Document Type : Research Paper
Authors
1
Ph.D. Student in Climatology, Department of Physical Geography, Faculty of Geography and Environmental Planning, University of Sistan and Baluchestan, Zahedan, Iran
2
Department of Physical Geography, Faculty of Geography and Environmental Planning, University of Sistan and Baluchestan, Zahedan, Iran.
3
Department of Physical Geography, Faculty of Geography and Environmental Planning, University of Sistan and Baluchestan, Zahedan, Iran
4
Postdoctoral Researcher, Department of Physical Geography, Faculty of Geography and Environmental Planning, University of Sistan and Baluchestan, Zahedan, Iran
5
Department of Water Engineering, Faculty of Water and Soil Engineering, Gorgan University of Agricultural Sciences and Natural Resources, Gorgan, Iran
10.22059/jwim.2026.408919.1280
Abstract
Missing values in hydrology studies are a common challenge for hydrologists, especially in statistical analyses that require complete datasets. This research evaluates the performance of the Multiple Imputation by Chained Equations (MICE) method in predicting and reconstructing daily river flow values. The study area is the Kajo River basin in southeastern Iran, and the statistical period covers the hydrological years from 1972-1973 to 2021-2022. To investigate and validate the effectiveness of the MICE approach in managing missing flow data, complete historical daily flow records from the hydrological years 2011–2012 to 2021–2022 were used. Subsequently, the MICE method along with Multiple Linear Regression (MLR) was applied to reconstruct all missing daily flow values. The best-performing estimation methods were evaluated using criteria such as the adjusted coefficient of determination (Adj R²), residual standard error (RSE), and mean absolute percentage error (MAPE). The findings indicated that the Classification and Regression Trees (CART) method combined with MLR outperformed other tested methods, achieving the highest AdjR² value and the lowest RSE and MAPE values. The RSE and MAPE values for the CART-MLR method at the Pirsehrab station are 0.472 and 0.583, respectively, and at the Chandokan station are 0.475 and 0.588, respectively.
Keywords
Main Subjects
)