Propagating Uncertainty in Urban Tree Trait Measurements to Estimate Socioeconomic Inequities in Ecosystem Service Accessibility: A Machine Learning and Simulation Framework.

Abstract

Achieving Sustainable Development Goal 11 requires addressing inequities in access to ecosystem services provided by urban trees to ensure a fair distribution of environmental benefits across socioeconomic groups. Ecosystem services estimates based on urban forest inventories often face challenges related to missing data or traits recorded in numerical ranges. Measurement-level uncertainty, if unaccounted for, can lead to overly optimistic estimates and limit their utility for urban planning. This study presents a methodological framework for propagating uncertainty in ecosystem services estimation with i-Tree. The approach integrates machine learning models to impute missing data and employs copulas and Monte Carlo simulations to assess the impact of trait uncertainty on provision. These simulations are further incorporated into Bayesian Hierarchical Models to evaluate how trait uncertainty influences estimates of inequities in ecosystem services accessibility across socioeconomic groups at the census tract level. Results indicate substantial uncertainty in tree-level estimates, which decreases with increasing spatial scale due to the central limit theorem. Model findings reveal inequities in benefits accessibility across social groups, even after accounting for census tract tree density, unobserved covariates and spatial autocorrelation via random effects, and tree trait uncertainty. Higher-income areas with greater income inequality and lower proportions of minority ethnic populations tend to have greater access to ecosystem services. This study provides a replicable methodology applicable to urban tree inventories with similar data limitations and offers urban planners an analytical tool for designing targeted tree-planting strategies that promote equitable ecosystem services provision distribution and foster healthier, more resilient urban communities.