CASSIE CoRe ®

EUS-Coast (prototype being tested together with C-Squeeze module)

(Projection of Urban Evolution in the Coastal area)

The calculation of expansion, projection of the urban area, can be carried out using two different methodological approaches, described below.

1. Using the FUTURES tool: The FUTURES tool (Meentemeyer et al., 2013, <https://doi.org/10.1080/00045608.2012.707591>) consists of a geospatial model for predicting urban expansion, which requires a detailed set of information of entry, and which, for this reason, limits its applicability to medium and large urban areas (over 200 thousand inhabitants). This will be applied mainly to the Babitonga Bay case study and will be used as validation data for the machine learning algorithm that will be built and implemented in the [EUS-Coast] module.
2. Using machine learning method: In order to develop a methodology that is based on a smaller number of input variables, an algorithm will be developed using machine learning techniques (e.g.: convolutional neural networks – deep learning). With a neural network based on the approach proposed by Guan et al. (2005,<https://doi.org/10.1559/152304005775194746>), which uses historical data from the region (land cover, digital elevation models, road network, etc.) added to satellite images, it is possible to train with based on past behavior, a network capable of predicting future urban occupation for a region, thus enabling the prediction of urban expansion in urban areas of all sizes.

The FUTURES approach will be applied to a small number of large coastal cities, and will serve as a reference for validating the Neural Networks methodology, which will be used to expand the prediction of urban expansion to any urbanized region in the coastal zone of Brazil. As a product of the application of the method, a case study will be generated, with projections of urban expansion in cities on the Brazilian coast (for the time intervals mentioned above) in vector format (polygons demarcating the urban areas for each scenario).