Flood damage model bias caused by aggregation

Reducing flood risk through improved disaster planning and risk management requires accurate and reliable estimates of flood damages.  Damage models commonly provide such information through calculating the impacts or costs of flooding to exposed assets, such as buildings within a community. At large scales, computational constraints or data coarseness leads to the common practice of aggregating asset data using a single statistic (e.g., the mean) prior to applying non-linear damage models. While this simplification has been shown to bias model results in other fields, like ecology, the influence of object aggregation on flood damage models has so far not been investigated. This study quantifies such errors in 12 published damage function sets and three levels of aggregation using simulated water depths. Preliminary findings show bias as high as 20% (of the damage estimate), with most damage functions having a positive bias for shallower depths (< 1 m) and a negative bias for larger depths (> 1 m). In other words, compared to an analogous model with object-specific asset data, aggregated models overestimate damages at shallow depths and underestimate damages at large depths. These findings identify a potentially significant source of error in large-scale flood damage assessments introduced, not by data quality or model transfer, but by modelling approach. With this information, risk modellers can make more informed decisions about when, where, and to what extent aggregation is appropriate.