Document Type : Research Paper
Authors
1 MSc Student, Department of Civil Engineering, University of Birjand, Birjand, Iran.
2 Professor, Department of Civil Engineering, University of Birjand, Birjand, Iran.
3 Assistant Professor, Department of Water Engineering, Lorestan University, Khorramabad, Iran.
Abstract
In this research, modeling and estimation of dew point temperature values in eight meteorological stations located in the eastern regions of Iran were done. These stations, including Bam, Birjand, Iranshahr, Kerman, Mashhad, Tabas, Zabol and Zahan, are all characterized by a dry climate. First, the correlation of different weather parameters with dew point temperature was investigated and then the parameters of mean temperature, maximum temperature and minimum temperature were selected as the parameters with the highest correlation to dew point temperature. These selected parameters then incorporated into a VAR (Vector Autoregression) model as inputs for estimating dew point temperature values. This modeling approach allows us to capture the interdependencies between these variables and enhance our accuracy in predicting dew point temperature. Then the stability of the residual series of the VAR model was investigated and the residual series of this model was developed using the generalized ARCH model. The result of the development of the VAR model was the investigation of the dew point temperature in eight meteorological stations with the VAR-GARCH model. The results indicated that this combined model outperformed VAR model in both the train and test phases. Specifically, the VAR-GARCH model demonstrated higher accuracy and improved results compared to solely using a VAR model. The incorporation of GARCH allowed better modeling of the residual series, leading to an overall increase in accuracy ranging from 5% to 30% during the test phase. These findings suggest that considering both autoregressive dynamics and conditional heteroskedasticity is crucial for accurately predicting dew point temperatures. By incorporating GARCH into our modeling approach, we were able to capture additional information about volatility and further enhance our predictions.
Keywords
Main Subjects
)