The rapid increasing in population density, rising in unmanaged and haphazard development of informal housing, economic crises, among others have been addressed as the principle source for the rapid urban sprawl in many African cities leading the large area of land being occupied for housing ( Nithesh, 2017) thus impact natural resources and peoples, and that authorities need to be informed for quick response to maintain sustainable management of natural resources (Jean Hug?a, 2018). While majority of researchers putting their effort in studying urban sprawl, either by focusing only on the trend of urban growth and future prediction, or by combining both trend and applying Suitability analysis to propose locations for new cities, still there is a lack in including both growth trend and patterns in proposing these new proposed areas, instead they mainly focusing in the general multi influence factors including roads, soil, settlement, railways and others in their analysis, resulting unfitting results for controlling future urban sprawl.
The optimal proposal for future controls in urban sprawl, should consider both multi influence factors together with trend and patterns reflecting the growth rate at that particular location (Zella, 2017), this will provide a holistic overview on where the controls should be taken in advance, and where there is a room to open more sites.
In this study we aim to combine Remote Sensing, GIS and applying python programing, first to explore the trend and rates at a particular patterns regarding the urban sprawl from 1994 – 2018 in Zanzibar city, and letters use these patterns to predict future trends and proposing the suitable area for developing new cities.
Although the study has been allocated to specific case of Zanzibar, the idea and methods can be applied anywhere facing similar problem.
In recent years the extent and expanding rates in urban cities have been reported to grow in extensive range, leading majority of researchers and other authorities taking explicit attention in Land Cover Land Use (LCLU) management to ensure the appropriate control for the future trend (Zella, 2017). GIS, remote sensing and machine learning has been widely used in variety aspect in this particular phenomena providing their mutual benefits linking researchers patterning conservation and management of ecosystem services (lHaque, 2017). The issue like using satellite images and applying machine learning algorithms including a Random Forests (Patil, 2017) in performing land cover classification and predicting future land change trends, Convolutional neural networks (Dragi?evi?, 2019) in purposes of identifying land use changes even from crowd sources such as geo-tagged images, gradient boosted machine learning algorithm (Aldakheel, 2011) to assess and compare the health of vegetation cover in different seasons, together with emerging GIS techniques like applying Site Suitability analysis to allocate the new suitable areas for city development ( Nithesh, 2017) or Geostatistical approaches including kriging interpolation, spatial autocorrelation and overlapping neighborhood statistics to access the correlation between land cover changes with population density of a particular region, along with estimating the changes that could affect their neighborhood (Tobar, 2012), appear to be the common approaches applied in land cover land use management resulting a massive studies deployed in this particular field.
Implication of these techniques in a number of case studies reveals their significance strength and applicability in multiple LCLU domains, thus influence our interest to apply some of those techniques in our current study