Maps, rates, and fuzzy mountains: Generating meaningful risk maps

Creating meaningful maps that represent rates and risks in the population is a challenge. Risk rates are often computed for small area units such as census entities that may contain small population counts. Due to the unstable nature of such estimates, maps produced using such data are likely to misrepresent the risk of an event´s occurrence over geographic space. This paper introduces two systems based on distinct approaches to generate risk maps that are not biased by the underlying population distribution of a given region: the adaptive kernel density estimation procedure implemented in WebDMAP and the population-uniform partitioning method included in UPAS. Comparison of both systems shows that qualitatively similar results can be obtained by both approaches.