:: Volume 10, Issue 3 (2-2023) ::
jgit 2023, 10(3): 121-141 Back to browse issues page
Evaluation of remote sensing-based drought monitoring indexes using support vector regression and random forest models (Case study: Marivan city)
Jamal Seyedi Ghaldareh , Salman Ahmadi * , Mehdi Gholamnia
University of Kurdistan
Abstract:   (1394 Views)
Drought is a natural and climatic phenomenon that occurs in large areas around the world every year, and its occurrence is caused by the shortage of rainfall and increased evaporation and transpiration at high temperatures. The purpose of this research is evaluating the remote sensing data in drought monitoring for Marivan city and analyzing the spatial-temporal distribution of the drought conditions and identifying its severity. In this study, we used different drought indicators produced from MADIS and TRMM satellite data, which were extracted from Google Earth Engine platform to analyze the drought conditions in Marivan city from February to November for the years 2001 to 2017 .In this research, remote sensing indices such as normalized difference index of vegetation, index of vegetation conditions, index of temperature conditions, index of improved vegetation, index of evaporation and transpiration and index of rainfall status were selected as independent variables. Furthermore. the standard rainfall index obtained from meteorological data has been calculated as a dependent variable to evaluate drought conditions. Random forest methods and support vector regression were used to compare the remote sensing data and the ground data and to check the correlation between them and the importance of the remote sensing indicators for drought monitoring. The result of the modeling was obtained using the support vector regression algorithm with the values of the explanatory coefficient of 0.88 and the mean square error of 0.313.The results of the random forest model with the values of the coefficient of explanation of 0.909 and the mean square error of 0.259 indicated the high efficiency of this model. Then, the correlation between  the remote sensing indices and  the meteorological index was investigated. And PCI, ET, EVI, NDVI indices had the most correlation among the other variables.Therefore, the remote sensing indicators can be used to predict the drought situation in the research area.
Keywords: Drought, Remote sensing images, Random forest, Support vector regression
Full-Text [PDF 1506 kb]   (614 Downloads)    
Type of Study: Research | Subject: RS
Received: 2022/09/7 | Accepted: 2023/01/17 | Published: 2023/02/6

XML   Persian Abstract   Print

Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Volume 10, Issue 3 (2-2023) Back to browse issues page