Geospatial multi-indicator assessment of flood risk in a wetland ecosystem, using a PCA-weighted risk framework

Flood risk in wetland ecosystems is increasing due to hydro-climatic variability, land-use changes, and socio-economic vulnerability, particularly in low-lying regions of South Asia. This study presents a geospatial multi-indicator framework to assess flood risk in the Chalan Beel, one of the largest freshwater wetland ecosystems in Bangladesh. The assessment integrates four major risk components (hazard, exposure, vulnerability, and capacity) using the analytical relationship. A total of 23 spatial indicators derived from satellite data, digital elevation models, socio-economic datasets, and open geospatial sources were incorporated. All variables were normalized and resampled to ensure spatial consistency. Indicator weights were determined using Principal Component Analysis (PCA), allowing integration of multi-dimensional variables. Results indicate that 22.85% of the study area falls in the moderate hazard class, while 20.74% and 16.93% fall in the high and very high hazard classes, respectively, under quantile classification. Each exposure class covers approximately 19 to 21% of the area, which is a consequence of the equal-count property of quantile classification and not evidence of a balanced spatial distribution. Elevation (23% weight) and unemployment rate (18%) received the largest exposure weights. Children and elderly population densities received the highest PCA weights (27% each) and therefore contributed most to the spatial variance of the vulnerability component. These weights describe statistical contribution to spatial differentiation and not causal importance. Capacity indicators reveal limited adaptive infrastructure in several locations, and distance to flood shelter (39%) received the largest capacity weight. The integrated assessment places 23.47% of the study area in the relatively high flood risk class and 21.01% in the relatively moderate class. These are relative rankings within the study area under quantile classification and not absolute risk levels, and they indicate spatial heterogeneity in relative risk. Comparison with observed inundation maps of the Bangladesh Water Development Board for 2022 to 2024 gave an area under the ROC curve of 0.84, and the index was compared with alternative weighting and aggregation schemes. Areas characterized by low elevation, proximity to rivers and beels, and higher population sensitivity fall in the relatively highest risk classes. The findings provide a spatially explicit framework to support wetland-focused disaster risk reduction, climate adaptation planning, and sustainable management of wetland ecosystems in Bangladesh.