Investigating spatial heterogeneity by Implementing the mgwr python package, a case study: southwestern of Tehran Plain
محورهای موضوعی : فصلنامه علمی پژوهشی سنجش از دور راداری و نوری و سیستم اطلاعات جغرافیاییAli Soleymani 1 , Parham Pahlavani 2
1 - Geographic Information Systems, School of Surveying and Geospatial Engineering, College of Engineering, University of Tehran, Tehran, Iran
2 - Geographic Information Systems, School of Surveying and Geospatial Engineering, College of Engineering, University of Tehran, Tehran, Iran
کلید واژه: Multi Linear Regression, Subsidence, InSAR, Geographically weighted regression,
چکیده مقاله :
Land Subsidence often causes irreversible damage to infrastructures and costs lots of expenses for governments annually; Hence, studying and monitoring subsidence in either plains or urban areas has become necessary in last decades. Studies have introduced excessive depletion of aquitards as the dominant factor in the occurrence of this hazard. In this study, the main aim was to take the impact of other spatial factors involving land subsidence into consideration. To devise a plan whether to pause or reduce the subsidence rate, we need to understand the mechanism of each factor inducing land subsidence. Here, we show the outcomes of a Geographically Weighted Regression (GWR) method with a fixed Gaussian kernel to identify the impact of each of the spatial factors inducing subsidence compared with the results from a Multi Linear Regression (MLR). In this regard, outputs of a compiled Interferometric Synthetic Aperture Radar (InSAR) time series analysis of the 15 Envisat ASAR images consumed to capture displacement from 2003 to 2005. Afterward, a kriging interpolation method is implemented to generate a surface of subsidence. The Python package "mgwr" is used to compile both GWR and MLR models. Several statistical diagnostics are performed to assert the GWR superiority over other non-geographical methods when dealing with spatial data. Finally, the GWR results show that just six factors out of 10 tend to be the dominant factors.
Land Subsidence often causes irreversible damage to infrastructures and costs lots of expenses for governments annually; Hence, studying and monitoring subsidence in either plains or urban areas has become necessary in last decades. Studies have introduced excessive depletion of aquitards as the dominant factor in the occurrence of this hazard. In this study, the main aim was to take the impact of other spatial factors involving land subsidence into consideration. To devise a plan whether to pause or reduce the subsidence rate, we need to understand the mechanism of each factor inducing land subsidence. Here, we show the outcomes of a Geographically Weighted Regression (GWR) method with a fixed Gaussian kernel to identify the impact of each of the spatial factors inducing subsidence compared with the results from a Multi Linear Regression (MLR). In this regard, outputs of a compiled Interferometric Synthetic Aperture Radar (InSAR) time series analysis of the 15 Envisat ASAR images consumed to capture displacement from 2003 to 2005. Afterward, a kriging interpolation method is implemented to generate a surface of subsidence. The Python package "mgwr" is used to compile both GWR and MLR models. Several statistical diagnostics are performed to assert the GWR superiority over other non-geographical methods when dealing with spatial data. Finally, the GWR results show that just six factors out of 10 tend to be the dominant factors
