How Much Did It Really Rain During Harvey?

Hurricane Harvey was an extraordinary rain event that killed dozens and caused over $100,000,000,000 in property damage. Flooding involved street flooding, sheet flow, and interbasin flow within both rural and highly urbanized areas, and many bayous set new discharge records. Key to dealing with future disasters of this type is accurately estimating the temporal and spatial distribution of rainfall, both for a proper understanding of the exceptional nature of the event and for proper hydrologic and hydraulic modeling of the actual flood and the impacts of proposed flood mitigation measures.

We collected Stage III and Stage IV radar-based data and rainfall analyses, conventional rain gauge observations, and unconventional observations from volunteer networks and private citizens. We then quality-controlled and integrated these data into high-resolution rainfall analyses using two state-of-the-art extreme rainfall analysis algorithms: MetStorm and the Storm Prediction Analysis System. We also explored the sensitivity of the results to the inclusion and weighting of data from various networks. These experiments allow us to characterize aspects of structural and measurement uncertainty inherent to the rainfall analyses.

We compare the resulting area-integrated rainfall totals at various spatial and temporal scales to those of previous extreme rainfall events occurring within the United States. We discuss the rarity of the event from a probabilistic perspective and use the event to update and assess the validity of a recently-completed analysis of probable maximum precipitation (PMP) for Texas.