An Integrated Framework for Assessing the Adaptive Capacity of Cities to Implement LIDs Using LULC Mapping

Urban flooding is a significant challenge in most cities worldwide, largely due to intensified rainfall extremes and unplanned urban development. Dense settlements have resulted in limited scope for augmenting or retrofitting existing stormwater management structures. Low Impact Development (LID) can restore the hydrology of urban areas by using decentralised stormwater control and transforming it into its natural state. It's critical to understand the extent to which a city can employ a LID, often known as its flood adaptive capacity.

While several studies have attempted to quantify urban adaptive capacity for LID implementation, much of this work focuses on broad vulnerability or resilience indicators rather than the realistic placement of LIDs. This creates a gap between high-level assessments and the practical realities of placing LID measures in specific urban contexts.  In practice, LID effectiveness is highly condition-dependent and requires high-resolution spatial information like land use, available space, slope, and surface characteristics. However, most cities lack the high-resolution spatial information and systematic assessment frameworks needed to determine where different LID measures can be realistically implemented. To address this limitation, we have developed a framework that derives those data and assesses the adaptive capacity of a city for implementing LIDs.

As part of this approach, the framework requires high-resolution urban land use/land cover (LULC) data, which we have generated through a multi-stage mapping framework that integrates SegFormer (Vision Transformer) based semantic segmentation with OpenStreetMap (OSM) geometric refinement. The model combines the Sentinel-1 SAR backscatter with Sentinel-2 optical composites to create a multi feature stack which is then used by SegFormer-B0/B1 model. From this refined LULC, key hydrologic indicators were derived, including percent imperviousness, runoff coefficients and available rooftop/pervious area for LIDs.

Further to derive the adaptive capacity an Analytical Hierarchy Process (AHP) in combination with entropy and fuzzy method of based weighted overlay is applied to compute suitability maps for major LIDs using historical flood extent, slope, impervious surface ratio, soil infiltration characteristics (HSG), groundwater depth, derived LULC map, road width and traffic intensity. These suitability maps were converted into an adaptability metric by estimating the fraction of locations that remain feasible for implementation after applying LID-specific constraints.

The result highlights that in highly urbanized cities like Delhi where there’re is no place for Nature Based LIDs, implementing decentralized Rain barrels with only 37mm capacity per household can reduce the flood by 75%. Overall, the proposed framework provides a pathway from LULC mapping to city scale LID adaptability assessment and enables evidence-based decision making for sustainable decentralised stormwater management.