Application of Random Forest for Identification of an Appropriate Model for Predicting Meteorological Drought
Advances in Meteorology · 2025-01 · 2025 (1)
DOI: 10.1155/adme/7674140 ↗Kurz gefasst
Fachartikel in der Zeitschrift Advances in Meteorology, online erschienen am 10.02.2025. Verfasst von Anwar Hussain und 5 weiteren Autor:innen. Die Arbeit wurde bisher 23-mal zitiert (Crossref) und nennt 67 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: King Khalid University, Prince Sattam bin Abdulaziz University.
Abstract (Originalsprache)
This research aims to find the best model for predicting the Standardized Precipitation Index (SPI) and the Standardized Precipitation and Evapotranspiration Index (SPEI) in the future. The study estimates SPI and SPEI at different time scales, ranging from 1 to 48 months. To predict drought, Random Forest (RF) models are used based on lag times of 1–12 months for the estimated drought indices (SPI and SPEI). Accuracy and error metrics like Nash–Sutcliffe efficiency (NSE), root‐mean‐square error (RMSE), producer accuracy (PA), user accuracy (UA), and Choen’s kappa are used to assess the models. The NSE values for the SPI at varying time scales (1, 3, 6, 9, 12, and 48 months) indicate that Bahawalpur, Rawalpindi, Murree, and Sargodha stations have the highest NSE values of 0.1148, 0.5868, 0.8302, 0.9196, 0.9516, 0.9801, and 0.9845, respectively. Similarly, the RMSE values for SPI at these time scales show that Bahawalpur, Rawalpindi, Murree, and Sargodha stations have the lowest RMSE values of 0.6187, 0.6094, 0.4091, 0.2865, 0.2275, 0.1594, and 0.1106, respectively. The NSE and variance explained for SPI and SPEI at a 1‐month time scale were found to be poor, but they improved as the time scale increased. On the other hand, the RMSE values for SPI and SPEI at a 1‐month time scale were found to be high but decreased with longer time scales. The stations that exhibit the highest values of the NSE for the SPEI at various time scales (1, 3, 6, 9, 12, and 48 months) are Rawalpindi, Jhelum, Murree, Mianwali, Rawalpindi, and Sargodha, respectively. These stations have NSE values of 0.0784, 0.6074, 0.8353, 0.9225, 0.9542, 0.9760, and 0.9896, respectively. Similarly, the stations with the lowest RMSE values for SPEI at these time scales are Sargodha, Murree, Murree, Murree, Murree, and Sargodha, with RMSE values of 1.002, 0.5909, 0.3993, 0.2626, 0.2132, 0.1546, and 0.0941, respectively. The analysis reveals a distinct pattern indicating that stations situated at higher elevations exhibit a more pronounced correlation between the SPI and SPEI indices in comparison to stations at lower elevations. Notably, Murree, Jhelum, Sialkot, and Rawalpindi demonstrate a statistically significant and strong correlation between the SPI and SPEI. Overall, the results show that SPEI is a better drought index for classifying and monitoring meteorological drought in stations with lower elevations. However, in stations with higher elevations, the selected indices provide similar information, but with some differences.
Förderung
- King Khalid University
- Prince Sattam bin Abdulaziz University
Zitieren
APA
Hussain, A., Niaz, R., Al-Rezami, A. Y., Omer, A. M., Al-Duais, F. S., & Almazah, M. M. A. (2025). Application of Random Forest for Identification of an Appropriate Model for Predicting Meteorological Drought. Advances in Meteorology, 2025(1). https://doi.org/10.1155/adme/7674140
Vancouver
Hussain A, Niaz R, Al-Rezami AY, Omer AM, Al-Duais FS, Almazah MMA. Application of Random Forest for Identification of an Appropriate Model for Predicting Meteorological Drought. Advances in Meteorology. 2025;2025(1). doi:10.1155/adme/7674140
BibTeX
@article{Hussain2025,
author = {Hussain, Anwar and Niaz, Rizwan and Al-Rezami, A. Y. and Omer, Adam Mohamed and Al-Duais, Fuad S. and Almazah, Mohammed M. A.},
title = {Application of Random Forest for Identification of an Appropriate Model for Predicting Meteorological Drought},
journal = {Advances in Meteorology},
year = {2025},
volume = {2025},
number = {1},
doi = {10.1155/adme/7674140}
}
EUVIMED