Spatiotemporal dynamics of urbanization and cropland in the Nile Delta of Egypt using machine learning and satellite big data: implications for sustainable development

The Nile Delta of Egypt is increasingly facing sustainability threats, due to a combination of nature- and human-induced changes in land cover and land use. In this paper, an analysis of big time series data from remotely sensed satellite images and the random forests classifier was undertaken to assess the spatial and temporal dynamics of urbanization and cropland in the Nile Delta between 2007 and 2017. Out of thirteen variables, five spectral indices were chosen to build 500 decision trees, with a resulting overall accuracy average of 91.9 +/- 1.5%. The results revealed that the urban extent in the Nile Delta has increased, between 2007 and 2017, by 592.4 km(2) (1.92%). Particularly, the results indicated that the years 2011 and 2012, which coincided the 2011 political uprising in Egypt, so-called the Arab Spring, were associated with significant land-use changes in the Nile Delta, both in rate and scale. As a result, the cropland area in the region decreased between 2010 and 2011 by 1.63% (502.21 km(2)). Moreover, the results showed that during the period 2012-2017, the mean annual urbanization rate in the region stood at 60 km(2)/year. In contrast, croplands decreased during the same period at an average annual rate of 2 km(2)/year. At the governorates' level, the results suggested that top agricultural producing governorates in the Nile Delta, such as Elmonoufia, Elkalubia, Elbouhyra, and Elghrbia, witnessed the highest rates of decrease in cropland areas during the period 2012-2017. Over the same period, urban areas increased the most in Elkalubia, Domiate, and Elmonoufia by 1.98%, 1.72%, and 1.34%, respectively. The f indings from this analysis are discussed along with their implications for sustainable land-use and urban planning policies.