QAQC scripts for GIS verifications

[popovs]

For the dens 'field visits' layer, automate GIS verifications for the forestry columns.

First, develop functions to pull the following GIS layers from BCGW:

• [x] Vegetated Land Cover - Proj_Age_Class
• [x] Cutblock layers
• [x] Base Mapping > Transportation roads
• [x] Licenses & Permits > Road Sections / Segments'''
• [x] Not BCGW, but also Serena's depletion layer

Next do some basic GIS operations to make these files usable for our particular verifications:

• [x] Subset each layer by each year of study -> make yearly GIS files of each layer (applies to all below steps)
• [x] For forest age class areas, ~rasterize~ select the age class polygon by age class. (PROJ_AGE_CLASS_CD_1 column) (kept this as polygon operation)
• [x] For forest status, divide age class polygon into >40yr forest and <40yo forest. Pull out two data products:
  • [x] Single large merged multipolygon of >40yo and <40yo forest. Each feature is for one year.
  • ~[ ] Rasterstack of combined >40yo and <40yo forest (>/<40yo cells having different cell values, ofc). Each layer in the stack is one year.~ Keeping as polygons. Still works fairly fast.
• [x] ~Also rasterize depletion areas.~ Staying w polygons. Moved GDB to local machine for faster processing.
• [x] Intersect/add up forest status ~raster~ polygons with depletion raster for each year. This result will serve as the forestry verification data layer for all GIS verification operations except the roads one.
• [x] For roads, merge together all relevant road datasets''', again subsetting by year. Ensure the roads are actually used/active, if that's deducible from the data.

'''Segments & sections might be identical - check this

GIS verifications

• [x] Proportion forestry within 60m of den

• Buffer points by 60 m

• Intersect with the forestry verification raster layer.

• Exact extract the proportion forested (%) within 60m of the den.

• Note <40yo areas OR cutblock areas OR rocky areas = "non forested"; >40yo areas = "forested".

• [x] Distance to <40yo forest

• For each den & year, point distance to nearest raster value where forest <40yo (follow MAMU methodology for distance from sea to inland areas)

• Confirm if we care about distance around barriers

• If the raster cells are too 'blocky'/low res and produce weird values for distance, use alt method with distance to single <40yo multipolygon feature.

• [x] Distance to >40yo forest

• Same methodology as above, but for the older >40yo forest age class.

• [x] Distance to nearest built road

• Point distance to nearest line.

Misc GIS

Mary additionally requested the following: "Proportion of forest age classes within 1.5 km of each den."

• [x] Proportion age class within 1.5 km of each den

• Buffer den by 1.5 km

• Exact extract with rasterized age class polygons

• Convert number of cells to percentages for each age class

• [x] Road density within 1.5km of each den

• Buffer den by 1.5 km

• Intersect/select touching roads

• Buffer highways by 30m and other road segments by 10m > dissolve boundaries

• Intersect 1.5km circle with buffered roads

• Calculate sq km of road within that intersection

[popovs]

Currently have scratch code written out to process all forestry layers except for rasterized age class layer. Need to wrap into neat functions to then apply functions to whole den list.

Also, removed cutblocks layer from analysis. The data is simply too inaccurate to use - it relies on polygons of cut block approval areas, but doesn't necessarily align with actual areas cut IRL, e.g. wildlife retention patches aren't necessarily included in the layer.

[popovs]

Create the following functions for yearly data subsetting:

• Subset roads to correct dates
• Subtract year of interest from VRI projected date to get age class of the polygon at the time of den date inspection
• Select depletion layer years at the time of den date inspection

Then when calling GIS operations, user can specify whether they'd like to subset geometries by year or not.

[popovs]

Question for Mary: for %age age class within 1.5 km of the den, are we talking about % of forest cover, or % of TOTAL area. E.g. if a den is by the coast (OUT_Kung_3 in screenshot), and you expand 1.5km to include the ocean, the % age class coverage by total area will likely be small, even though 100% of the forest is old (age class 9).

For now, assuming percentages as a proportion of forest cover, rather than as proportion of total area.

[popovs]

Question for Mary, regarding road density & distance to nearest road: what to do with railroads?

[popovs]

First draft of all scripts written up and pushed to main. Will open new ticket for any script cleanups or new script requests.