[21pt] 1st-order catchment max stages

[CarsonPruitt-NOAA]

Since most of our catchments are 1st-order and most 1st order catchments might not require the 25 meter maximum stage that we use as a global max, we should take a look at how we might lower the max stage for these catchments to see what kind of efficiencies we can gain for both processing and storage.

This task has 2 parts:

• [ ] Configure fim_pipeline to run SRC using two different max stages. delinate_hydros....sh uses the TauDEM tool catchhydrogeo to compute SRCs and accepts a catchment list and a stages list. We'll probably have to run this tool twice: once for 1st order catchments with the lower max and again for catchments 2 and up. Modifications to make_catchment_and_stage_lists.py to output another set of catchment/stage files will have to be done to accomplish this.
• [ ] Figure out what the max stage should be for 1st order catchments. I'm thinking that using the NWM retrospective dataset will probably be the best for this. Another option is to use the HydroVIS cached FIM databases to see the stages used in operational FIM, but I think the retrospective is more comprehensive. For each 1st order NWM reach, find the max discharge of the entire retrospective dataset and use the current SRCs to figure out what the corresponding stage should be. Taking maximum stage for all reaches should give us a good idea of what max stage to choose -- probably 20 to 50% higher than the maximum retrospective stage should cover what we need.

[CarsonPruitt-NOAA]

Zahra, I've been thinking about this a little lately, and I think it may be better to start with Greg's log-fit 100-yr flows instead of trying to look through the entire NWM retrospective dataset.

[ZahraGhahremani]

Sure, I am still working on the first part and testing how it works. I will probably start working on the second part tomorrow. Could you please point me to where the file is? I also have an idea for finding the max stages. Since the NWM data does not perform well for small streams, the NWM 100-year stage value could be far from the actual max stage value. My idea is to use NWM 100-year stage values along with network attributes (e.g., stream order, slope, channel sinuosity) as inputs to the ML model and use USGS max stage values as target variables. I believe there are ~7000 or more USGS sites that we can extract for training. This will be a better estimate even if model accuracy is not high compared to the NWM 100-year stage.

[CarsonPruitt-NOAA]

@GregoryPetrochenkov-NOAA could you put your 100-yr flow file into our EFS drive with the other NWM v3 files and let Zahra know what the name of the file is?

[GregoryPetrochenkov-NOAA]

I went ahead and sent an email with the location of the 100-yr recurrence flow file on the EFS Drive.

[ZahraGhahremani]

I used 100-yr flows and fim_4_5_4_4 (236 HUCs excluding AK) to determine max stage for 1st order catchments. Here is the results. About 84% HUCs can use max stage < 6 m.