Towards High Resolution Urban Heat Analysis: Incorporating Thermal Drones to Enhance Satellite Based Urban Heatmaps

As remote-sensing becomes more actively utilized in the environmental sciences, our research continues the efforts in adapting smart cities by using civilian UAVs and drones for land surface temperature (LST) analysis. Given the increased spatial resolution that this technology provides as compared to standard satellite measurements, we sought to further study the urban heat island (UHI) effect - specifically when it comes to heterogeneous and dynamic landscapes such as the Charleston peninsula. Furthermore, we sought to develop a method to enhance the spatial resolution of publicly available LST temperature data (such as those measured from the Landsat satellites) by building a machine learning model utilizing remote-sensed data from drones. While we found a high correlation and an accurate degree of prediction for areas of open water and vegetation (respectively), our model struggled when it came to areas containing highly impervious surfaces. We believe, however, that these findings further illustrate the discrepancy between high and medium spatial resolutions, and demonstrate how urban environments specifically are prone to inaccurate LST measurements and are uniquely in need of an industry pursuit of higher spatial resolution for hyperlocal environmental sciences and urban analysis.