An increase of urban flood risks is expected for the following decades not only because climate is becoming more extreme, but also because population and asset densities in cities are increasing. There is a need for models that can explain the damage process of urban flooding and support damage prevention. Recent improvements in flood modeling have highlighted the importance of urban topography to properly describe the built environment. While such modeling has mainly focused on the hazard components of urban pluvial floods, the understanding of damage processes remains poor, mainly due to a lack of flood impact information. Citizen’s reports about flood incidents can be used to describe urban flooding impacts. In this study a database of such type of reports and a digital elevation model are used as main inputs to analyze the relationships between urban topography and occurrence of pluvial flood impacts. After a delineation of urban subwatersheds at a district level, the amount of reports along the overland flow-paths is studied. Then, the spatial distribution of reports is statistically assessed at district and neighborhood levels, in Euclidean and network-constrained spaces. This novel implementation computes the connections of a network of subwatersheds to calculate overland flow-path gradient distances, which are used to test whether the location of reports is constrained by those gradients. Results indicate that while reports have a clear clustered spatial distribution over the study area, they are randomly distributed along overland flow-path gradients, suggesting that factors different from topography influence the occurrence of incidents.