Scholarlink

<p>Predicting runoff plays a key role in flood warning systems in regions like Kendrapara district, where major rivers such as the Mahanadi, Brahmani, and Baitarani meet. These areas pose difficulties to develop a precise flood forecasting model because of their mountainous landscape and frequent flash floods. The Ensemble Framework (EF5) for flash flood forecasting shows promise as a technique without satellite rainfall data. This research applied the EF5 model to build a flood forecast system for the Kendrapara district, focusing on zones prone to recurrent flash floods. Several benchmark machine learning models namely Support Vector Machine (SVM), Random Forest (RF), Gradient boosting machines (GBM) and Long Short-Term Memory Network (LSTM) are also applied for proving effectiveness of the EF5 model. Results confirmed that the EF5 model performed well during validationwith the Nash-Sutcliffe efficiency (NSE) and Pearson’s correlation coefficient (PCC) routinely exceeding 0.82. Notably, spline interpolation yields the best results, with PCC and NSE values above 0.9,followed by Inverse Distance Weighting (IDW), which comes in second place with PCC and NSE of approximately 0.8; Kriging performs the worst, with PCC and NSE of approximately 0.4 to 0.6. Nonetheless, disparities in efficiency are observed across several interpolation techniques. Furthermore, lead and lag lengths have a considerable impact on the model’s predicted accuracy. Shorter lead times correspond to greater PCC and NSE values, which surpass 0.8. However, as the projected lead time rises, these values rapidly decrease, nearing zero or even becoming negative. Given the EF5 framework’s simplicity and validity in Kendrapara’s tiny mountainous basins, further study and verification are required to improve flood prediction skills in this region.</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

Flash Flood Forecasting Based-EF5 Model using Distinct Interpolation Methods: An Ensemble Framework

  • 期刊论文

  • 发表日期：

    2026年7月30日

查看全文
  • Abhisek Mishra,
  • Abinash Sahoo,
  • Sandeep Samantaray,
  • Deba Prakash Satapathy,
  • Sajjad Firas Abdulameer,
  • Mayadah W. Falah,
  • Zaher Mundher Yaseen

摘要

Predicting runoff plays a key role in flood warning systems in regions like Kendrapara district, where major rivers such as the Mahanadi, Brahmani, and Baitarani meet. These areas pose difficulties to develop a precise flood forecasting model because of their mountainous landscape and frequent flash floods. The Ensemble Framework (EF5) for flash flood forecasting shows promise as a technique without satellite rainfall data. This research applied the EF5 model to build a flood forecast system for the Kendrapara district, focusing on zones prone to recurrent flash floods. Several benchmark machine learning models namely Support Vector Machine (SVM), Random Forest (RF), Gradient boosting machines (GBM) and Long Short-Term Memory Network (LSTM) are also applied for proving effectiveness of the EF5 model. Results confirmed that the EF5 model performed well during validationwith the Nash-Sutcliffe efficiency (NSE) and Pearson’s correlation coefficient (PCC) routinely exceeding 0.82. Notably, spline interpolation yields the best results, with PCC and NSE values above 0.9,followed by Inverse Distance Weighting (IDW), which comes in second place with PCC and NSE of approximately 0.8; Kriging performs the worst, with PCC and NSE of approximately 0.4 to 0.6. Nonetheless, disparities in efficiency are observed across several interpolation techniques. Furthermore, lead and lag lengths have a considerable impact on the model’s predicted accuracy. Shorter lead times correspond to greater PCC and NSE values, which surpass 0.8. However, as the projected lead time rises, these values rapidly decrease, nearing zero or even becoming negative. Given the EF5 framework’s simplicity and validity in Kendrapara’s tiny mountainous basins, further study and verification are required to improve flood prediction skills in this region.