ohm-np / masters-thesis Goto Github PK

Post challenges (at least every week) and ideas on how to solve them

AOIs:

• WDPA polygons (7495)
• simplified WDPA polygons (4722)
• grid cells (~10/20K)

Rasters:

• worldpop (1 km resolution)
• high resolution rasters (create a demo raster)

Tests:

• sequential
• parallelization (8 cores)
• parallelization (16 cores)

Functions:

• terra::zonal
• terra::extract
• exactextractr::exact_extract

Results:

• save csv for time taken processing all aois at once
• several plots

Creating bubble plots might show all the relevant results into a single plot since we can visualize 4 types of variables within a single plot.

Example bubbleplot

Create factor plots for all the functions - in particular

Think of how final table/chart should look like. It should help users to choose the most efficient and effective combination of R tools and techiques to process their resources for corresponding polygons.

• outlines
• story (abstract to conclusion)
• graphics
• codes
• references

Append here in this issue.

AOIs:

• wdpa polygons
• simplified wdpa polygons
• grid cells

Target vector:

• ecoregions

Functions:

• terra::crop/intersect
• sf::st_intersect

Results:

• save time elapsed as csv
• several plots

% of CPU and memory usage

Variables

• populaton count (raster)
• ecoregions (vector)

1. track for one hour (60601)
2. plot the graphs

Prepare the results using mapme.biodiversity package and process following datasets:

• SRTM Elevation
• Terrain Ruggedness Index using SRTM DEMs
• population count
• Accessibility
• Mangrove Extent
• Terrestrial Ecoregions of the World

Study Areas?

• WDPA polygons
• Simplified WDPA polygons
• Fishnets

Engine choices?

• terra zonal
• terra extract
• exact extract

Create line plots for these combinations:

1. 0.01 raster for 1000, 5000, and 10000 sqkm fishnets
2. 0.001 raster for 1000, 5000, and 10000 sqkm fishnets

Also, study the accuracy of the results obtained from this combination:

1. population 2020 raster for allPAs (zonal, extract, and exact)

Use sequential (0) and parallel (1-16 cores) approach