[5pt] Test out larger agreedem buffer on flatter floodplains

[CarsonPruitt-NOAA]

One of the underlying issues that contributes to the catchment boundary glass wall occurs when a floodplain is very flat. Oftentimes HAND will attribute floodplain pixels to stream pixels several km downstream because of small variations in the DEM or natural levees near the streambank. In the illustration below, the flow direction at green X is likely being routed along the green arrow, causing its relative elevation to be too high. If it were routed to the stream along the more direct white arrow, the relative elevation would be much more realistic. (HUC 11070203)

This task is just to test the hypothesis to see if the agree buffer distance has any effect on mitigating the catchment boundary glass wall. Eventually, we may want to adjust the buffer by stream order or floodplain width.

[CarsonPruitt-NOAA]

inundation-mapping/config/params_template.env

Line 5 in dc67758

export agree_DEM_buffer=70

[CarsonPruitt-NOAA]

Another smaller example in HUC 06010108

[RyanSpies-NOAA]

Here is the experimental depth output I generated for 11070203 (near Strong City and Cottonwood Falls) using the NWM 25yr recurrence flow datafile:

This image displays the GMS branch "volume m3" for each inundated catchment polygon. Note there are significant fluctuations in the volume extracted from the SRCs between these "hot dog" catchments: