Inputs and intermediates required for sensitivity analysis

[lauren-herbine]

What I'm trying to do

In #684, @joewheaton laid out some channels for sensitivity analysis of analysis window sizes. We want to be able to get to something like below to be able to test different window sizes.

The following sizes were requested:

• [ ] 200 m (100 up 100 down)
• [ ] 400 m
• [ ] 600 m
• [ ] 800 m
• [ ] 1000 m
• [ ] 1200 m
• [ ] 1400 m
• [ ] 1600 m
• [ ] 1800 m
• [ ] 2000 m
• [ ] 2200 m
• [ ] 2400 m
• [ ] 5000 m
• [ ] 10,000 m

Then you can sort by streamsize (0, 1 or 2)

To accomplish this, I need necessary intermediates in the DGO/IGO format in attribute tables of at least one complete project (mainstems included). To cut down on the calculation time, I have split attributes up into categories according to their necessary lines of evidence:	Category	Elevations + a length	Areas of polygons at DGO scale	Sinuosities	VBET polygons	Stream network counts	N/A
Metrics	Stream gradient	active floodplain area	Channel Sin	Count of floodplain RR	confluence count	Stream Type
	Valley gradient	active channel area	Planform Sin	Count of floodplain RL	diffluence count	Headwater
		Active Channel Area to Valley Bottom Ratio	VB Sin		tributary junciton count	Ecoregion III
		Active Floodplain Area to Valley Bottom Ratio			Stream order	Drainage Area

I will only be testing one metric from each category

What I need

I will need the following inputs or intermediates to be included in the attribute table of a GNAT project at the DGO/IGO scale:

stream gradient

• [ ] Minimum stream elevation
• [ ] Maximum stream elevation
• [ ] Channel length

active floodplain area

• [ ] Area of active floodplain polygons

count of floodplains RR

• [ ] Count of floodplain polygons RR

confluence count

• [ ] Count of confluences

channel sinuosity deprioritize

• [ ] Channel length
• [ ] Channel straight line distance

The biggest thing I need is these at the DGO scale so that I can write scripts that test them at different analysis window sizes. Additionally, doing this for a complete project (one where the mainstems are working) would be best, as I need to test these for each of our different stream sizes.

Thank you!

[philipbaileynar]

Action items for GNAT:

• Enhance GNAT calculation so that if a measurement or metric does not specify a window size then the code simply uses the footprint of the DGO polygon as the analysis window.
• Add the following as new measurements with NULL window size:
  • minimum elevation
  • maximum elevation
  • total blue line length of the main path
  • area of active floodplain
  • count of confluences

Problematic metrics in @lauren-herbine request above:

• Count of floodplain polygons RR is problematic for two reasons:
  • Dependent on the digitized direction of the blue line which is potentially unreliable.
  • A single floodplain pocket could get counted multiple times if it spans multiple DGOs. This means that the count won't scale properly when compositing larger window sizes.
• Sinuosity will be problematic because the straightline distances of each DGO will not add up to the correct straight line distance over longer, manually composited window sizes.
  • One potential solution would be to use metrics that represent the lat and long of the top and bottom of DGO. The manual composition of larger window sizes might be able to figure out how to use these coordinates to derive a straight line distance. This sounds easy but is fraught with problems because the metric code will not be able to identify which coordinates are the top and bottom of the polyon; it will just be pair 1 and pair 2 (note that each metric can only be one number, so these will need to be stored as Lat1, Lng1, Lat2, Lng2)

[lauren-herbine]

I will discuss some of these points with @joewheaton tomorrow, and then I will follow up with @philipbaileynar and @KellyMWhitehead at the end of next week (after new warehouse launch and run of the GNAT 30).

[lauren-herbine]

Talking with @joewheaton

• We're going to get rid of count of floodplains as a metric
• I am going to switch out channel sinuosity for a riverscape channel segment sinuosity - more info coming on that, but it will solve the need for straight line distances over multiples of DGOs