Improve our understanding of urban carbon sequestration by estimating vegetation productivity at a fine spatial scale across the Melbourne metropolitan region. 

Lead Researcher:  Associate Professor Ruwini Edirisinghe

Researcher: Bawantha Rathnayaka, Lahiru Gunathilake, Chayn Sun

Keywords: Vegetation carbon modeling, Remote sensing, Carbon sequestration

Overview

Net Primary Productivity (NPP) is a critical indicator of terrestrial carbon uptake, central to carbon accounting, climate mitigation, and ecosystem management. In urban environments high-resolution NPP data are essential for accurate quantification of carbon sequestration and planning decisions. However, the widely used MODIS NPP product is limited to 500 m resolution, insufficient for heterogeneous urban landscapes, and existing carbon modelling frameworks are not readily applicable to urban contexts. This study develops high resolution (10m) urban vegetation carbon model using MODIS NPP using predictors such as Sentinel-2 data, vegetation indices, including NDVI and EVI, land-cover, vegetation indices and vegetation datasets. The project investigated multiple predictor configurations and machine-learning algorithms to identify an effective modelling approach, with model performance evaluated using appropriate validation metrics. The model was validated regionally and used to generate 10 m carbon maps for the Greater Melbourne area. Vegetation carbon estimates were aggregated to SA1, suburb, and council level administrative units which enable to identify priority areas for urban greening. The interactive dashboard supports local government decision-making by benchmarking vegetation against comparable LGAs, helping identify priorities for environmental planning and investment and enabling more accurate assessment of urban vegetation contributions to carbon sequestration. 

Outcomes, Engagement and Impacts 

• Urban Vegetation Carbon Assessment Model - Melbourne Case Study |  Video

An Urban Vegetation Carbon Assessment Model that leverages artificial intelligence, geospatial analytics, and environmental data to estimate carbon storage and sequestration potential across urban landscapes. Featuring a case study from Melbourne, Australia, this project demonstrates how urban trees and green infrastructure can be assessed at scale to support evidence-based planning, sustainability initiatives, and climate resilience strategies. The model integrates spatial data, vegetation characteristics, and machine learning algorithms to identify areas with the greatest potential for carbon sequestration and environmental benefits.

Contact:

Associate Professor  Ruwini Edirisinghe - ruwini.edirisinghe@rmit.edu.au