Mistakes in GreenALS_SLPdeg_PD FIS file

[juliagrabowski]

The GreenALS_SLPdeg_PD FIS file in the default library doesn't run. It defines three categories for slope and point density, but then in the rule set refers to a missing fourth category for slope. It also defines an elevation uncertainty range of [0 0.75], but then the high uncertainty category maxes out at 0.6. I am developing a FIS for a green Lidar dataset and would like to use this ruleset as a reference. I am curious what the author originally intended for the ruleset and if the mistakes could be rectified.

[joewheaton]

Hi @juliagrabowski. So, that FIS was made a long, long time ago (back in the Matlab DOD version). Jim McKean and I slapped it together for an AGU poster. It was based on the old EARL platform, which is totally out of date. If you want to build one, you might use one of the other LiDAR's as a starting point.

The elevation range should actually probably extend further and yes the high value is generally better to make match the upper end of the range (but it should still work).
https://github.com/Riverscapes/fis-dem-error/blob/bb7cc8884eda42b3db2857e7da4d923840a2d58e/BySurveyType/ALS/GreenALS_ZError_SLPdeg_PD.fis#L30-L36

As we worked more with LiDAR later, we definitely realized how misleading point density can be as it is often a product of overlapping flight lines and more doesn't mean better (often worse with overlap). Anyhow, you might look at:

Schaffrath KR, Belmont P, Wheaton JM. 2015. Landscape-scale geomorphic change detection: Quantifying spatially variable uncertainty and circumventing legacy data issues. Geomorphology 250 : 334–348. DOI: 10.1016/j.geomorph.2015.09.020

Without testing this one, but knowing I got it to work previously in Matlab, I'm wondering what you mean by "doesn't run"? That can be the case if the input rasters you are providing fall outside the specified range? This is often the case for more recent flights.

If you are doing one for Green LiDAR, I would go back to the drawing board entirely and likely deal with in-channel and out-of-channel separately (e.g. with seperate FIS).

[joewheaton]

Thanks @fluviotect for those... spot on suggestions for work-arounds. It is annoying, but I guess that's the price we pay for still allowing/working with shapefiles ;). It is so annoying how many things we work with on GCD where things have been ever so stable, but then it becomes unstable because of some stupid inconsistency or change by ESRI. All in keeping with their commitment to over-charging and under serving all of us ;).

[fluviotect]

Amen brother!

[fluviotect]

I finally bit the bullet and committed to QGIS and pure python about 9 months ago. It slowed me down a bit initially, but even at somewhat reduced productivity, I didn't/don't get angry the way I did of the course of 22 years of suffering with ESRI products. Nearly every single time I've dipped back into Arc since, I got reminded of why I've been so eager to get away.

The only down-side has been that I use shapefiles much more than I had since about 2003, but I look forward to transitioning to sqlite and geopackages...after thesis submission ;-).

[joewheaton]

Me too, but only about 5 months ago. I have been surprised how straight-forward transition has been. The geopackages are amazing! Really powerful and way simpler then the 5-8 files that make up a shapefile. Plus, if you go to PostGIS, its an easy transition.

[fluviotect]

I've found the transition pretty smooth as well....certainly much easier with versions since 3.10 than when I made earlier attempts at like v2.x and v3.6. Editing was a tough transition and I have some intermittent frustration with layout bugginess, but I really like how I can implement almost anything I can think of with a reasonable amount of effort. Makes GIS fun again.

Good to hear about geopackages. I've only encountered when I inadvertently created one while accidentally mixing geometries in geopandas. Seemed a bit odd as a datatype at first, but easily addressed once I wrapped my head around jumping through the extra (in pure python) hoop. Definitely seems to have good potential.

Cheers for the tip on PostGIS. Will look into it for sure. (also after thesis submission)