Quantifying greenspace using deep learning in Karachi, Pakistan

Abstract

Greenspaces in communities are critical for mitigating effects of climate change and have important impacts on our health. We demonstrate a deep learning-based approach which includes green augmentation for measuring and delineating types of greenspace in a city, with satellite imagery. Our method outperforms gold standard methods which use vegetation indices; it segments 80.68% of the greenspace, and correct assigns 83.63% of image pixels as greenspace or not, while the best performing vegetation index only detects 58.11% of the greenspace of which 56.71% of predictions are correct. Detection across the city can inform planning needs based on where greenspaces exist and in what form; we find that greenspaces in Karachi are often linked to road surface coverage (Pearson’s correlation coefficient (r) shows a significant 0.26 correlation between greenspaces and roads, p < 0.001), with a slightly higher correlation between roads and trees versus roads and grass. Quantifying greenspace in Karachi illuminates an important need; the mean per capita greenspace across union councils in Karachi is 2.84 m2/person which significantly lags World Health Organization recommendations.