Summarize jumps in land cover from FORE-SCE to NLCD transition

[jds485]

This issue is to document how many catchments have large jumps (> 5% proportion) for each land cover class.

Plot suggestions (might already exist)

• boxplots for the change in land cover proportion from 2000 to 2001. One boxplot per land cover class.
• facet plots with time on x-axis and land cover proportion on y-axis. One line per PRMS segment to visualize changes over time

[msleckman]

Initiated in branch msleckman/drb-inland-salinity-ml/review_lulc with starting point graph of diff between 2001 and 2000 explore_land_cover_2000_2001.md

Box plot of change between 2000 and 2001 tbd.

[msleckman]

Reviewed a couple different plots for this issue which can be viewed in full in the explore_land_cover_2000_2001.md. (plots don't render when pushed to github for some reason but hoping to resolve this soon).

Showing one of the plot here, but there are others in the md (forked branch here)

The boxplot below shows the distribution of proportions across all lc categories at the catchment level. When nested between adjacent years 1990 and 2008, we can better capture the difference. Generally 2001 and 2008 actually show greater mean similarity compared to 2000 and 2001. Vis versa, 1990 and 2000 show to be more similar. This is most likely due to the discrepancy between NLCD and FORESCE. We also notice across all proportion values that there are many outliers above the mean, is this a concern? I'd say it is expected that some catchments are fully covered by one lc class, so the outliers are not unexpected in my view given that our range is 0 to 1.

Moving forward, would it be preferred to only select either2000 or 2001 as a interval year to take forward. Any other ideas on how to handle this? @jds485 would be interested in your thoughts and please let me know if you have other questions.