Assessment of how uncertainty representation in flood maps can affect geographic-based decisions

Abstract

Listen

Translate article

Flood maps that show predicted flood extents will always be uncertain regardless of how the modelling is conducted. It is therefore important that these uncertainties are represented and communicated in the maps, and that map users understand the presented information. Through an online user survey, this study evaluates how users make geographic decisions based on nine flood uncertainty maps, represented and designed according to data scheme and semantics associated with their values (dual-ended, sequential and binary), and applied with different mapping techniques (continuous surface, choropleth and graduated symbol mapping). The results show that the type of map and the visual variable used for representation (in terms of colours and values) became important when deciding locations. Higher decision confidence was shown when dual-ended and sequential probability maps were used. Medium-to-dark blue regions in these maps made participants avoid locations, while white, brown and the lightest blue colours made them select locations. The usage of a sequential map represented by grey scale colour showed to be less intuitive for the participants, leading to lower task performance and less confidence in decisions. Despite the different backgrounds of participants, comprehension of the uncertainty maps and the tasks did not vary much from each other. Differences among them were observed in location preferences and time to solve the tasks. The user group that had the most professional experience with maps and GIS was most conservative in their site choices, and took longest time to solve the tasks. Students, on the other hand, opted to take more risk in their decisions and preferred more uncertain locations. Apparently, the effectiveness of the flood uncertainty maps used in this study varied mainly on the representations used. Appropriate design made them comprehendible by different users. However, making decisions based on these maps, as well as confidence in decisions and time to solve the task, may also be dependent on other factors such as domain knowledge, line of work, practical experience in handing problems or making decisions, and possibly culture.