.png){kind=icon}
.png){kind=icon}
Return
Multi-scale drought analysis and machine learning-based completion of missing streamflow data in the Aras Basin
.png){kind=icon}
PRE
AI
C
B
DOI：10.1016/j.pce.2026.104410.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Drought is occurring as an inevitable consequence of climate change. Long-term changes in precipitation patterns impact water resources, resulting in ecological and socioeconomic challenges. Monitoring the long-term changes in precipitation, temperature, and streamflow data plays a significant role in the early detection of drought. However, long-term, consistent, and complete data records are important for the effectiveness and reliability of developed drought models. In this study, meteorological and hydrological drought analyses were performed using precipitation, temperature, evaporation, and streamflow data from 1980 to 2023 in the provinces of Ardahan, Kars, and Erzurum, where a continental climate prevails over the Aras basin. Streamflow data not recorded after 2011 was predicted for the provinces of Kars and Erzurum using machine learning methods (RF, KNN, SVR, and XGBoost). The XGBoost model, exhibiting the highest performance metrics, was selected to impute the gaps in the streamflow time series. The SPI, SPEI, and RDI methods were employed in the study to monitor the temporal evolution of meteorological drought, while the SDI method was utilized to track the changes in hydrological drought within rivers. Additionally, the Mann-Kendall and Sen's Slope techniques were utilized to identify the trend in temperature, precipitation, and streamflow data Results indicate that accurately completing missing streamflow data increases analysis reliability, contributing significantly to early hydrological drought detection in continental climates. In this respect, the study provides a scientific basis for future water resource management and climate change adaptation strategies.
Keywords:
Meteorological drought
Hydrological drought
Missing data
Machine learning
Reconnaissance drought index
Standardized precipitation index
Aras basin
Journal
IF:
4.1
Papers:
3.3K
Citations:
6.9K
Organization
Cited Papers