Scholarlink

<p>This study analyzed the drought characteristics of the Great Zab River via the standardized streamflow index over 6- and 12-month time scales between 1990 and 2021. The objective of this study was to analyze drought characteristics in terms of severity, frequency, and duration, as well as to identify the optimal model for prediction. The results indicated that the highest drought severity at the 12-month scale occurred in February 2008, with an SSI value of − 2.11. The average frequency of drought was 54%. Conversely, the minimum drought intensity occurred from October to December 2002. The longest duration of occurrence was from April 2007 to April 2013, with a total duration of 65.3 months. In this context, nine droughts occurred, with an average duration of 22.3 months. For the 6-month index, the highest drought intensity occurred in January 1999, with a value of 4.1. The frequency of drought increased to 80.5%, resulting in 17 events. The longest drought duration in this index lasted 122.9 months, from October 2006 to April 2013. One of the driest years according to the SSI-6 index was 2010, which is in agreement with results derived from climate drought studies. Traditional models were used to forecast future droughts via the autoregressive integrated moving average (ARIMA) model and two exponential smoothing models: the Brown linear trend model and the Damped trend model. The accuracy of the SSI-6 results from the ARIMA model of the order [3,0,11] was extremely high, with a coefficient of determination of 92.4%. Statistical tests revealed that this model was the best amongst all the models. The damped trend model of SSI-6 had a lower efficiency than ARIMA did. In the SSI-12 index, the application of models presents an ARIMA of [1,1,1], which matches real data, with 87% as the coefficient of determination. Comparing the models between the conventional ARIMA method and the exponential smoothing model, the best result was obtained from the ARIMA model. The ARIMA model performed better in SSI-12 than in SSI-6, with 3% improvement in R<sup>2</sup> in SSI-12 during the validation, indicating higher accuracy and stability. SSI-12 is a more reliable drought forecast since its data span is longer. The final comparison was performed with findings from another study via a hybrid artificial intelligence (AI) model and K-nearest neighbours. K-means clustering performed better than the AI hybrid model did. Traditional models such as ARIMA outperform other models in hydrological drought predictions.</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

Hydrological drought characteristics prediction using traditional models for the Great Zab River in Iraq

  • 期刊论文

  • 发表日期：

    2025年5月6日

查看全文
  • Omar M. A. Mahmood Agha,
  • Khansaa Abdulelah Ahmed,
  • Alaa I. Naser

摘要

This study analyzed the drought characteristics of the Great Zab River via the standardized streamflow index over 6- and 12-month time scales between 1990 and 2021. The objective of this study was to analyze drought characteristics in terms of severity, frequency, and duration, as well as to identify the optimal model for prediction. The results indicated that the highest drought severity at the 12-month scale occurred in February 2008, with an SSI value of − 2.11. The average frequency of drought was 54%. Conversely, the minimum drought intensity occurred from October to December 2002. The longest duration of occurrence was from April 2007 to April 2013, with a total duration of 65.3 months. In this context, nine droughts occurred, with an average duration of 22.3 months. For the 6-month index, the highest drought intensity occurred in January 1999, with a value of 4.1. The frequency of drought increased to 80.5%, resulting in 17 events. The longest drought duration in this index lasted 122.9 months, from October 2006 to April 2013. One of the driest years according to the SSI-6 index was 2010, which is in agreement with results derived from climate drought studies. Traditional models were used to forecast future droughts via the autoregressive integrated moving average (ARIMA) model and two exponential smoothing models: the Brown linear trend model and the Damped trend model. The accuracy of the SSI-6 results from the ARIMA model of the order [3,0,11] was extremely high, with a coefficient of determination of 92.4%. Statistical tests revealed that this model was the best amongst all the models. The damped trend model of SSI-6 had a lower efficiency than ARIMA did. In the SSI-12 index, the application of models presents an ARIMA of [1,1,1], which matches real data, with 87% as the coefficient of determination. Comparing the models between the conventional ARIMA method and the exponential smoothing model, the best result was obtained from the ARIMA model. The ARIMA model performed better in SSI-12 than in SSI-6, with 3% improvement in R2 in SSI-12 during the validation, indicating higher accuracy and stability. SSI-12 is a more reliable drought forecast since its data span is longer. The final comparison was performed with findings from another study via a hybrid artificial intelligence (AI) model and K-nearest neighbours. K-means clustering performed better than the AI hybrid model did. Traditional models such as ARIMA outperform other models in hydrological drought predictions.