cont-limno / bathymetry Goto Github PK

Categorical

• lake depth by glaciation status
• lake depth by connectivity
• Landform (mountains, prarie, etc.)

Continuous

• lake depth by lake surface area (is there a piecewise relationship?)
• Elevation s.d.
• Lakeshape (convexhull) versus depth

Misc

• Does an "elipitic sinusoid" predict lake depth?

Top candidate is mean absolute percent error (yardstick::mape). Need to calculate in scripts/05_depth_model.R

See 01_geometry.Rmd (showing median slope calculations)

mdsumner recommends:

I'd geotiff them with lzw and tiling on (cog) and ship with a Geopackage index, filename and coverage polygon. Store the foreign crs as string if the rasters use different ones. This would be a breeze in and for R

https://twitter.com/mdsumner/status/1252606856976982018?s=20

https://x.com/FreyaHolmer/status/1750950122718962127?s=20

Need to save an old bbl file to get references to work. See https://tex.stackexchange.com/a/172319/52459

https://tex.stackexchange.com/a/336026/52459

Solved with: ftilmann/latexdiff#174 (comment)

This should go in 01_geometry.Rmd chunk 01_geometry_base

Looks like the most efficient way would be to move one table (probably the lake characteristics) and one figure (probably the conceptual diagram) to the appendix, and cut about 200 words or so. The ms would be at 3 figures, 1 table, and ~4000 words

Are lakes with bathymetry:

• deeper
• larger
• closer to urban centers?

Also acknowledge GLEON

Look at the recipes and rsample packages specifically the step_downsample function.

https://youtu.be/dbXDkEEuvCU

This should go in 06_stats.R computed using lagosus_depth_predictors.csv

The elevatr topo downloads from AWS were timing-out...

Currently this is done using a brute force moving window interpolation

Could use a more involved method like those explored at: https://fishandwhistle.net/post/2019/bathymetry-lake-volume-estimation-using-r/

Big Island Pond, ME

Triple the rows of table 1 by adding sections for only concave, convex, and all lakes

See:

• 05_depth_model.R

Which alternative (i) does Table 1 represent?

• 02_model_metrics.Rmd

make tables/02_model_metrics.pdf

that would be a 20 line table

Alternative nearshore slope calculations are already done in 00_get_nearshore.R. Alternative in-lake slopes already done in 00_get_geometry.R.

Current models in 05_depth_model.R ~L13 use slope_mean (the mean slope of the entire buffer) and inlake_slope (the point slope from a single deepest point) to make the depth_grid_metrics.rds object. Package the code to make this object into a function, Create alternative depth_grid_metrics_*.rds files.

in-lake slope alternatives (00_get_geometry.R)

• inlake_slope_mean - mean slope of the entire lake basin

• inlake_slope_online_mean - mean slope of a line extending from the max depth point to the shoreline to account for variations in slope

• inlake_slopes_online_mean - mean slope of line(s) extending from max depth point(s) to the shoreline

• inlake_slope_pnt - a single max depth point divided by the shortest distance to the shoreline

• inlake_slope_pnts - depth of (possibly multiple) max depth points divided by their average distance to the shoreline

deepest point distance alternatives (00_get_geometry.R)

• dist_deepest - distance from a single deepest point to the shoreline

• dists_deepest - distance of (possibly multiple) max depth points to the shoreline

nearshore slope alternatives (00_get_nearshore.R)

• nearshore_slope_mean - mean of all points in the buffer

• nearshore_slope_online_mean - mean of all points in the buffer in a 20m width strip extending from the deepest point

• nearshore_slopes_online_mean - mean of all points in the buffer in a 20m width strip extending from the deepest point(s)

Also, need to make depth_grid_metrics.rds, and/or depth_grid.rds and/or depth_fits.rds part of the Makefile pipeline.

The (Table 1) proxy-proxy-model should be shown in a graph, and its full statistics (regression equation) should be given.

Make it second to last SI fig (S9). Discuss at L341 in the LnO submission.

I think there is some manual crs assignment that will break with sf 0.9:
https://www.r-spatial.org/r/2020/03/17/wkt.html

This produced no interesting results

Essentially Figure S4-C split out by depth and size classes. Compute classes using lagosus_depth_predictors.csv. Place code in figures/01_geometry.Rmd