Extreme events and alluvial-fan flood hazards: An assessment using hydrological modeling, field mapping, and remote sensing in Arizona

Alluvial fans pose a significant challenge to flood-hazard assessment because of their complex and dynamic topography. The complexity of fan topography renders many standard flood-routing methods inapplicable. As a result, there is currently no widely-used analytical method for predicting flood-inundation extents on alluvial fans. To fill this gap, we propose and test a new flood-hazard methodology based on a 2D hydrological model capable of modeling flows over complex topography. The model solves the continuity equation and Manning's equation simultaneously using an implicit numerical method. A high-resolution DEM and peak flood stage are required inputs. The model can be used to reconstruct inundation extents and flow regimes of historical floods for comparison with field- and satellite-based flood-inundation maps or to predict the inundation extents for a range of peak discharges in order to construct a probability-of-inundation map. This map can then be compared with a surficial-geologic map to determine flood-prone areas. To test the accuracy of the model for reconstructing historical floods, we compared model predictions for inundation extent against field- and satellite-based flood-inundation maps for two extreme historical floods on the southern Tortolita and Harquahala piedmonts in Arizona. Predicted inundation extents match field- and satellite-based maps for the Tortolita and Harquahala floods by 83% and 86%, respectively. Probability-of-inundation maps predict a spatially-complex flood hazard that strongly reflects small-scale topography and surficial geology. These results provide further evidence that alluvial-fan flood hazards are not well resolved and generally overpredicted in FEMA Flood Insurance Rate Maps (FIRMs). The hydrological model is particularly useful for identifying the thresholds in input flood stage required to "activate" flood hazards on different segments of the fan. The model also enables the impact of extreme events on the flood hazard to be assessed. In some areas, uncertainty in the probability of extreme events does not greatly impact flood hazards because incised channels convey all of the flow within a broad range of extreme flood sizes.