Economic aspects of urban greenness along a dryland rainfall gradient: A time-series analysis

Abstract

Studies at the global scale show that urban greenness is not equally distributed across and within cities. Yet, quantification of urban greening trends in drylands is still lacking. We have modeled urban greenness dynamics and its determinants, using Landsat-based time-series analyses of NDVI and census data, of 23 localities along a dryland climatic gradient in southern Israel, between 1997 and 2019. NDVI trends and their associated temporal changes in local average wages, age of locality, and average annual rainfall were analyzed by exploiting a panel structure and model parameters estimation with fixed effects to control for unobserved differences between localities, and to strengthen causal interpretation. Results show positive NDVI trends in all localities—indicating that urban construction increases greenness in drylands. However, the localities varied greatly in the slopes of their respective linear NDVI trends (0.05 <α < 0.161). Specifically, we found that, ceteris paribus, the differences in average wages between localities is the most important factor in explaining spatio-temporal differences between the respective NDVI of localities. Given the well-known high correlation between individuals’ wages and other characteristic such as education and income (all important determinants of the Socio-Economic classification of local authorities), this leads us to conclude that residents’ economic characteristics are an important predictor of the level of greenness. Harlin’s Granger causality test for wages and NDVI panel data indicate that wage is a Granger cause of urban greenness (p < 0.001)—however, greenness was not found to be the driving force of wage.