Assessment and Projection of Meteorological Drought in Pakistan Using SPEI and Bayesian Neural Networks under CMIP6 Climate Change Scenarios
International Journal of Innovations in Science and Technology · 26.09.2026 · 3014
DOI: 10.33411/ijist/2044 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Innovations in Science and Technology, online erschienen am 26.09.2026. Verfasst von Aftab Ali Khan und 3 weiteren Autor:innen. Die Arbeit nennt 34 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Pakistan’s water and food security depend on a highly variable and increasingly stressed hydroclimatic system, yet most existing drought assessments in the country are deterministic and provide little indication of the confidence that should be placed in their projections. This study developed a probabilistic, Bayesian Neural Network (BNN)-based framework to assess and project meteorological drought across Pakistan using the Standardized Precipitation Evapotranspiration Index (SPEI) at 1-, 3-, 6-, and 12-month timescales. Monthly precipitation and temperature records from 62 stations (1981–2018) were used to compute historical SPEI, and K-means clustering identified five spatially coherent, timescale-invariant homogeneous drought zones spanning the humid northern highlands to the arid southern plains. A BNN using Monte Carlo Dropout was trained for each zone and timescale, achieving strong predictive accuracy (test R² = 0.839-0.993; NSE = 0.69-0.99), with predictive uncertainty widening appropriately during rare, severe drought events. Bias-corrected precipitation and temperature from three CMIP6 global climate models under SSP2-4.5 and SSP5-8.5 were used to project SPEI through 2099. Results show that drought risk intensifies disproportionately at already arid or topographically distinct stations (Pattan, Bannu, Ormara) under the high-emission scenario, while wetter northern stations remain comparatively insensitive to emission pathway; conversely, the extreme northern station Gupis shows a projected long-term (SPEI-12) wetting trend, a finding that should be treated as a hypothesis generated by the present three-GCM ensemble rather than an established regional climate signal, pending confirmation from a larger multi-model ensemble, in contrast to persistent or worsening deficits further south. These findings indicate that Pakistan’s future drought risk will become increasingly spatially concentrated rather than uniformly distributed, and to our knowledge, they provide among the first uncertainty-quantified, station-resolved drought projections for the country, offering an evidence base for targeted, risk-informed water-resource and agricultural adaptation planning.
Zitieren
APA
Khan, A. A., Ahmad, S. F., Saeed, S., & Khan, A. U. (2026). Assessment and Projection of Meteorological Drought in Pakistan Using SPEI and Bayesian Neural Networks under CMIP6 Climate Change Scenarios. International Journal of Innovations in Science and Technology, 3014. https://doi.org/10.33411/ijist/2044
Vancouver
Khan AA, Ahmad SF, Saeed S, Khan AU. Assessment and Projection of Meteorological Drought in Pakistan Using SPEI and Bayesian Neural Networks under CMIP6 Climate Change Scenarios. International Journal of Innovations in Science and Technology. 2026:3014. doi:10.33411/ijist/2044
BibTeX
@article{Khan2026,
author = {Khan, Aftab Ali and Ahmad, Syed Furqan and Saeed, Salman and Khan, Afed Ullah},
title = {Assessment and Projection of Meteorological Drought in Pakistan Using SPEI and Bayesian Neural Networks under CMIP6 Climate Change Scenarios},
journal = {International Journal of Innovations in Science and Technology},
year = {2026},
pages = {3014},
doi = {10.33411/ijist/2044}
}
EUVIMED