|
Development of a hybrid model for generating a precipitation map of Iran
|
Kazem Saber Chenari    , Kazem Saber Chenari * |
| Associate Professor, Department of Hydrology and Water Resources Development, Soil Conservation and Watershed Management Research Institute, Agricultural Research, Education and Extension Organization, Tehran, Iran |
|
|
Abstract: (80 Views) |
| Precipitation is one of the most important components of the hydrological cycle. Therefore, the development of accurate and up-to-date precipitation map at the national scale is of great importance. The aim of this study was to develop an appropriate hybrid approach for generating a new national-scale precipitation map using long-term precipitation data. For this purpose, precipitation data from 1,275 selected rain gauge were collected, quality-controlled, and reconstructed for the period 1988–2017. The long-term annual average precipitation of each station was then calculated and used as input data for interpolation models to generate the annual precipitation map of Iran. After assessing data normality and applying the Box–Cox transformation, the performance of inverse distance weighting and geostatistical methods using different variogram models, was evaluated. Subsequently, several hybrid models were developed based on the best-performing individual models. To simultaneously exploit the capability of different approaches in representing local variations and modeling spatial dependence structures, hybrid models were developed through linear combination of the outputs of the selected interpolation methods and subsequently evaluated. The performance of the investigated methods was assessed using the correlation coefficient, mean absolute error, mean bias error, and normalized root mean square error. The results indicated that among the individual models, ordinary kriging with the spherical variogram provided the best performance, achieving a correlation coefficient of 0.951, an MAE of 57.2 mm, and an NRMSE of 20.4%, and was identified as the most accurate single interpolation method. The IDW method with a power of 2.5 ranked second after the ordinary kriging model. Furthermore, combining the outputs of the best-performing models resulted in several hybrid models that substantially improved the accuracy of precipitation estimation. Among these models, the hybrid ordinary kriging and IDW (power=2.5) model achieved the highest accuracy, with an R=0.999, an MAE of 2.98 mm, and an NRMSE of 1.23%, outperforming all other investigated models. In addition, the spatial distribution analysis of relative errors showed that approximately 98.9% of Iran’s area had an error lower than 10% using the proposed hybrid model, whereas this proportion was only 15.7% for the best individual model. |
Article number: 3 |
| Keywords: precipitation map, Geostatistics, Ordinary Kriging, Co-kriging, Hybrid Model, Spatial Interpolation, Iran. |
|
|
|
Type of Study: Research |
Subject:
Special
Received: 2026/07/10 | Revised: 2026/09/22 | Accepted: 2026/09/16 | ePublished ahead of print: 2026/09/17
|
|
|
|
|
|
|
| Send email to the article author |
|
|
|
| Add your comments about this article |
|
|
|
