Development of a Multi-Scale Observation Product to Aid Numerical Model Simulations During Flood Emergencies
Open Access
- Author:
- Sava, Elena
- Graduate Program:
- Geography
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- November 26, 2018
- Committee Members:
- Guido Cervone, Dissertation Advisor/Co-Advisor
Guido Cervone, Committee Chair/Co-Chair
Alexander Klippel, Committee Member
Anthony C Robinson, Committee Member
Andrea H Tapia, Outside Member - Keywords:
- remote sensing
flood
numerical models
social media
data fusion - Abstract:
- This dissertation introduces a new Multi-Scale Observation Product (MOP) for flood assessment. The MOP integrates traditional observations with new, low cost, opportunistic data to create inundation maps at a high spatial and temporal resolution. A fusion technique is developed that combines satellite remote sensing imagery with aerial photography and user-generated data shared through social network outlets and provides a characterization of uncertainty. The goal is to augment remote sensing imagery, which is characterized by high spatial but low temporal resolution, with near real-time data that are often spatially sparse, but stream at a high temporal velocity throughout a flood emergency. The MOP is tested on three separate cases of flood events that occurred throughout the U.S. Results show that MOP can be generated using a variety of different data sources, and flood inundations estimated are comparable with a simulated output from numerical models, Hydrologic Engineering Center River Analysis System (HEC-RAS), Flood2D-GPU, and WRF-Hydro. Likewise, results showed that inundation and confidence levels derived from the MOP correspond with observation data from the Federal Emergency Management Agency (FEMA) products. Lastly, water level heights extracted from photographs shared through social media showed encouraging results as a strong agreement was seen between model output and field measurements. This research was driven by the need for high spatial and temporal resolution flood products for flood modeling. More specifically, hydrological models are now run at very fine resolutions (hyper-resolution modeling), but there is a lack of observation-based products at these high resolutions. The MOP product could be used to assess inundation extents, calibrate model simulations, initialize conditions, and identification of optimal model parameters.