Building a historical flooding map through spatial analysis

Due to the increase of flooding cases around the world, there is a need for producing even more accurate flood susceptibility mapping. For this, different models, software and frameworks have been developed for many years to assist local authorities and policy-makers for forecasting hazards and mitigating flooding impacts. However, spatially model flooding in real-world systems remains considered as a difficult task. Forecasting flooding requires knowledge from past events and mapping flood locations is crucial to explain the correlation among the flooding and the influencing factors as well as model calibration and validation. In developing countries, a collection of flooding records and inventories remains challenging, either because the data is not available, or because it is not in the suitable scale and resolution. Building historical flooding map is considered a time-consuming process with multiple datasets and normally with costly field surveys. Besides, acquiring this data can be harder due to the inexistence of flooding insurance or civil protection agencies support. This work aims to contribute to this context by developing and assessing a GIS-based framework to map historical flooding cases through the use of spatial analysis. In this study, we used ArcGIS Pro software to construct a historic flooding map for Campina Grande, Brazil. The city faces recurrent flooding episodes, but there is not an available official map with flooding locations to guide decisions for mitigation. The GIS-based framework allows analysing and better understand the interactions of flooding locations and geographic features. The analysis obtained 230 flooding locations in different scales (buildings, streets and neighbourhoods) and sources for the period from 2004 to 2018. Topographic and hydrologic flood influencing factors (altitude, slope, distance to rivers and lakes, flow direction and accumulation) were selected and combined as layers in the GIS environment. Further, criteria were modelled based on spatial analysis and relations to estimate proximity areas around flood occurrence points with high probability of flooding conditions. These tools allowed to compare visual and data patterns of features and surfaces. All locations and factors were then integrated through Model Builder in order to generate a surface with flooding locations within the city. The final historical flooding map was evaluated and validated with 172 points of confirmed flood cases in the city. The GIS-based framework represents a way of analysing and producing historical inventory maps for flooding management using spatial analysis.