Spatial patterns of modelled trees and their association with air temperature in Halle (Saale)
Abstract
Urban Heat Islands (UHI) are a well documented phenomenon in urban areas across the globe. However, the effects of urban trees on air temperature and UHI are rather less studied. This is mostly due to the lack of datasets for urban trees, as manual collection of tree location and type data is costly and time intensive. Deep learning models offer an alternative in this area that can be applied to orthoimages to delineate individual trees over a given area.
The aim of the study is to use a deep-learning-based modelled tree dataset from the city of Halle (Saale) in Germany, crowd-sourced air temperature data, and a land cover dataset to study the spatial distribution of trees in the city and their effects on local air temperature ranges.
In this study, spatial distribution of trees is defined as the density and count of trees per land cover type in 250m x 250m grids across the study area. The aim is to ascertain whether there is a linear correlation between spatial distribution of trees in Halle to daily air temperature (mean and range) in the city.
Generally, there was a negative correlation between tree density and mean daily air temperature (not significant), as well as between tree density and mean daily temperature range (not significant). Whereas, there was a positive correlation between tree count and mean daily air temperature (significant), as well as between tree count and mean daily temperature range (highly significant). The results varied across land cover types.
Paper and Methods
Paper:
Data+Code: https://zenodo.org/records/10792650