Satellite Imagery Analysis of The Gaza Strip Using Remote Sensing and Machine Learning Techniques

Background: Urban Sprawl is a significant issue in both developed and developing countries. Urban expansion has led to multiple consequences, including the loss of productive agricultural lands, open green spaces and reduction of groundwater levels. Objective: The goal of this study is to integrate remote sensing and machine learning technologies to examine the decline of agricultural land in the Gaza Strip caused by urban expansion. The use of satellite imagery provides detailed data to better understand these changes, particularly when a variety of satellite images spanning multiple periods are used. Methods: The study employed an analysis of six different satellite images captured every 10 years from 1970 to 2020. The findings were used to predict land use for 25 centers in Gaza Strip in 2030. Results: The analysis of satellite images by utilizing controlled classification produced a set of data and tables showing the evolution of the urban expansion across the periods of 1970-2020. Using this data, two models were employed to predict the areas of urban and agricultural use up to 2030: the Markov chains model and a regression model using the random forest algorithm. The predictions of the random forest were very close to that of the Markov model, indicating the efficiency of the algorithm. Conclusion: The study revealed a significant and rapid increase in urbanization in city centers, which was estimated to be around 28.9 km2 by 2030. While in more rural areas, there was an increase in annual growth at the expense of agricultural land, particularly in the eastern and western regions of the Gaza Strip.