Extended Abstract



Introduction:

Surface water reservoirs are among the most critical infrastructures for water supply, particularly in arid and semi-arid regions where water security is increasingly threatened by climate variability, prolonged drought, and intensified human pressure. In such environments, monitoring reservoir storage dynamics is essential for understanding hydrological drought and for supporting sustainable water resources management. The Chah-Nimeh reservoir system, located in the Sistan region of southeastern Iran, is a vital source of drinking and agricultural water. However, this system has experienced substantial storage decline in recent years, raising serious concerns about regional water security and environmental sustainability. Because Chah-Nimeh is strongly dependent on inflow from the transboundary Helmand River, its hydrological regime is influenced not only by local climatic conditions but also by upstream water regulation and cross-border management. Recent developments in remote sensing and cloud-based geospatial analysis have greatly improved the capacity to monitor reservoir surface area and storage variations with high temporal and spatial resolution. Therefore, this study aimed to investigate the temporal dynamics of reservoir storage and hydrological drought in the Chah-Nimeh reservoir during 2017-2025 using Sentinel-2 imagery and multi-source climatic datasets within the Google Earth Engine platform, while also examining the extent to which local climatic indicators can explain observed storage variations.

Materials and Methods:

This study was designed to evaluate the temporal dynamics of water storage and hydrological drought in the Chah-Nimeh reservoir using an integrated remote sensing and multi-source climate data framework. All analyses were conducted in the Google Earth Engine (GEE) cloud-computing platform, which enabled efficient processing of long-term geospatial datasets. Four main data sources were employed based on their open accessibility, scientific reliability, and appropriate spatial and temporal resolution. Sentinel-2 Level-2A imagery was used as the primary optical dataset for water surface extraction. A total of 345 atmospherically corrected images acquired between March 2017 and December 2025 were processed. Cloud and cloud-shadow contamination were removed using the Scene Classification Layer (SCL), and only images with less than 15% cloud cover were retained. Monthly composites were then generated using the median spectral values, yielding 83 valid monthly observations.

To delineate the water body, three spectral water indices, namely NDWI, MNDWI, and AWEI, were calculated for each monthly composite. The use of multiple indices allowed both robust water detection and uncertainty estimation in the derived surface area. Reservoir volume was subsequently estimated from the extracted water surface area using the stage-area-storage relationship of the reservoir. A second-order polynomial regression was fitted between water surface area and storage volume, providing a reliable basis for monthly volume estimation. Water level was also derived through inverse interpolation from the stage-area curve.

To assess climatic conditions, CHIRPS daily precipitation data, ERA5-Land monthly temperature and potential evapotranspiration data, and TerraClimate monthly products were incorporated. These datasets were used to compute climatic and hydrological indicators, including SPI, SPEI, PDSI, SRI, SSI, and SWLI. Among them, SWLI, derived directly from reservoir storage anomalies, was considered the principal indicator of hydrological drought, whereas SPI, SPEI, and PDSI were used only for comparative interpretation. The reliability of CHIRPS and ERA5-Land was evaluated against observations from the Zabol meteorological station using correlation coefficient, coefficient of determination, bias, and normalized root mean square error.

Trend analysis was performed using the non-parametric Mann-Kendall test and Sen’s slope estimator. Pearson correlation, coefficient of determination, normalized root mean square error, and mean absolute error were applied to examine the relationships between climatic variables and reservoir storage. In addition, lag-time analysis was conducted to identify the temporal response of reservoir storage to precipitation variability.



Results and Discussions:

The results revealed a pronounced decline in the storage status of the Chah-Nimeh reservoir during the study period, confirming the occurrence of significant hydrological stress. Analysis of the monthly time series showed that both reservoir volume and water surface area followed a statistically significant downward trend. Reservoir storage decreased at an average rate of 5.38 million m³ per year, while water surface area declined by 0.63 km² per year, indicating a persistent reduction in the effective capacity of the reservoir. These findings demonstrate that the reservoir experienced not merely short-term fluctuations, but a sustained long-term depletion trend.

The temporal pattern of storage dynamics further showed that the reservoir passed through several distinct phases. A relatively stable and wetter phase was observed during 2019-2020, followed by a gradual decline during 2021-2022, and then a severe depletion phase extending through 2023 to early 2024. The lowest storage value was recorded in February 2024, when reservoir volume dropped to approximately 26.27 million m³, representing the most critical hydrological condition during the study period. In contrast, a sharp recovery occurred in May 2024, when storage increased abruptly from 36.56 to 99.05 million m³ within a single month. Given that local precipitation during this period was negligible, this sudden increase cannot be attributed to local rainfall and is more plausibly associated with changes in upstream inflow conditions or transboundary water release.

Hydrological drought analysis provided further evidence of this pattern. Based on the SWLI index, which was adopted as the principal indicator of reservoir hydrological drought, 12 out of 82 analyzed months were classified as drought periods, of which 83% fell into the severe or extreme drought categories. The year 2023 represented the most critical hydrological drought condition, with an average SWLI of -1.35, indicating severe drought. However, a key finding of the study was the clear mismatch between reservoir-based hydrological drought and locally derived climatic drought indicators. During periods when SWLI indicated severe depletion, the meteorological indices SPI and SPEI often remained close to normal. This divergence was especially evident in 2023, when hydrological drought intensified despite the absence of a comparable meteorological drought signal.

Correlation analysis supported this interpretation. None of the examined local climatic variables or drought indices showed sufficient explanatory power for reservoir storage variations. Even the relatively better-performing indicators, such as SPEI-6 and SPEI-12, exhibited limited agreement with observed storage dynamics, as confirmed by low R² values and relatively high NRMSE. These results suggest that the Chah-Nimeh reservoir is not governed primarily by local climatic forcing. Rather, its storage behavior is strongly influenced by upstream inflow from the Helmand River and by transboundary and managerial controls. Therefore, the hydrological drought observed in the reservoir should be interpreted as a process driven

Conclusion:

The findings of this study demonstrated that hydrological drought in the Chah-Nimeh reservoir cannot be adequately interpreted using local meteorological indicators alone. Reservoir storage showed a significant declining trend during 2017-2025, while the SWLI index identified severe hydrological drought conditions, particularly in 2023 and early 2024. In contrast, SPI, SPEI, and PDSI did not show consistent agreement with the observed storage depletion. This mismatch indicates that reservoir behavior is primarily controlled by regulated inflow from the Helmand River and transboundary management factors. The integrated Google Earth Engine-based framework proved effective for continuous monitoring of reservoir dynamics in data-scarce and transboundary dryland systems.

Keywords: Remote Sensing, Hydrological Drought, Chah-Nimeh Reservoir, Spectral Water Indice