Abstract
<title>Abstract</title> <p>Urban flood assessment identifies where hazards, exposure, or vulnerability are greatest, but it less often asks whether realized flood burden is unusually high or low relative to the environmental conditions under which it occurs. This study develops Hazard-Normalized Flood Burden Deviation (HNFBDev), an observed–expected geospatial diagnostic that identifies where urban flood burden is higher or lower than expected under represented hydro-environmental conditions. Using Jakarta as a case study, the framework combines hydro-environmental forcing, Sentinel-1 flood observations, population exposure, environmental expectation modelling, and spatial diagnostics. The hydro-environmental baseline represents inland and coastal conditions, while the population-exposure-weighted flood-burden proxy captures the intersection of flood observations and exposed population. Environmentally expected burden is estimated from the forcing baseline, and HNFBDev is derived as the standardized deviation between observed and expected burden. The results show that compound hydro-environmental forcing was significantly associated with observed burden and explained only part of its spatial variation. The HNFBDev surface identified locations where observed burden was higher or lower than expected under comparable conditions, with significant spatial clustering and robustness under alternative weighting assumptions. These findings demonstrate that flood-burden disparities cannot be interpreted from hazard, exposure, or observed inundation alone. HNFBDev provides a transferable observed–expected geospatial diagnostic, subject to local calibration and validation, for identifying places where flood outcomes warrant hydraulic, infrastructure, institutional, and local verification. It should be interpreted as a screening and prioritization tool, not as a direct measure of adaptation performance, resilience, governance quality, infrastructure effectiveness, social vulnerability, flood depth, or causal mechanisms.</p>