Skip to content

Latest commit

 

History

History
125 lines (92 loc) · 3.95 KB

README.md

File metadata and controls

125 lines (92 loc) · 3.95 KB

Experiment - Viewshed

Sebastian T. Brinkmann (31.3.2022)

Libraries

library(protoVS)
library(terra)
library(sf)
library(tidyverse)

We used a theme by Koundinya Desiraju to visualise our results.

Elevation Data and Observer Locations

For the viewshed experiment we used LiDAR derived a Digital Surface Model (DSM) and Digital Terrain Model (DTM) with 1m spatial resolution of the Vancouver metropolitan area (Natural Resources Canada 2019).

DTM <- rast("data/DTM_Vancouver_1m.tif")
DSM <- rast("data/DSM_Vancouver_1m.tif")

We’ll calculate viewsheds at random observer locations across the complete study area. To avoid edge effects we’ll apply a negativ buffer of 1000 m to the area of interest (AOI) shapefile.

# Read shapefile and apply negative buffer
aoi <- read_sf("data/Vancouver.gpkg") %>% 
  st_buffer(-1000)

# Points every 100 m
aoi_grid_100 <- sf::st_make_grid(aoi, 100, what = "centers")
aoi_grid_100 <- aoi_grid_100[aoi]

# Extract observer height (DTM + 1.7m) and height of DSM
h_dtm <- terra::extract(DTM, st_coordinates(aoi_grid_100)) + 1.7
h_dsm <- terra::extract(DSM, st_coordinates(aoi_grid_100))

# Heigth at observer location should be smaller than height of DSM
# which would be the case if observer was inside a house, underneath a tree etc.
# In this case, the algorithm would have lower computation time
set.seed(1234)
observer <- aoi_grid_100[which(h_dtm > h_dsm)] %>% 
  sf::st_sample(1000) %>% 
  sf::st_cast("POINT") %>% 
  sf::st_as_sf(crs = st_crs(aoi)) %>% 
  dplyr::rename(geom = x)

Click the figure for an interactive version

Viewshed calculation

Viewsheds for all points have been calculated at multiple distance levels from 1 to 1000 m. We measured total runtime for each distance level and scaled it to computation time per viewshed. The times represent the average of 20 runs and are given in milliseconds.

max_distance_vec <- c(
  1:4, seq(5, 20, 5), 25, seq(50, 1000, 50)
)

vs_time <- viewshed_comparison(observer = observer,
                               dsm_rast = DSM, dtm_rast = DTM,
                               max_distance_vec = max_distance_vec,
                               observer_height = 1.7, sample_size = 10,
                               ncores = 10, progress = TRUE)

Results

Click the figure for an interactive version

RFVS vs. RFVS + prototyping algorithms - Level of Improvement

At ~250 our novel algorithm has similar (but not lower), otherwise better performance compared to the benchmark RFVS algortihm. Overall, mean single-core computation time per viewshed was improved by 34.2% (±20.9%) using our novel algorithm. Smaller improvements of 19.8% (±18.4%) were measured for small viewsheds (r < 300 m), using a large radius (r >= 300 m) led to an improvement of 39.9% (±19.4%).

Bibliography

Natural Resources Canada. 2019. “High Resolution Digital Elevation Model (HRDEM) - CanElevation Serie.” https://open.canada.ca/data/en/dataset/957782bf-847c-4644-a757-e383c0057995.