Mon Oct 13, F. Flechtner
• Sorted by Date • Sorted by Last Name of First Author •
Kashani, Ali and Safavi, Hamid R., 2025. Assessing groundwater drought in Iran using GRACE data and machine learning. Scientific Reports, 15(1):14671, doi:10.1038/s41598-025-99342-9.
• from the NASA Astrophysics Data System • by the DOI System •
@ARTICLE{2025NatSR..1514671K,
author = {{Kashani}, Ali and {Safavi}, Hamid R.},
title = "{Assessing groundwater drought in Iran using GRACE data and machine learning}",
journal = {Scientific Reports},
keywords = {GRACE, Groundwater drought, Machine learning, Downscaling, GGDI, XGBoost, CanESM5, Teleconnection, Earth Sciences, Physical Geography and Environmental Geoscience},
year = 2025,
month = apr,
volume = {15},
number = {1},
eid = {14671},
pages = {14671},
abstract = "{Groundwater serves as a critical freshwater reservoir globally,
essential for ecosystem conservation and human well-being.
Drought conditions adversely impact groundwater systems by first
reducing recharge, followed by declines in groundwater levels
and withdrawal potential, which can result in agricultural
setbacks and irreversible consequences such as land subsidence.
The introduction of the Gravity Recovery and Climate Experiment
(GRACE) project marked a significant advancement in monitoring
terrestrial water storage anomalies (TWSA), encompassing both
surface and subsurface water. Traditional methods for assessing
groundwater storage anomalies (GWSA), such as piezometric wells,
have proven to be costly and inefficient, often lacking
sufficient spatial and temporal coverage. Although GRACE data
offers valuable insights, its large-scale nature presents
challenges for localized basin and aquifer studies, compounded
by data gaps resulting from a 15-month interruption during the
transition to the GRACE-FO project. This study investigates the
status of groundwater across six major river basins in Iran
utilizing data from GRACE and its complementary Global Land Data
Assimilation System (GLDAS) over a 255-month period from 2002 to
2023. The Extreme Gradient Boosting (XGBoost) algorithm is
employed for downscaling TWSA to a resolution of
0.25{\textdegree}, achieving a high Pearson correlation (R) of
0.99 and a root mean square error (RMSE) of 22 mm. The
downscaled GWSA, derived from the balance equation, exhibits an
average correlation (R) of 0.93 and RMSE of 39 mm with
observational data. Following the application of the Seasonal
Autoregressive Integrated Moving Average (SARIMA) model to fill
GWSA time series gaps, this study models and forecasts GWSA
trends through 2030 using historical data and SSP2 scenario
projections of the canESM5 climate model. Results indicate an
average groundwater depletion of 29 cm per year across Iran's
aquifers from 2002 to 2023, with the Caspian Sea basin
experiencing the most significant decline. The GRACE Groundwater
Drought Index (GGDI) is calculated and compared with the
Standardized Precipitation Index (SPI), revealing an 8-month lag
in drought propagation from meteorological to groundwater
sources in Iran. Furthermore, correlations between the GGDI and
teleconnection indices highlight their substantial influence on
drought conditions in basins adjacent to major water sources.
The results of this study, by emphasizing the reliability of
satellite data and machine learning models in groundwater
drought monitoring, can assist policymakers in enhancing
groundwater resource management, strategic planning, and
identifying critical basins, particularly in regions with
limited observational data.}",
doi = {10.1038/s41598-025-99342-9},
adsurl = {https://ui.adsabs.harvard.edu/abs/2025NatSR..1514671K},
adsnote = {Provided by the SAO/NASA Astrophysics Data System}
}
Generated by
bib2html_grace.pl
(written by Patrick Riley
modified for this page by Volker Klemann) on
Mon Oct 13, 2025 16:16:53
GRACE-FO
Mon Oct 13, F. Flechtner