[21pt] Adjustment of SRCs using optimized manning number

[hhs732]

Initial scatter plot of optimized in-channel and overbank manning roughness:

More analyses coming

[hhs732]

And here is the median for in-channel and over bank roughness values:

n_ch = 0.0711 n_ob = 0.1106

[CarsonPruitt-NOAA]

Very cool, Heidi! I actually love the idea to use the HUC on the x-axis of your plot because it sort of separates the HUCs into their geographic sectors. I never would have thought of that.

[CarsonPruitt-NOAA]

Your next steps should be:

1. Create a new Manning's n file with your new optimized median roughness values
2. Run an Alpha domain of FIM pipeline with the new Manning's n file
3. Compare eval plots of these 4 runs: (You should have already run the first 3 in your previous analysis)
   • v4.5 (calibration on)
   • v4.5 (calibration off)
   • ML-bathymetry (calibration off, default roughness)
   • ML-bathymetry (calibration off, new optimized roughness) <-- this is the new run from step 2 above

[hhs732]

Optimized in-channel roughness map

[hhs732]

Optimized over-bank roughness map

[hhs732]

Pink: Median optimized roughness

[hhs732]

[hhs732]

[hhs732]

Apparently just working for BLE sites, as it was optimized on them. Also, we only considered bathymetry on stream order 7 and higher in optimization process. Right now we are optimizing on bathymetry for SO4 and higher. Let's see what we get in this round for optimized roughness values. I am still wondering if we could use a classification approach to assign one manning number for each HUC.

[RobHanna-NOAA]

Apparently just working for BLE sites, as it was optimized on them. Also, we only considered bathymetry on stream order 7 and higher in optimization process. Right now we are optimizing on bathymetry for SO4 and higher. Let's see what we get in this round for optimized roughness values. I am still wondering if we could use a classification approach to assign one manning number for each HUC.

We could implement that pretty easily and I could see us eventually having more HUC level unique values or metadata too. I like the idea. I wonder if one mannings per reach is too broad, but if we go more grandular, having it based on a HUC specific file or somethign is still very valuable. I bet there are some post processing in the parquet files we can wrap into this.

[hhs732]

Optimized roughness values for BLE HUCs with bathy included for SO4 and higher:

[hhs732]

Median_optz_ch_mannN = 0.0725 Median_optz_ob_mannN = 0.1011

0.07245 | 0.10106

[hhs732]

In-channel roughness map

[hhs732]

Over-bank roughness map

[hhs732]

Alpha test metrics comparison:

[hhs732]

[hhs732]

[hhs732]

Let's look at the stacked bars:

[hhs732]

Optimizing for NWS and USGS (AHPS) HUCs with ahps objective function. So far, mannN for 22 HUCs was optimized.