The HAND of flood mapping: multi-dimensional evaluation across data-rich and data-poor basins using existing maps, ground observations, and remote sensing data

Abstract Computational bottlenecks continue to limit the scalability of hydrodynamic models, prompting broader adoption of rapid, low-complexity flood mapping methods such as height above nearest drainage (HAND). Despite HAND’s growing popularity, existing evaluations are often fragmented, geographically concentrated to one region, or narrow in scope. We conducted a comprehensive assessment of HAND’s flood mapping accuracy focusing on three Gulf Coast Basins in the United States, the Meghna River Basin in northeast India-Bangladesh, and the Limpopo River Basin spanning multiple countries in Southern Africa. Our assessment systematically targeted four dimensions: (1) geographically diverse domains including data-rich watersheds and data-poor transboundary regions, (2) spatial extents ranging from 1300 km 2 to 39 000 km 2 including semi-arid headwater rivers, mixed land use urban watersheds, and humid wetland dominated regions, (3) event types including historic flood events and 100 year design floods, and above all (4) heterogeneous reference data sources including existing flood hazard maps, ground observations of high water marks, and remote sensing data. Results show that HAND delivers flood maps with reasonably high accuracy across all these dimensions covering broad range of geophysical settings. While reduced accuracy in low-relief terrain is consistent with prior work, our findings underscore the often-overlooked influence of stream network density—revealing it to be as critical as topographic data resolution and river channel roughness in mediating HAND’s performance. Rather than positioning HAND as a replacement for hydrodynamic models, we highlight its value as a rapid decision-making tool within flood modeling workflows, especially in near real-time flood emergency situations when traditional hydrodynamic models often struggle.