A novel regional forecastable multiscalar standardized drought index (RFMSDI) for regional drought monitoring and assessment
Drought is a complex recurrent natural phenomenon. It is the main outcome of climate change. It has long-term impacts on agriculture, human life as well as the environment. Therefore, quantifying drought at the regional level is essential for developing sustainable policies. This study introduced a...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-03-01
|
| Series: | Agricultural Water Management |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S0378377425000034 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832586829394608128 |
|---|---|
| author | Aamina Batool Veysi Kartal Zulfiqar Ali Miklas Scholz Farman Ali |
| author_facet | Aamina Batool Veysi Kartal Zulfiqar Ali Miklas Scholz Farman Ali |
| author_sort | Aamina Batool |
| collection | DOAJ |
| description | Drought is a complex recurrent natural phenomenon. It is the main outcome of climate change. It has long-term impacts on agriculture, human life as well as the environment. Therefore, quantifying drought at the regional level is essential for developing sustainable policies. This study introduced a new drought index for regional drought forecasting called the Regional Forecastable Multiscalar Standardized Drought Index (RFMSDI). The RFMSDI methodology is based on Forecastable Component Analysis (FCA) and K-Component Gaussian Mixture Distribution (K-CGMD). FCA reduce dimension by focus on components that are inherently more predictable. It ensures that reduced data has a built-in ability to predict future trends by selecting the maximized forecastable components. K-CGMD is utilized to model the multimodel time series data. The study application incorporates eight meteorological stations in Türkiye's Elazig province (Baskil, Agin, Elazig, Karakocan, Keban Maden, Palu and Sivrice). The effectiveness of RFMSDI is evaluated by analyzing precipitation data over these meteorological stations of Türkiye. The comparative assessment of the research signifies the superiority of FCA for regional data aggregation. In this research, the comparative assessment of RFMSDI is evaluated against the Standardized Precipitation Index (SPI) by analyzing Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) metrics across different time scales using various machine learning and traditional time series models. The research findings include the following: 1) K-CGMD is a better fitting approach for standardizing RFMSDI and SPI based on reduced BIC values. 2) RFMSDI has superior performance over SPI based on the lower values of RMSE and MAE. 3) Both machine learning and classical methods reveal that RFMSDI outperforms SPI in predicting droughts. 4) SPI shows localized advantages with the ELM training set at 1- and 6-month time scales but RFMSDI offers a more comprehensive and consistent tool for drought prediction, especially when tested on unseen data. In general, the findings endorse the effectiveness of RFMSDI for monitoring drought on a regional level. |
| format | Article |
| id | doaj-art-ebe42b70242241b7afd0bb3dfe00a967 |
| institution | Kabale University |
| issn | 1873-2283 |
| language | English |
| publishDate | 2025-03-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Agricultural Water Management |
| spelling | doaj-art-ebe42b70242241b7afd0bb3dfe00a9672025-01-25T04:10:46ZengElsevierAgricultural Water Management1873-22832025-03-01308109289A novel regional forecastable multiscalar standardized drought index (RFMSDI) for regional drought monitoring and assessmentAamina Batool0Veysi Kartal1Zulfiqar Ali2Miklas Scholz3Farman Ali4College of Statistical Sciences, University of the Punjab, Lahore, PakistanCivil Engineering, Engineering Faculty, Siirt University, Siirt, TurkeyCollege of Statistical Sciences, University of the Punjab, Lahore, Pakistan; Corresponding author.Department of Civil Engineering Science, School of Civil Engineering, and the Built Environment, Faculty of Engineering and the Built Environment, University of Johannesburg, Kingsway Campus, PO Box 524, Aukland Park, Johannesburg 2006, South Africa; Kunststoff-Technik Adams, Specialist Company According to Water Law, Schulstraße 7, 26931, Germany; Nexus by Sweden, Skepparbacken 5, Västerås 722 11, SwedenWater Science and Environmental Research Centre, College of Chemistry and Environmental Engineering, Shenzhen University, Shenzhen 518060, ChinaDrought is a complex recurrent natural phenomenon. It is the main outcome of climate change. It has long-term impacts on agriculture, human life as well as the environment. Therefore, quantifying drought at the regional level is essential for developing sustainable policies. This study introduced a new drought index for regional drought forecasting called the Regional Forecastable Multiscalar Standardized Drought Index (RFMSDI). The RFMSDI methodology is based on Forecastable Component Analysis (FCA) and K-Component Gaussian Mixture Distribution (K-CGMD). FCA reduce dimension by focus on components that are inherently more predictable. It ensures that reduced data has a built-in ability to predict future trends by selecting the maximized forecastable components. K-CGMD is utilized to model the multimodel time series data. The study application incorporates eight meteorological stations in Türkiye's Elazig province (Baskil, Agin, Elazig, Karakocan, Keban Maden, Palu and Sivrice). The effectiveness of RFMSDI is evaluated by analyzing precipitation data over these meteorological stations of Türkiye. The comparative assessment of the research signifies the superiority of FCA for regional data aggregation. In this research, the comparative assessment of RFMSDI is evaluated against the Standardized Precipitation Index (SPI) by analyzing Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) metrics across different time scales using various machine learning and traditional time series models. The research findings include the following: 1) K-CGMD is a better fitting approach for standardizing RFMSDI and SPI based on reduced BIC values. 2) RFMSDI has superior performance over SPI based on the lower values of RMSE and MAE. 3) Both machine learning and classical methods reveal that RFMSDI outperforms SPI in predicting droughts. 4) SPI shows localized advantages with the ELM training set at 1- and 6-month time scales but RFMSDI offers a more comprehensive and consistent tool for drought prediction, especially when tested on unseen data. In general, the findings endorse the effectiveness of RFMSDI for monitoring drought on a regional level.http://www.sciencedirect.com/science/article/pii/S0378377425000034DroughtForecastable component analysisMachine learning methodsClassical methodsTürkiye |
| spellingShingle | Aamina Batool Veysi Kartal Zulfiqar Ali Miklas Scholz Farman Ali A novel regional forecastable multiscalar standardized drought index (RFMSDI) for regional drought monitoring and assessment Agricultural Water Management Drought Forecastable component analysis Machine learning methods Classical methods Türkiye |
| title | A novel regional forecastable multiscalar standardized drought index (RFMSDI) for regional drought monitoring and assessment |
| title_full | A novel regional forecastable multiscalar standardized drought index (RFMSDI) for regional drought monitoring and assessment |
| title_fullStr | A novel regional forecastable multiscalar standardized drought index (RFMSDI) for regional drought monitoring and assessment |
| title_full_unstemmed | A novel regional forecastable multiscalar standardized drought index (RFMSDI) for regional drought monitoring and assessment |
| title_short | A novel regional forecastable multiscalar standardized drought index (RFMSDI) for regional drought monitoring and assessment |
| title_sort | novel regional forecastable multiscalar standardized drought index rfmsdi for regional drought monitoring and assessment |
| topic | Drought Forecastable component analysis Machine learning methods Classical methods Türkiye |
| url | http://www.sciencedirect.com/science/article/pii/S0378377425000034 |
| work_keys_str_mv | AT aaminabatool anovelregionalforecastablemultiscalarstandardizeddroughtindexrfmsdiforregionaldroughtmonitoringandassessment AT veysikartal anovelregionalforecastablemultiscalarstandardizeddroughtindexrfmsdiforregionaldroughtmonitoringandassessment AT zulfiqarali anovelregionalforecastablemultiscalarstandardizeddroughtindexrfmsdiforregionaldroughtmonitoringandassessment AT miklasscholz anovelregionalforecastablemultiscalarstandardizeddroughtindexrfmsdiforregionaldroughtmonitoringandassessment AT farmanali anovelregionalforecastablemultiscalarstandardizeddroughtindexrfmsdiforregionaldroughtmonitoringandassessment AT aaminabatool novelregionalforecastablemultiscalarstandardizeddroughtindexrfmsdiforregionaldroughtmonitoringandassessment AT veysikartal novelregionalforecastablemultiscalarstandardizeddroughtindexrfmsdiforregionaldroughtmonitoringandassessment AT zulfiqarali novelregionalforecastablemultiscalarstandardizeddroughtindexrfmsdiforregionaldroughtmonitoringandassessment AT miklasscholz novelregionalforecastablemultiscalarstandardizeddroughtindexrfmsdiforregionaldroughtmonitoringandassessment AT farmanali novelregionalforecastablemultiscalarstandardizeddroughtindexrfmsdiforregionaldroughtmonitoringandassessment |