Sentinel-1 Estimated Maximum Flood Extent for Sub-District Flood Exposure Mapping in Makassar City, Indonesia

Abstract

Sub-district level information on flood exposure is basic requirement for disaster mitigation and spatial planning, yet it is rarely available for Indonesian cities because satellite imagery seldom captures the inundation peak. The February 2023 flood in Makassar City is a clear case: with only Sentinel-1A operating on 12 day repeat cycle, the peak of 13-17 February was never imaged and nearest acquisition on 21 February recorded receded condition. This study reconstructs the Estimated Maximum Flood Extent (EMFE) for that unobserved peak using surrogate variables and compares two reconstruction pathways: EMFE-GRD, in which backscatter thresholding is corrected to peak conditions using external hydrological data and EMFE-SLC in which interferometric coherence change carries the peak information inherently and receives only a partial correction. The novelty of this study is the explicit use of surrogate variables to reconstruct an unobserved flood peak, combined with paired sample significance testing to select between reconstruction pathways. Both products were validated against 90 paired field points. EMFE-GRD was the more accurate pathway (overall accuracy 62.2%, 95% CI 52.2-72.2%, Kappa 0.25), and the paired difference against EMFE-SLC was statistically significant (McNemar p = 0.028), although both products remain only moderate agreement and EMFE-SLC was not distinguishanle from chance. EMFE-GRD indicates that about 41.6 % of the mainland area is potentially inundated, concentrated in the low-lying eastern and northern expansion zones. The results are presented as a two tier screening instrument, in which a core zone of inter method agreement guides evacuation and resource placement while an anticipatory zone suppots worst case planning, giving BPBD Makassar a field validated basis for mitigation priorities and for integration into the city spatial plan.