Assessment and Enhancement of National Water Model Height above Nearest Drainage Flood Inundation Mapping using Planet CubeSat for the 2017 Bear River Flood event

Flood inundation remains stubbornly challenging to map, model, and forecast with high precision for decision making because it requires a detailed
representation of the hydrologic and hydraulic processes, which are computationally demanding, and data limited. Recently, an empirical approach,
Continental-Scale Flood Inundation Mapping (CFIM), having fewer data demands and perhaps offering a more practical alternative, has been
presented as a scientific workflow where a Height Above Nearest Drainage (HAND) terrain model along with the National Water Model (NWM)
forecast discharge is employed for near real-time flood inundation mapping. In February 2017, a record flood occurred on the Bear River in Box
Elder County due to rapid snowmelt and rain on snow. In this study, we evaluated the CFIM method over the reach of the Bear River where this
flooding occurred. We evaluated the performance of the CFIM in terms of its accuracy in representing flooded and non-flooded areas when
comparing the results with flood inundation observed in imagery from the high-resolution Planet CubeSat RapidEye Satellites. The results indicate
that there were differences between CFIM flood inundation predictions and flooded area recorded by CubeSat Imagery. We used evaluation of these
differences to address challenges of CFIM and present a set of improvements to overcome some of the limitations and advance the outcome of
CFIM. The improvements utilize (1) the high-resolution (1:24,000) National Hydrography Dataset (NHD) to provide an obstacle-removed and
hydrologically conditioned topography, and (2) a higher-resolution Digital Elevation Model (DEM) dataset available for this area. The results indicate
that differences between CFIM flood inundation predictions and flooded area recorded by CubeSat Imagery were attributed to differences in observed
and forecast discharges, but also notably due to shortcomings in the HAND method and the derivation of HAND from the national elevation dataset
as implemented in CFIM. Examination of the causes for these differences has led us to develop proposed improvements to the CFIM methods,
which in this study were evaluated only for this single location. Nonetheless, the proposed improvements have the potential, following further
evaluation, to improve the broad application of the CFIM methodology.

PLAIN LANGUAGE SUMMARY:
Flood inundation is difficult to map, model, and forecast because of the data needed and computational demand. Recently an approach based on
the Height Above Nearest Drain (HAND) derived from a digital elevation model along with using the National Water Model forecasts has been
suggested, for both flood mapping and obtaining reach hydraulic properties. This approach was tested for a recent snowmelt flood on the Bear River
and compared to inundated area mapped using CubeSat satellite imagery. Initial differences found were reduced by addressing shortcomings in the
terrain analysis evaluation of HAND both in terms of the digital elevation model resolution and method used to condition the digital elevation model
using streamline information.

Slides for AGU Fall Meeting 2018 presentation H34G-08 at Washington D.C., December 12, 2018
Session: H34G: Research, Development, and Evaluation of the National Water Model and Facilitation of Community Involvement II