ACTIVE AUTOMATED SENSING SYSTEM FOR URBAN TURBULENT AND RADIANT HEAT

Abstract

Geographic Information Systems serve as automated systems for capturing, storing, retrieving, analyzing, and displaying spatial data, crucial for understanding and mitigating the effects of climate change on urban areas. Urban Heat Island effect, intensified by urbanization and climate change, presents significant challenges to urban sustainability and human health. The Final Year Design Project focuses on the impact of micro-climate change in urban areas and explores the use of GIS and real-time sensing systems to monitor and manage these effects. The approach involved collecting data using QField for tree species, and IoT sensors for temperature and humidity, including four sensor types. Additionally, 45 species of trees and 8000 samples were digitized. Furthermore, 37 drone flights were conducted over NUST, capturing 7000+ images. Combining these datasets, a real-time active automated system was developed and integrated into a React-based web portal. GIS has been utilized as a service, in which, as one of the applications, Machine Learning has been utilized to predict indoor temperature. This has been performed through Extreme Gradient Boosting (XGBoost), by hyper tuning the parameters so that they learn the underlying relationship between indoor temperature, tree species, distance to building, distance to water bodies, tree density, building density, and outdoor temperature. The developed system demonstrates the potential of GIS and real-time sensing technologies in providing valuable insights for sustainable urban planning and climate change adaptation. Additionally, this approach can be further utilized in various use cases, including bio-climatic variables, insect prone areas, heat-zones across study area, and understanding the cooling effects of different species. The findings of the project identified several tree species across NUST that were capable of lowering the temperature, such as red bottle brush, and river tea tree, however, in the contrast, trees as pilkhan and bishop tree were seen to have had little to no effect on mitigating the temperature extremes. The ideal new building site was identified to have a cooling effect if planned at a little distance. It was observed that the tree species gave a certain cooling effect as long as they were not densely planted. Machine Learning Regression Algorithm XGBoost was fine-tuned and gave a resultant R2 Score of 0.8593. Altogether, a service has been created that is capable of collecting from any domain, places it a hosted platform, and queries data from Nodejs APIs and Flash APIs (hosted on AWS). The project has bought together several components and applied them to study the patterns that vary in the spatial domain, harnessing the power of GIS as a service.