Simulation of the climatic changes around the coastal land reclamation areas using artificial neural networks

Abstract

For the last 20 years, Istanbul has intensely experienced land use/cover change (LUCC) as a result of both the direction of investment strategies to the construction sector and urban transformation works. Within this process, the demand for the disposal of debris from building demolitions with minimum cost has made the creation of land reclamation areas a current issue. Land reclamation areas that pose a threat to the marine ecosystem also have effects on the local climate, depending on the LUCC experienced on the urban surface. In this study, two coastal reclamation areas of Istanbul (Yenikapı, Maltepe) were addressed, and the predictability of changes in the thermal environment after the landfill was examined using Artificial Neural Networks (ANN). When the relationship between the simulation data and the actual changes was statistically tested, correlations between the original and the simulated images of Maltepe and Yenikapı are 0.650 and 0.710 respectively were obtained. Also, it was determined that the simulations provided exact results in the range of 37–55%, and accurate results in the range of 66–87% with a sensitivity of 100 m. These results revealed that the simulations performed by the ANN have sufficient sensitivity for monitoring the thermal changes in urban areas.