accurately identify advection, growth, and decay processes in storm evolution. To achieve this, reconstructed radar reflectivity has been utilized, derived from integrated observations of INSAT-3D and spaceborne radar from the Global Precipitation Mission (GPM). The model effectively captures convective storms with high reconstructed radar reflectivity, and its performance has been assessed through multiple case studies. The nowcasts were validated against observed reconstructed radar reflectivity and corresponding precipitation intensity from satellite data. For verification, key statistical metrics were computed, including the confusion matrix and skill scores such as Fractional Skill Score (FSS), Heidke Skill Score (HSS), False Alarm Ratio (FAR), Probability of Detection (POD), and Critical Success Index (CSI). The model exhibited high accuracy, achieving scores of 0.87 (FSS), 0.84 (HSS), 0.90 (POD), and 0.81 (CSI) for a 30-minute lead time at a 1 dBZ threshold, with a FAR of 0.19. Further analysis of FSS across different reflectivity bins showed consistently high scores (~ 0.8 or above), indicating the model’s ability to nowcast a wide range of reconstructed radar reflectivity levels with minimal variation in accuracy. Overall, the model exhibits strong predictive capability and holds significant potential for operational nowcasting of convective storms over India.