Curvature-aware urban terrain earthwork volume estimation: a multi-city remote sensing framework

Accurate estimation of earthwork material volume is essential for achieving circular and resource-efficient urban development, enabling material reuse in construction planning. Traditional survey techniques cannot efficiently cover large areas, while grid-based approaches often overlook terrain curvature. This study presents a novel and scalable framework for earthwork volume estimation across five cities over eleven years. We introduce a curvature-aware methodology using sub-meter digital terrain models (DTMs) derived from Maxar stereo imagery. DTMs were validated against commercial datasets (RMSE 1.12–2.10 m) and literature benchmarks. Our approach combines adaptive grid refinement with spline-based volume integration, yielding 20.6% lower error than grid-based methods. Pre-earthwork terrain surfaces were reconstructed using Natural Neighbor Interpolation (RMSE 0.06 m in Astana). Temporal analysis reveals diverse urban development trajectories, including volume decline in Astana, stable growth in Almaty and Porto, and seasonal drops in Prague. These findings support sustainable construction strategies through precise city-scale earthwork volume estimation.