Integrating Public Participation GIS in Climate Adaptation: A Case Study of Flood Risk Mapping in Berrien County and Milwaukee

Abstract

This study investigates the use of Public Participation Geographic Information Systems (PPGIS) as a tool for validating hydrological flood models. Two web-based PPGIS platforms were developed and deployed in Berrien County, Michigan, and Milwaukee County, Wisconsin. These two study areas are facing increasing flooding risk due to climate changes. The main objectives were to assess whether PPGIS can effectively collect spatial validation points for flood models, evaluate participant bias and expertise variation. A total of 311 responses were recorded for Berrien County and 88 for Milwaukee County. Overall, 56% of Berrien responses aligned with model predictions, with correct points concentrated along the St. Joseph River. In Milwaukee County, 85 of 88 points validated the model, primarily along Lake Michigan’s shoreline. Agreement between participant reports and modeled flood extents were analyzed using Jensen-Shannon Divergence (JSD). Results showed that one user exhibited a low JSD value (0.06), indicating high correspondence with the majority of user inputs. Three users displayed high divergence (>0.78). This finding suggests that one expert with comprehensive knowledge for the flooding may be enough to validate the model. Survey results indicated that 50% of Berrien participants found the tool very easy to use, though users suggested improvements such as better instructions and support for selecting multiple points. Expert feedback during stakeholder sessions highlighted the need for a "damaged" category and alternative input methods, such as area selection. Limitations included small sample size, potential participant bias, and technical issues such as session timeouts. Generally, our tool can clearly collect spatial data and is easy for the user to use. The tool we designed increased the ease of data collection by experts and facilitated communication between hydrologists and other experts to validate the flooding model outputs. The ease of use can be contributed for the future study to calculate data from the general public.