Remote Sensing-Based Urban Sprawl Modeling Using Multilayer Perceptron Neural Network Markov Chain in Baghdad, Iraq

Abstract

The global and regional land use/cover changes (LUCCs) are experiencing widespread changes, particularly in Baghdad City, the oldest city of Iraq, where it lacks ecological restoration and environmental management actions at present. To date, multiple land uses are experiencing urban construction-related land expansion, population increase, and socioeconomic development. Comprehensive evaluation and understanding of the effect of urban sprawl and its rapid LUCC are of great importance to managing land surface resources for sustainable development. The present research applied remote sensing data, such as Landsat-5 Thematic Mapper and Landsat-8 Operation Land Imager, on selected images between July and August from 1985 to 2020 with the use of multiple types of software to explore, classify, and analyze the historical and future LUCCs in Baghdad City. Three historical LUCC maps from 1985, 2000, and 2020 were created and analyzed. The result shows that urban construction land expands quickly, and agricultural land and natural vegetation have had a large loss of coverage during the last 35 years. The change analysis derived from previous land use was used as a change direction for future simulation, where natural and anthropogenic factors were selected as the drivers’ variables in the process of multilayer perceptron neural network Markov chain model. The future land use/cover change (FLUCC) modeling results from 2030 to 2050 show that agriculture is the only land use type with a massive decreasing trend from 1985 to 2050 compared with other categories. The entire change in urban sprawl derived from historical and FLUCC in each period shows that urban construction land increases the fastest between 2020 and 2030. The rapid urbanization along with unplanned urban growth and rising population migration from rural to urban is the main driver of all transformation in land use. These findings facilitate sustainable ecological development in Baghdad City and theoretically support environmental decision making.

Geospatial Analysis of Land Use/Cover Change and Land Surface Temperature for Landscape Risk Pattern Change Evaluation of Baghdad City, Iraq, Using CA–Markov and ANN Models

“…Baghdad's population increased at an alarming rate over the last 70 years (between 1950 and 2020), starting with 579,167 people in 1950 and consistently increasing to 3,606,844 people in 1985, with its population exponential growth reaching 7,144,260 by 2020 [40]. There are concerns that future urban development may be threatened by natural disasters; however, the US Geological Survey Earthquake datasheet from 1985 to 2020 (https://earthquake.usgs.gov/earthquakes/search/ [accessed on 8 March 2021]) showed no reported earthquakes and fault lines in the study area, while the adjacent cities and eastern Iraq were at risk of earthquakes [32].…”

“…In this study, the image classification and interpretation used the combined classification method based on ERDAS IMAGINE 2015 and ArcGIS 10.8. The combined method, which included supervised and unsupervised classification methods, is widely used for LUCC classification [15,32]. The maximum likelihood algorithm is the most frequently employed approach for remote sensing image interpretation using supervised classification, and it calculates the subsequent probabilities of fitting pixels to their corresponding classes according to the Bayes Theorem [46].…”

“…The dependent variables, such as digital elevation model (DEM), slope, road network, road distance, distance from the urban, distance from the river, and population density. The other independent variables, such as BL change to CL, WB change to CL, AL change to CL, and change from all LU to CL, are accounted as driving factors for the Land Change Modeler (LCM) in this study [32] (Figure A1a-l).…”

“…In this research, the past LUCC classification from 1985 to 2020 was processed by using ArcGIS, and the CA-Markov model was applied for future LUCC predictions for 2030, 2040, and 2050 using Terrset software [32]. The LCM window is used for data managing in Terrset software [15].…”

“…Qassim et al and Al-Ramahi et al analyzed the LUCC from 1986 to 2019. The spatial concentration of Radon Gas distribution in Baghdad City was assessed by using RS and GIS [32]. Hamdy et al [33] mapped the LUCC in Al-Karkh, Baghdad.…”